Category Archives: Uncategorized

The risks and rewards of ML as an API

One of the projects I’ve been working on is a study on COVID-19 misinformation in Saudi Arabia. So far we’ve downloaded over 100,000 tweets. To expand the range of analytic tools that can be used, and to open up the dataset for non-Arabic speakers (like me!), I wrote a ML-based translation program, and fired it up yesterday morning. It’s still chunking along, and has translated over 27,000 tweets so far.

I think I’m seeing the power and risks of AI/ML in this tiny example. See, I’ve been programming since the late 1970’s, in many, many, languages and environments, and the common thread in everything I’ve done was the idea of deterministic execution.  That’s the idea that you can, if you have the time and skills, step through a program line by line in a debugger and figure out what’s going on. It wasn’t always true in practice, but the idea was conceptually sound.

This translation program is entirely different. To understand why, it helps to look at the code:


This is the core of the code. It looks a lot like code I’ve written over the years. I open a database, get some lines, manipulate them, and put them back. Rinse, lather, repeat.

That manipulation, though…

The six lines in yellow are the Huggingface API, which allow me to access Microsoft’s Marian Neural Machine Translation models, and have them use the pretrained models generated by the University of Helsinki. The one I’m using translates Arabic (src = ‘ar’) to English (trg = ‘en’). The lines that do the work are in the inner loop:

batch = tok.prepare_translation_batch(src_texts=[d['contents']])
gen = model.generate(**batch)  # for forward pass: model(**batch)
words: List[str] = tok.batch_decode(gen, skip_special_tokens=True)

The first line is straightforward. It converts the Arabic words to tokens (numbers) that the language model works in. The last line does the reverse, converting result tokens to english.

The middle line is the new part. The input vector of tokens is goes to the input layer of the model, where they get sent through a 12-layer, 512-hidden, 8-heads, ~74M parameter model. Tokens that can be converted to English pop put the other side. I know (roughly) how it works at the neuron and layer level, but the idea of stepping through the execution of such a model to understand the translation process is meaningless. The most important part of the program cannot be understood in the context of deterministic execution.

In the time it took to write this, its translated about 1,000 more tweets. I can have my Arabic-speaking friends to a sanity check on a sample of these words, but we’re going to have to trust the overall behavior of the model to do our research in, because some of these systems only work on English text.

So we’re trusting a system that we cannot verify to to research at a scale that would otherwise be impossible. If the model is good enough, the results should be valid. If the model behaves poorly, then we have bad science. The problem is right now there is only one Arabic to English translation model available, so there is no way to statistically examine the results for validity.

And I guess that’s really how we’ll have to proceed in this new world where ML becomes just another API. Validity of results will depend on diversity on model architectures and training sets. That may occur naturally in some areas, but in others, there may only be one model, and we may never know the influences that it has on us.

Welcome to the future of software development

Some good(?) insights from COVID-19 death rates

April 11, 2020 – I’ve put together a website with some friends that show these values for all countries dealing with the pandemic:

March 23, 2020

I think I found a way of looking at COVID-19 data in a way that makes intuitive sense to me – growth rate. For some context, let’s start with the Johns Hopkins scary dashboard:

Screenshot from this morning

This is a very dramatic presentation of information, and a good way of getting a sense of how things are going right now, which is to say, um… not well.

But if we look at the data (from here), we can break it down in different ways. One of the things we can do is look at trends in some different ways. Here, I’m going to focus on the daily death rate. In other words, what is the percentage of deaths from one day to the next? First, let’s look at Italy and Iran, two countries that are currently struggling with the worst of the crisis so far:

These still look horrible, but things do not appear to be getting worse as fast as they were  in early February. The curves are flattening, but it’s still hard to see what might happen in the future. We’re just not good at understanding exponential charts like the one on the left  much more subtle than “OMG!” Logarithmic charts like the one on the right can be misleading too – that’s a big jump between 1,000 deaths and 10,000 deaths at the top of the chart on the right. And at the bottom, we’re looking at nine deaths. 

What happens if we look at the same data as a rate problem though?

That looks very different. After a big initial spike, both countries have a rate of decrease that fits pretty well to a linear trend. So what do we get if we plug the current rate of increase into those linear approximations and solve for zero? In other words, when are there zero new deaths?

As we can see from the far right of the chart, as of today, Italy’s rate of new deaths is 11.89%, or 0.1189. Iran is 7.66% or 0.0766. Using those values we get some good news:

  • Italy: 27 days, or April 19th
  • Iran: 15 days, or April 7th

Yay! But…

Let’s look at the US. There’s not really enough data to do this on a state-by state basis yet, but there is plenty of data for the whole country. Again, we get a good linear approximation. The problem is, it’s going the wrong way:

The increase in our death rate (0.69% per day) is more than either Iran’s or Italy’s rate of decrease. At this point, there is literally no end in sight.

Let’s look at the world as a whole using death rates. This time the best fit is a second-degree polynomial, which produces U-shaped curves:

Also not good. Things clearly improved as China got a handle on its outbreak, but the trends are now going the other way as the disease spreads out into the reset of the world. It’s clearly going to be a bumpy ride.

I’d like to point out that there is no good way to tell here what caused these trends to change. It could be treatments, or it could be susceptibility. Italy and Iran did not take the level of action that China did, yet if trends continue, they will be clear in about a month. We’ll know more as the restrictions loosen, and there is or isn’t a new upturn as the lid comes off.

Notes from The History of Cartography, Vol 3: Cartography in the European Renaissance

The History of Cartography, Volume 3: Cartography in the European Renaissance

David Woodward

History of Cartography project


  • This is an enormous series. This single volume is over 2,000 pages, so this is going to be a slow read. I’m interested in this period of time because this is when maps transitioned from a kind of pictographic story to instruments of navigation. The snippets that follow reflect that focus.


  • The thematic essays raise important issues in the history of cartography that both set an agenda for future research on Renaissance maps and take stock of the growing role of cartography as a way to organize social, political, and cultural space. These essays are meant to be thought provoking, rather than exhaustive, and reflect some of the multilayered approaches that the study of maps has adopted in the past two decades. They show how the authority of maps became an essential factor in influencing the ways in which Renaissance Europeans saw and imagined the geographic layout, order, and substance of the world – with “world” meaning not only an external object to be represented, but also a stage on which internal human aspirations could be played out.  (p. xxxix)
  • The investigation of how maps were conceived, made, and used in this period provides a case study highlighting some of these historiographical issues in a new way. Indeed it is surprising that Burckhardt completely ignored these cartographic aspects even when stressing the importance of the discovery of the world and its relationship to the discovery of the self, both topics on which the history of cartography has much to say. (p. 6)

Table (p. 7)

  • In Volume 1 of this History, the point was made that the word mappa or mappamundi in the Middle Ages could be used to describe either a text or a map. This practice continued into the sixteenth and seventeenth centuries, as with Sebastian Munster’s Mappa Evropae (Frankfurt, 1537), John Smith’s A Map of Virginia (Oxford, 1612), or Thomas Jenner’s A Map of the Whole World (London, 1668). Indeed the metaphorical use of the word “map” to describe not only geographical descriptions but also other activities has exploded even in our own day, as we hear almost daily of the “road map” to peace in the Middle East. (p. 7)
  • Likewise, the classical and medieval written land itineraries continued to be a robust tool for wayfinding, and these were by no means replaced by their graphic equivalents. Although we have a famous example of an assemblage of graphic and written itineraries in the Tabula Peutingeriana, an image whose pedigree goes back to the fourth century, written directions of how to get from one place to another predominated over maps in the medieval period. One may even question the extent to which graphic itineraries were actually used on the road. (p. 8)
  • Finally, textual sailing directions, known as periploi in classical times and portolans (portolani) in the Middle Ages, continued to be favored by many sailors over their graphic equivalents into the sixteenth and seventeenth centuries, particularly in northern European waters, where they became known as rutters. The confusion still persists today, as the term “portolan” is often used when “portolan chart” is intended, leading some to propose that the term be abolished altogether. As Fernandez-Armesto argues in this volume, maps and charts were not used for navigation in the Renaissance as much as -written sailing directions. (p. 8)
  • It was not derived from Ptolemy’s Geography, for Ptolemy stressed that local maps (chorographies) should not be based on measurement, but should instead be made by artists. (p. 10)
  • Between 1400 and 1472, in the manuscript era, it has been estimated that there were a few thousand maps in circulation; between 1472 and 1500, about 56,000; and between 1500 and 1600, millions. The significant increase in the sheer number of maps available for viewing calls for an explanation. Certainly maps began to serve a huge variety of political and economic functions in society. (p. 11)
  • The change in the abstract conception of space-from the center-enhancing mappaemundi to the Ptolemaic isotropic structure of mapmaking-has often been called the quintessential modernity of Renaissance cartography. The evidence for this lies in the relative scarcity of terrestrial maps bearing longitude and latitude before the fifteenth century. No terrestrial maps using longitude and latitude survive from thirteenth- and fourteenth-century Europe, despite Roger Bacon’s description of one on a sheepskin with cities shown by small red circles in the “Opus maius” (ca. 1265). In comparison, by the mid-seventeenth century, the observation of latitude and longitude as control points for topographical surveys had been introduced in France. What happened in the intervening four centuries is routinely ascribed to the rediscovery of Ptolemy’s manual of mapmaking in the first decade of the fifteenth century. (p. 12)
  • The notion of a bounded uniform space also implies that the objects placed in it are co-synchronous, a concept that, as we shall see, led to the idea that historical and “modern” maps could and should be separate documents. Since the surface is represented as a uniform space, scale and proportion are also possible. (p. 13)
  • Measurements of sufficient precision to take full advantage of the Ptolemaic paradigm were not available until astronomical measurements of latitude and longitude had become routine. (p. 13)
  • Geographic coordinates were thus mainly of scholarly and not practical concern until reliable astronomical measurements of both longitude and latitude became available in the late eighteenth century, after a satisfactory chronometer had been developed. Coordinates and projection grids certainly were powerful rhetorical devices in the fifteenth and sixteenth centuries, but the data behind them was often questionable. (p. 13)
  • The adoption of systematic map projections introduced a variety of centering and framing issues. The center of a projection did not usually imply either the author’s viewpoint or the most important feature to be portrayed. Unlike mappaemundi, in which Jerusalem, Delos, Rome, or some other holy place might be at the center of the map, a map such as Rosselli’s ovoid “World map” was centered on no particular place (the center is off the coast of modern Somaliland). What could be manipulated was the field of view of the projection. Since graduation in longitude and latitude forced the hand of the cartographer to some extent, the area to be covered by a projection had to be carefully calculated. (p. 14)
  • The tense of medieval mappaemundi usually covered a broad span of historical time. No strong distinction between a location and an event was drawn. Places that had once been important in history but no longer existed were shown side by side with currently important places. The map told a story, often a very long one. In the fifteenth and sixteenth centuries, as the atlas became a major genre, this storytelling role was still enormously important in maps. (p. 16)
  • The use of intersections of longitude and latitude that Ptolemy proposed as control points for mapmaking is not unlike the process by which a researcher gathers observations about the world and compares them against the framework of the laws of nature. It is not surprising that the map has been used as a metaphor for modern science. If “science” in the Renaissance meant the pursuit of knowledge about the natural world, the model of cartography built upon the cumulative observations of others. (p. 17)
  • An illustration of this approach to compilation using widely different sources is provided by Nicolaus Cusanus’s intriguing image of the cosmographer as creator, which we find in the Compendium, written in the year of his death, 1464. Nicolaus chose the metaphor of a cosmographer as a man positioned in a city with five gates, representing the five senses. Messengers bring him information about the world using these senses, and he records the information in order to have a complete record of the external world. He tries to keep all the gates open so as not to miss information gathered by any particular sense. When he has received all the information from the messengers, he “compiles it into a well-ordered and proportionally measured map lest it be lost.” He then shuts the gates, sends away the messengers, and turns to the map, meditating on God as the Creator who existed prior to the entire world, just as the cosmographer existed prior to the appearance of the map. Nicolaus concludes that, “in so far as he is a cosmographer, he is creator of the world,i” a carefully worded phrase whose sentiment would get cosmographers such as Gerardus Mercator and Andre Thevet into trouble with the church a century later. Nicolaus’s story illustrates the notion that by creating maps people saw, perhaps for the first time, that they could influence events and create worlds, that they could have the freedom to do things, rather than accept passively whatever God had ordained. Implicit in this passage is the realization that the world and the human representation of it were two different things. (p. 17)
  • Renaissance cartography has often been linked to the colonial and religious expansion of Europe. Mapping supported a sense of territorial self-entitlement that allowed religious and political leaders to claim vast areas of land overseas in the name of Christian European states. In Brian Harley’s words, “Maps were also inscriptions of political power. Far from being the innocent products of disinterested science, they acted in constructing the world they intended to represent …. Cartographic power was also a metaphor. It was expressed as imperial or religious rhetoric, as part of the creation ritual of taking possession of the land.” Such ceremonies of possession varied with the colonial power. The Portuguese relied on the abstract means of description, measured latitudes, to claim land. Their argument was that they had developed the technological knowledge to do so and hence had the right to wield it to their advantage. Mapping and surveying knowledge seem such an obvious form of evidence for colonial claims that their lack of treatment in some works is puzzling. (p. 19)
  • Ptolemy’s positive influence was far subtler, implying through a mathematization of the known inhabited world by means of longitude and latitude a measured-albeit faulty-estimate of what remained beyond the Greco-Roman inhabited world. Marco Polo’s book, on the other hand – even granted its author’s penchant for exaggeration – provided a narrative description of renewed trading possibilities with the East. Marco’s travels, in turn, were prompted by the Crusades (1096-1270), which enormously widened the geographical horizons of many classes of people, increased mobility, and fostered a culture of trade and travel. (p. 20)
  • Harris has tnade the point that cartography was a paradigmatic “big science” in the sense that it employed long-distance networks. He uses the concept of the “geography of knowledge,” by which he means the spatial connections between artifacts and people associated with a particular branch of knowledge, to explain how large corporations operated. He gives four examples, all of which have strong cartographic associations: the Casa de la Contratación de las Indias, the Consejo Real y Supremo de las Indias, the Verenigde Oostindische Compagnie (VOC), and the Society of Jesus. (p. 20)
  • Eisenstein’s thoughtful commentary on lvins’s dictum on the exactly repeatable pictorial statement was particularly welcome to historians of cartography as it used the example of printed maps to enlarge the context. She introduced the topic by stating that “the fact that identical images, maps and diagrams could be viewed simultaneously by scattered readers constituted a kind of communications revolution in itself.” Eisenstein’s view of the importance of printing for the cumulative gathering of information is echoed by Olson, whose general book on the implications of writing and reading unusually contains a section on maps. According to Olson, “The 600 or so maps which have survived from the period before 1300 show no sign of general developmental progression towards a comprehensive map of the world. The principal stumbling block to such a map was the lack of reliable means of duplicating maps, an obstacle overcome only with the invention of printing and engraving, and the invention of a common, mathematical, frame of reference which would permit the integration and synthesis of information being accumulated on the voyages of discovery.” (p. 21)
  • The concept of publishing did not depend on printing; Pliny the Younger refers to an “edition” of a thousand copies of a manuscript text. But when viewed as conveyors of information, Ivins and Eisenstein argue that the advantage of printed images lay more in the production of versions free from the corruption of the copyist, which could be used for comparative study. When map compilers had at their fingertips several standard printed sources of geographical data, such study was bound to benefit. As maps from different regions, scales, and epochs were brought into contact with each other in the course of compiling successive editions of atlases, contradictions became more visible, and divergent traditions more difficult to reconcile. (p. 21)
  • However, if one focuses not on the content of maps but on their economic role as consumer commodities, a different picture emerges. Here their graphic form as well as their function was important in establishing a holistic vision of the world. Such a vision of the general layout of countries and continents might not have been particularly accurate (a limitation that persists today not only in the general population but also in political leaders), but it engendered a culture of cosmopolitanism in a larger range of social classes. Geography also became an essential part of general education, and the accoutrements of the cartographer (surveying instruments, globe, and armillary spheres) became icons of learning. (p. 22)
  • One could maintain that the use of maps to plot observations lagged as much as the sacred uses of maps persisted. (p. 23)
  • This is not to say that profound changes in cartographic method and practice did not take place in the Renaissance. The fact that the abstract theory of geographical coordinates was accepted as a way to make maps was in itself a significant change, as was the construction of maps orthogonally, from an infinity of impossible human viewpoints in space. The implications of this geometric view of cartography for the centering, framing, and orientation of maps were far reaching in the public perception throughout the world. (p. 23)
  • Coincident with this new way of plotting data arose an awareness of the representation itself and of how it related to the world, or an awareness that representations of the world and the world -itself were two different things. This resulted in a greater reliance on or more thought given to using artificial codes in cartographic representation. (p.23)
  • The Middle Ages has been described as a period that “knew little of maps,” and indeed the number of surviving examples, even if allowances are made for what was probably an extremely high rate of loss, do not suggest that maps were produced and consumed in particularly large numbers between the fifth and fourteenth centuries. (p.25)
  • Although medieval maps often used to be described as copying a few standard models and repeating a tired assortment of information drawn from classical and biblical sources, it is becoming increasingly clear that they, like all other maps, should instead be understood as tools for thinking and as flexible means of communicating ideas. In the Middle Ages, as in other periods, maps could be shaped and manipulated to meet particular needs as their authors drew from graphic and textual traditions, from experience, and from their own ideas to create individual artifacts suited to given contexts. As Gautier Dalche has emphasized, maps, like other representations, do not inform us generally about contemporaries’ perceptions of space, but rather about the mental and technical tools available to the mapmaker. Medieval maps must, in short, be approached not as transparent windows into their creators’ and users’ minds but as rhetorically constructed documents belonging to specific times and specific contexts. Recent studies have emphasized the importance of exploring these contexts, whether the specific codicological context of a particular manuscript or the larger social and cultural setting in which the map was conceived, as essential to understanding the full meaning of a given map within its society. (p. 26)
  • One of the most influential contributions to the study of medieval cartography has been the idea that world maps were intended to describe time as well as space. Since the publication of two highly influential articles by von den Brincken on the close relationship between universal chronicles-those that attempted to sum up aJl of human history in one work . and world maps, it has been widely accepted that one function of these maps was to give an overview of the world, understood as the theater of human, and especially Christian, history. (p. 30)
  • Within their broad function as representations of space and time, world maps could serve a wide variety of more specific rhetorical needs. One way to explore the functions of the world map in medieval society is through the multivalent meanings of the world itself in the learned culture of the time. Part of the curiosity about the physical world that characterized the twelfth-century Renaissance was the desire to understand the earth as a part of a system. The concern among philosophers for the machina universitatis or the machina mundi led them to focus on the system underlying the universe and the laws that governed it. The details of the earth itself (terra, both the planet and the element earth) were of less interest to them than the grand mechanism of the world (mundus). Contrasted with this interest in the machina mundi was the equally vibrant idea of contemptus mundi (renunciation of the world), which drew on a related but different definition of the “world” to contrast the ascetic life with the life of ordinary secular affairs. “Secular” recalls the term saeculum that contrasted “the world of men and of time” with the eternal world of the Christian God. Between these extremes were the views of historians, pilgrims (whether armchair or actual), and other travelers, for which locations and events on the earth did matter and needed to be recalled. (p. 31)
  • Roger Bacon’s discussion of a figura or drawing showing major cities located according to their longitude and latitude. Bacon has in the past been credited with considerable innovations in geographical thought, most particularly in his understanding of the use of coordinates to create an accurate graphic representation of the world’s places. (p. 33)
  • Bacon was thus not unique in his interest in locating the places of the world accurately within a system that connected them to the heavens. (p. 34)
  • This was due in part to the heightened attention given in the twelfth century to the literal sense of biblical exegesis: understanding the names, places, and history described in the Bible was seen as the necessary foundation for examining other meanings (moral, Christological, or eschatological). (p. 34)
  • In conclusion, the surviving examples of world maps, along with other texts, images, and references to maps, bear witness to-the passionate interest in the real world described by Gautier Dalche. The variety of functions that these maps could play reflects the multifarious meanings of the world in medieval culture, as the maps served to describe, analyze, summarize, and create knowledge and perceptions about the fundamental spaces of human existence. These were works destined for both elite and somewhat more popular audiences-including pilgrims, parishioners, and consumers of romances-to whom they helped provide visual, intellectual, and imaginative access to the larger world. As we have seen, the sensitivity of recent scholarship to the specific contexts in which maps appeared and the ways in which they were used has given us new insights into the complexity and subtlety of the potential meanings of medieval world maps, although much remains to be uncovered about the perception and representation of space in this fertile period. (p.36)
  • It is, however, extremely important to remember the comminnent of time and resources involved in producing the medieval copy: the question then arises of what this map meant to the society that found the human and financial resources to copy it. It has been plausibly explained in the context of the strong interest in the classical world that we have already seen influencing the toponyms of later medieval world maps; however, more research should be done to elucidate the importance and the influence of this map. (p. 38)
  • In addition to these surviving examples of itinerary maps, the significance of the itinerary – especially the written or narrated itinerary – is demonstrated by the frequency with which itineraries served as at least one source for other types of maps. For example, some of the information on the Hereford world map was based on an itinerary that may show a route familiar to English traders in France.  Even more substantial is the role played by itineraries in the creation of regional maps. These interconnections are especially striking in the rich cartographic production of Matthew Paris, although, as we will see when we turn to earlier maps of Britain, his work is far from unique in this respect. (p.39)
  • It is possible to describe the sources and creation of Paris’s maps of England in considerable detail, an approach very welcome in the study of medieval cartography, thanks to a study by Harvey.  According to his reconstruction, Paris began by adopting the outline of the island from a world map, probably of Roman origin. He then drew on an itinerary from Dover to the Scottish border to develop his representation of the interior, filling in extra place-names around this core. His subsequent revisions of the map reflect his discoveries of new sources, providing the river network, for example, and improvements in the coastline. Collectively, these maps demonstrate how powerful a process the compilation of geographical information from various sources could be and how central a role itineraries and world maps could play in the elaboration of regional maps. (p. 39-40)
  • most maps were made to aid in understanding, not primarily to represent space in a geometrically correct way. As Delano-Smith and Gruber point out, “a diagram is the most appropriate style for any map used in explanation,” a dictum with which medieval cartographers would have agreed wholeheartedly. (p. 44)
  • Both the examples of the smooth incorporation of the portolan chart with the world map and its more tentative acceptance in the Aslake world map indicate that, in spite of regional limitations of access to these map forms, mapmakers were eager to adapt new cartographic information to their own purposes when it came their way. The readiness of even a hesitant northern mapmaker to adopt a radically new depiction of space suggests that, in the fourteenth century, the idea was becoming fairly widely accepted that world maps could and should contain at least some detailed topographic information in addition to the historical and toponymic information presented by earlier world maps. (p. 46)
  • If we turn to our second author and artist, Opicino de Canistris, we find a similar range of maps in the service of a very different project. Opicino was not writing history, with its well-known attention to the loci (places) in which historical events took place. Instead, he worked from the equally familiar idea of the created world as God’s book to develop an elaborate system for understanding and recognizing sin in the individual via an analysis of the places of his life as represented on maps. (p. 47)
  • I have argued at length elsewhere that [Opicino de Canistris] believed that maps were important because the very schematization of the image of the world that they proposed bridged the gap between the materialistic human imagination and man’s higher powers of reason. As such, maps, for Opicino, were a potential answer to the spiritual problems of his time and fitting tools for a priest concerned with analyzing and combating unbelief. (p. 48)
  • Far from a unified project, the mapmaking and map use of the late medieval and early Renaissance period reveals itself as abundant and chaotic growth as yet unpruned into the chaste mathematica topiary of seventeenth-century cartography. (p. 52)

The Clockwork Muse

The Clockwork Muse: The Predictability of Artistic Change (1990)

Author (publications)

Colin Martindale (March 21, 1943 – November 16, 2008) was a professor of psychology at the University of Maine for 35 years.

Martindale wrote and did research analyzing artistic processes. His most popular work was The Clockwork Muse (1990). Martindale argued that all artistic development over time in written, visual and musical works was the result of a search for novelty.

Martindale was awarded the 1984 American Association for the Advancement of Science Prize for Behavioral Science Research.


Although he idea of Digital Humanities, or the quantitative analysis of the arts, is starting to make its presence felt in the last half of the 2010s, The Clockwork Muse predates all this by decades. Martindale was doing computational analysis of poetry, prose, visual arts, architecture, and even scientific writing in the 1980s!

The central premise of this book is that human creative output is driven by what he calls The Law of Novelty, which consist of two components:

  1.  Arousal potential
  2. Habituation, or exposure fatigue

His main premise is that to be successful, arousal potential must increase over time, and at a rate that the audience desires. Too little, and the audience will become fatigued and find novelty elsewhere. Too fast and the audience won’t comprehend. Artists that produce novelty at the right rate for their communities are the most successful. At the same time, artists behave according to the principle of least effort, and will take the easiest path to increasing arousal potential.

How art is created for human audiences is by varying the ratio of two approaches:

  1. Primordial content: This represents a “going back to basics” approach to art. Forms become simpler and less refined. Increases in primordial content typically occur at the beginning of a movement. Consider impressionism, where art went from highly refined representation to vaguer representations that incorporated a larger interpretive and emotional component.
  2. Stylistic change: This is the process that we see within an artistic movement, where the process is incrementally refined by the artists in the movement. A syntax is developed, and a progressively more sophisticated “conversation” emerges.

The contribution of these two approaches in art movements, careers, and even within sequential art such as novels and music vary inversely to one another in roughly sinusoidal patterns. While each value may decrease individually, the combination of the two increases.

Martindale then goes on to show that this theory holds in a multitude of contexts and studies. There are two general forms of the studies. In his text analytics, he analyzes variability/incongruity, and groups terms into “Primordial”/”Stylistic” buckets and then does statistical analysis on the results, generally finding good correlations. For the study of visual arts ranging from painting to architecture, he collects canonical images across a given timeframe (often in centuries), divides this span into segments, randomises the segments, and has subjects classify the images across multiple dimensions which in turn can be collapsed into the primordial/stylistic overarching categories.


My more theoretical thoughts

The dichotomy of primorial content / stylistic change matches Kaufman’s concepts of long jumps / hillclimbing on rough fitness landscapes, and also Iain McGilchrist’s interpretation of the roles of the left / right hemispheres of the brain in the The Master and his Emissary, where the right hemisphere operates in global, general contexts and the left is local and specific.

Stylistic change depresses the hill it is climbing. That’s why large jumps in the fitness space represented by increased primordial content work – Once a movement is exhausted, a small jump will only sink the terrain further. But a large jump is almost guaranteed to increase arousal potential.

People (and other organisms?) get bored with incremental progress below a certain rate (Habituation), and become biased towards long jumps that create change and provide novelty. There is survival value in this that is addressed in the explore/exploit paradox. This pressure for change at a certain rate is what drives much of our clustering behavior, which is based on alignment. Insufficient arousal builds up greater potential for nonlinear long jumps, or revolution.

Speed affects the ability to change direction. Changing direction stylistically can create more arousal potential than the linear extension of a particular technique (Though reversing direction should lower arousal potential). That means the faster that stylistic change is happening, the more likely that progress will be linear (realism -> superrealism -> hyperrealism, etc)


Chapter 1: A Scientific Approach to Art and Literature

  • As for history, the whole point of universal history is to find general laws that apply no matter what nations are involved or at what time. Because the topic is complex, these laws seem less deterministic than those governing objects falling in vacuums. History is, however, not completely chaotic and unpredictable. Just as in chemistry, certain combinations in certain situations will explode. Just as in physics, certain configurations in certain places will collapse. (Page 15)
  • Styles going by different names show obvious similarities: for example, eighteenth-century neoclassicism and fifteenth-century renaissance style, and fourteenth-century gothicism and nineteenth-century romanticism. Though the earlier style ~ did not directly cause the later style, we must explain the similarities. Perhaps art oscillates along a continuum so that such similarities are inevitable. If so, we must explain why it oscillates, and why along that continuum rather than along some other one. Of course, we must explain what the continuum is and why it exists. (Page 20)
  • The psychologist D. T. Campbell (1965) argues for direct application of the principles of Darwinian evolution to cha11ge in cultural systems and products. Sociocultural change, he says, is a product of “blind” variation and selective retention. The three necessities for evolution of any sort are: presence of variations, consistent selection criteria that favor sorne variants over others, and mechanisms for preserving the selected variants. At any time, a number of variants of a given object are produced, and the n1ost useful, pleasing or rewarding arc chosen for retention. Then, at the next point in tirnc, there is variation of the new form, and the process continues. Though such theories provide a general framework for thinking about aesthetic evolution, they do not tell us why aesthetic variation exists in the first place, nor were they proposed to do so. (Page 33)

Chapter 2: A Psychological Theory of Aesthetic Evolution

  • Theories concerning an inner logic driving change in the arts were anticipated by Herbert Spencer’s quasi-Darwinian theory. This English philosopher, in his major statement on art (1910 [1892]), set forth the principle that art, like everything else, moves from simple to complex. By complex, he meant more differentiated and hierarchically integrated. The anthropologist Alfred Kroeber (1956) followed Spencer in proposing such a simple-to-complex law, as did Kubler (1962). (Page 31)
  • If you drop an object in a vacuum, gravitation will cause it to move in a specific direction at a specific rate. Just so, if art were created in a social vacuum, pressure for novelty would cause it to evolve in a specific manner The empirical evidence suggests that art tends in fact to evolve in a social vacuum, and that non-evolutionary factors are comparatively negligible Though art is not produced in a complete social vacuum, I believe there to he more of a social vacuum than is commonly thought. Furthermore, social forces are analogous to friction, in that they impede or slow down the progress of an artistic tradition. (Page 34)
  • In the case of biological evolution, selected variations are, of course, encoded in the “memory” of DNA configurations. We may assume that more important or more complex DNA configurations are “forgotten” less quickly than less complex ones. The analogue is most certainly the case in the arts. If for no other reason than our educational system, the average reader of a contemporary British poet has some rudimentary “memory” of the poet’s predecessors at least back to the time of Chaucer. In comparison, the average person who purchases clothing has a poor memory for prior styles of fashion. As likely as not, such a person knows little about the styles of even thirty or forty years ago. The better “memory” of the poetic system leads us to expect that it should change in a different manner than fashion. In the latter, novelty seems often to be obtained by reviving a forgotten style, an option not as available to the poet. (Page 37)
  • …no one liked the Moonlight Sonata, he did not understand why the person was bothering with such chitchat. It did not concern Beethoven what people thought of that sonata. He did not write it for others. He wrote it for himself, and he liked it. As I have said, the notion that an artist is trying to communicate with an audience is misleading. It leads to an “audience-centric” confusion. (Page 38)
    • This book is about how Nomads work. High art has to be exclusively Nomad, and in most cases, drives the audience away. Pop art is different, because it cares about the audience.
  • A good deal of evidence supports the contcnhon that people prefer stimuli with a medium degree of arousal potential and do not like stimuli with either an extremely high or low arousal potential. This relationship, described by the Wundt curve, is borne out by several genera) studies reviewed by Theodore Schneirla ( 1959) and Berlyne ( 1967) as well as by studies 0f aesthetic stimuli per se. The effect has been found with both literary (Evans 1969; Kamann 1963) and visual (Day 1967; Vitz 1966) stimuli. There is some question about the shape of the Wundt curve (Martindale et al. 1990 ), but there is no question that people do like some degree of intensity, complexity, and so on. (Page 42)
  • WundtCurve
    • Hedonic tone and arousal potential. According to Berlyne (1971), hedonic tone is related to the arousal potential of a stimulus by what is called the Wundt curve: stimuli with low arousal potential elicit indifference, stimuli with medium arousal potential elicit maximal pleasure, and stimuli with high levels of arousal potential elicit displeasure. (Page 42)
  • Habituation refers to the phenomenon whereby repetitions of a stimulus are accompanied by decreases in physiological reactivity to it. The psychological concomitant is becoming used to or bored with the stimulus. Habituation is not merely the polar opposite of need for novelty. Avoiding boredom is not necessarily the equivalent of approaching novelty. People who habituate quickly to stimuli do not necessarily have a high need for novelty (McClelland 1951 ). In fact, creative people like novelty but habituate more slowly than do uncreative people (Rosen, Moore, and Martindale 1983). Because of this fact and because habituation seems to be a universal property of nervous tissue (Thompson et al. 1979 ). (Page 45)
  • Ambiguity is a collative variable. Collative properties such as novelty or unpredictability can vary much more freely than meaning in all of the arts. One soon finds that to increase the arousal potential of aesthetic products over time, one must increase ambiguity, novelty, incongruity, and other collative properties. This is the reason for my theoretical emphasis on collative properties rather than upon other components of arousal potential. (Page 47)
  • Peak shift is a well-established behavioral phenomenon (Hanson 1959). Consider an animal that is rewarded if it responds to one stimulus (such as a 200-Hz tone) and not rewarded if it responds to another stimulus (a 180-Hz tone). After training, the animal will exhibit maximal responsivity at a point beyond which it was rewarded and in a direction away from the unrewarded stimulus (a 220-Hz tone). J.E.R. Staddon (1975) argues that peak shift serves as the force behind sexual selection in biological evolution. Because of peak shift, female birds that prefer to mate with males with bright rather than dull plumage will show even greater preference for males with supernormal or above-average brightness. As a result, such males will mate more often and produce more offspring. As a further result, and because peak shift operates during every generation, the brightness of male plumage in the species will increase across generations. (Page 47)
  • On the psychological level, habituation occurs gradually and the need for novelty is held in check by the peak-shift effect. Thus, an audience should reject not only works of art with insufficient arousal potential but also those having too much. Finally, the principle of least, effort assures us that artists will increase arousal potential by the minimum amount needed to offset habituation. The opposing pressures should lead to gradual and orderly change in the arts. (Page 48)
  • Berlyne ( 1971) pointed out that the evolutionary theory has difficulty in explaining cases such as Egyptian art that show extremely slow rates of change. Yet, though much Egyptian painting was sealed in tombs-hardly a place to bring about speedy habituation-it did evolve as predicted by the theory (see pages 212-19). In general, the more an audience is exposed to a type of art, the faster the art should change. This assumption leads to specific predictions: we should find higher rates of change in living room furniture than in bedroom furniture, in everyday dress than in formal dress, and so on. (Page 52)
  • A simpler explanation is that clothing can lose some of its aesthetic qualities because of functional reasons. Fashion is not a high art. There is a pressure to increase arousal potential, but other pressures can add so much noise that the theory of aesthetic evolution is only marginally — rather than continually –applicable. Consumers can retard change by passive resistance. They can extinguish styles altogether by boycott. Walking sticks and hats are more or less extinct. The audience did this. This fact does not refute the theory of aesthetic evolution. Evolution can occur only when the environment permits it. If a politician kills all the poets, poetic evolution obviously ceases. If poets had to make a living by writing poetry, there would be no poetry; it would have gone the way of the scarlet waistcoat. On the other side of the coin, if the makers of scarlet waistcoats had not held to the silly notion that the only thing they are good for is to wear, there would still be plenty of them , and they would be quite fancy. Their functional aspect put all kinds of constraints on what they could look. like. Recall Kubler’s definition that if something has a use, it is not art. If something has a use, people want it to work. That gets in the way of its aesthetic aspects. If something has a use, people can stop using it and destroy its aesthetic aspects altogether. Being useless has distinct advantages. Paintings don’t come with warranties. Customers can’t return them and say they don’t work–or that they have suddenly started to work. Customers can’t say anything, really. If something is useless, it doesn’t make any sense to say how long it will remain useless or to quibble about exactly how it should be useless. (Page 55)
  • The evolutionary theory can be construed in two ways. The weak version is that the theory explains a bit, but perhaps not much, about art history. The strong version is that the theory explains the main trends in art history. Because the main axes of artistic style-classic versus romantic, simple versus complex-are isomorphic with the main theoretical axes-conceptual versus primordial content, low versus high arousal potential– it is not unreasonable to think that the strong version may be true. (Page 69)

Chapter 3. Crucible in a Tower of Ivory: Modern French Poetry


(Page 104)

  • We know that the truth must lie somewhere between these two extremes: that is, We know that poets cluster together into groups-for example, some tend to write lyrical poetry, whereas others write epic poetry. We want to know how many dimensions are needed to account for the similarities among poets. Fortunately, once we have correlated all of the poets with one another, a procedure called multidimensional scaling will tell us just this. Multidimensional scaling (Scikit-learn) tells us that the twenty-one French poets differ along three· main dimensions. These three dimensions account for 94 percent of the similarity matrix. (Page 114)

Chapter 4. Centuries of British Poetry

  • When a poetic tradition is first coalescing, poets are less sure of the rules of the game and of who else is playing. (Page 123)
    • I think this is the case for most if not all group consensus under uncertainty. I see this a lot in the D&D data.
  • Thus, legislators are really under pressure to produce new combinations of words, called laws, just as are poets. Suppose Parliament enacts a set of far-reaching new laws (initiates a new style). Subsequent Parliaments are going to have to pass laws that elaborate upon and refine the general laws. Eventually the result is going to be a set of overly specific, complicated, and contradictory laws. At this point, another stylistic change is in order: that is, a new set of general laws. These laws will also need refinement, so the whole process will begin anew. If this is at all close to capturing what legislative bodies do, then we might expect oscillations in primordial content because of oscillations in generality versus specificity of laws- an argument that cursory reading of the British statutes does support. General statutes are often followed by more concrete and specific ones that clarify the earlier statute or limit it so as to prevent unintended consequences. On the other side of the coin, a variety of specific statutes may eventually be replaced by an overarching general law. If we had a more appropriate measure of impact or complexity, we would almost certainly find that the complexity of law has increased across time. In an abstract sense, then, legal and poetic discourse are subject to evolutionary pressures that are isomorphic. (Page 131)


  • The amount of primordial content in the two prior periods is negatively related to primordial content in a given period. The amount of stylistic change is positively related to stylistic change in the prior period and negatively related to stylistic change two period earlier. These influences of the past on the present set up oscillations. (Page 141)
    • The delay in the connections may matter a lot. There may have to be a graph matrix term included in the chain. Behaviors would be very different for a fast, low weight link and a heavy weight, delayed link. Or is that a property in the reactivity of the node? Or both?
  • Primordial content shows a significant linear increase across all four periods. This linear trend accounts for 59 percent of the overall variability in primordial content. There is no sign of leveling off of this trend during the later periods. The pattern of results is most consistent with the hypothesis that the late metaphysical poets were caught in an evolutionary trap: They engaged in deeper and deeper regression, but did not achieve the desired increases in arousal potential. Once entrapped, the metaphysical style perished and was replaced by the ensuing neoclassical style. (Page 148)
    • I think this is a case where the “elevation” of the fitness landscape is sinking, but the practitioners are hillclimbing the wrong way.

Chapter 5. On American Shores: Poetry, Fiction, and Musical Lyrics

  • I have been talking, I know, in rarefied or abstract terms, our 170 poets ending by being represented by points in four – or five-dimensional hyperspaces. It is where they are in these abstract spaces that I have aimed to account for by four or five equations. The exciting thing is that poetry moves through these spaces in very orderly ways across the course of time and I think we can eventually capture the beauty of this sweep of history with quite simple and beautiful equations. Of course, this is hardly the usual way of talking about the history of poetry. However, I think that it adds to or complements-rather than in any way contradicting or negating-the more usual approaches. (Page 152)
  • To see how much of the overall variability in poetic language the evolutionary theory accounts for, I followed the same procedures that were used for French and English poetry. I first intercorrelated the poets’ profiles on the Harvard III categories to measure their relative similarity. Then, through multidimensional scaling, I found that four dimensions accounted for 94 percent of the variation among the poets. Correlating these dimensions with the content categories suggests that the dimensions arc getting at lyric, narrative, and two sorts of emotional modes. We can account for 25 percent of variation in this space with our theoretical variables-a good bit less than the figure for French and British poetry. One reason for this discrepancy may be that the American series contains more minor poets-who may not really have a good sense of tradition than do the French and British series. They may be perfectly fine poets, but their concerns may be tangential to those of poets writing in the focal American tradition. In fact, if we examine only the four most eminent poets for each period, 43 percent of the similarity among them can be accounted for by the theoretical variables-about the same figure as for French and British poetry (Page 165)
    • American poets are/were more nomadic?
  • you can’t make money writing poetry, the poet has considerable freedom: nothing he or she writes is going to make money, so there is no sense in trying. Unfortunately, the case is different with fiction. You can make money with it, but only if a lot of people like it. Publishing firms publish poetry for prestige, but they want to make a profit with fiction. They resist publishing fiction that no one will buy. This situation puts something of a non-evolutionary pressure on the writer. Even worse, the writer may want to make money and thus write more for an external audience and less for other artists. Of course, the external audience habituates and wants novelty. However, it must habituate more slowly than artists because it is exposed less frequently to literature than they. The real problem with the external audience is that it may place all sorts of non-aesthetic pressures on literature. If the audience takes a puritanical tum, it won’t read pornography. If it gets obsessed with abolishing slavery, it wants Simon Legree stories. If the audience’s desires get really extreme, the environment will be distinctly unfavorable to aesthetic evolution. To use a biological analogy, sexual selection can proceed only in a benign environment. In an environment with a lot of predators, birds of paradise could not have evolved, for their brilliant plumage would have attracted those predators. (page 169)

Chapter 6. Taking the Measure of the Visual Arts and Music

  • After establishing that subjects agreed in their ratings, I obtained a mean score for cach painting on each of the variables by collapsing across subjects. The rating scales were subjected to factor analysis to see whether the scales were measuring, as hoped, a smaller number of underlying dimensions. Factor analysis (tutorial), which represents similarities among scales in a multidimensional space, indicated that the scales were really measuring five axes or dimensions. It was clear that two of these factors were, as hoped, getting at the theoretical variables. One factor-referred to as arousal potential-had high loadings on the scales Active, Complex, Tense, and Disorderly. Another-referred to as primordial content-had high loadings on Not Photographic, Not Representative of Reality, Otherworldly, and Unnatural. (Page 188)

Chapter 7. Cross-National, Cross-Genre, and Cross-Media Synchrony

  • This gives us a closed circle. When the longer series of French and British painting was considered, it was clear that French painting influences British painting more than vice versa. Now we see that American painting is predictable from British painting. Closing the circle, l found French poetry to be at least synchronously related to American poetry Painting seems to be a much more international enterprise than poetry. At least for the last two centuries, British, French, ancl American painting appear to be closely related even when we take statistical precautions to avoid spurious findings. The Italian, British, and French series overlap for several periods. (Page 240)
  • According to Berlyne (1971), a work of art has three types of properties: psychophysical properties intrinsic to the stimulus (such as pitch, hue, intensity), collative properties (such as novelty, complexity, ambiguity), and ecological properties (such as denotative and connotative meaning). Berlyne held that these properties taken together determine the arousal potential of a work of art. Each of these aspects of a work of art offers several possibilities for cross-media correspondence. The psychologist S.S. Stevens (1975) showed that cross-modal matching on the intensity dimension is reliable. Another psychologist, Lawrence Marks (1975), showed that there are reliable consistencies in the matching of the brightness of colors and the “brightness” ( defined by pitch and loudness) of sounds. Collative variables (such as complexity or novelty) are clearly defined across all media. Kenneth Burke ( 195 7) argued that cross-media styles are based on two dimensions that Berlyne would call collative variables: unity and diversity. Meaning can also serve as the basis for cross-media styles-as is obvious in art forms having specific content, such as representational painting and sculpture. In terms of connotative meaning, Charles Osgood, George Suci, and Percy Tannenbaum’s (1957) three factors of evaluation, potency, and activity show up in a variety of rating tasks irrespective of what is being rated. Thus, connotative meaning could serve as a basis for equating works in different media. Finally, overall arousal potential or resultant liking (see figure 2.1, page 42) could serve as a basis for cross-media styles. I have argued that it makes sense to talk about primordial content in all of the arts. The experiment I did with Ross and Miller (1985) suggests that this cross-media effect is more than giving the same name to different things. In our experiment, people wrote fantasy stories in response to paintings. Primordial content in the stories-as measured by content analysis-was correlated with primordial content-as measured by rating scales-in the paintings. (Page 251)

Clock 254

Page 254

  • Thus, artistically untrained subjects do show a clear and significant tendency to perceive cross-media styles. Both baroque and neoclassical styles differ from the romantic style on the first dimension: that is, people see baroque and neoclassic styles as being similar to each other and dissimilar to the romantic style. Baroque and neoclassical styles do not differ significantly on this dimension. The second dimension differentiates baroque and neoclassical styles and, to a lesser extent, neoclassical and romantic styles. Baroque and romantic styles do not differ significantly on this dimension. Thus, people see baroque and romantic styles as similar to each other and dissimilar to the neoclassic style on this dimension. In summary, people perceive the three styles as varying in two fundamental ways, or along two dimensions. Unfortunately, multidimensional scaling does not tell us how to label these dimensions. (Page 255)

Chapter 8. Art and Society

  • Marxism ultimately attributes much social and cultural change to the emergence of new technology: technological innovations create new means of production; a class struggle ensues for control of these means of production. (page 264)

Chapter 9. The Artist and the Work of Art

  • I have argued that the effect of the pressure for novelty arises not so much from its strength as from its persistence. In the first part of this chapter, I investigate whether the pressure for novelty is strong enough to affect individual artists. Does the evolutionary theory help us understand only the great sweep of history, or can it shed light on stylistic development in the individual artist? Studies of creators as diverse as Beethoven and Grieg, Picasso and Rembrandt, and Dryden and Yeats suggest that the evolutionary theory is, indeed, relevant to the individual artist. At the end of the chapter, I report on some quantitative studies of nonevolutionary forces, such as temperament and psychopathology, that also shape the content of poets’ verse. In the middle part of the chapter, I turn quantitative methods upon individual works of art, ranging from prose through poetry to music. We find coherent trends across the course of literary narratives-trends that arise, in part, from a need to keep the reader’s attention but, in larger part, from what an author is trying to say. In fact, the type of trend we find sheds valuable light on what an author is–consciously or unconsciously-trying to say. (Page 284)
  • Aesthetic evolution shows a surprising self-similarity under magnification: when we examine the course of primordial content across centuries, we find long-term trends with superimposed oscillations. If we examine the course of primordial content across the career of an individual artist, we see exactly the same thing on a smaller time scale. Theoretically, these patterns are caused by the need to increase arousal potential. This need is caused by habituation: the audience-and the artist-tires of old forms and wants new ones. Habituation corresponds, as I have said, to building up a set of expectations. After these expectations are well developed, a work of art that conforms too closely to them will elicit neither interest nor pleasure. It is to avoid this sad fate that the artist must increase the arousal potential of his or her works. (Page 312)
  • On the psychological level, the theme of the journey to hell and back hypothetically symbolizes a regression from the conceptual (abstract, analytic, reality-oriented) thought of waking consciousness to primordial (concrete, free-associative, autistic) thought and then a return to conceptual thought. Of course, the psychoanalyst Ernst Kris (1952) holds that any act of creation involves an initial stage of inspiration and a subsequent stage of elaboration. In the inspirational phase, there is a regression toward primordial thought, whereas in the subsequent elaboration stage there is a return to analytic thinking. The inspirational stage yields the “rough draft” of the creative product, whereas the elaboration stage involves logical, analytical thought in putting the product into final form. Thus, the theme of the night journey mirrors the psychological processes involved in the creation of art. (Page 314)
  • If a narrative describing a journey to hell or a similar region symbolizes . regression from a conceptual to a primordial state of consciousness and a subsequent return to a conceptual state, we might expect that words indicating primordial content would first increase and then decrease in the narrative. (Page 315)
  • These two passages from page 333 describe earlier concepts of “coordinate frames” that we have used (in western society only?) to position imagination:
    • The names will be different, but the idea is the same Since Galen’s theory of the temperaments was good enough for Kant, Wundt, and Pavlov, I’ll just use the old terms. The idea of temperament theory is that people differ along two dimensions: sanguine (optimistic and happy) versus melancholic (pessimistic and anxious) and choleric (active and quick to anger) versus phlegmatic (calm and passive) Galen connected the temperaments with bodily fluids which were seen in turn as expressions of the four elements: air versus earth and fire versus water. A nice theory, but it turned out to be wrong on the level of physiology and physics. Maybe, though, the connection between elements and temperaments is right on the psychic level.
    • At least in states of fantasy and reverie, there is a correspondence between temperaments and elements. As the literary critic Northrup Frye put it, “earth, air, water, and fire arc still the four elements of imaginative experience, and always will be” (Bachelard 1964, p. vii [1938]).

Clock 335

Page 335

Chapter 10. Science and Art History

  • To test these predictions, I had eighteen subjects rate their liking for the Italian paintings shown in chronological order and seventeen subjects rate them when shown in reverse chronological order (Martiudale 1986a). To disguise the real purpose of the experiments, the Like-Dislike scale was embedded in a set of seven other scales. Since subjects showed highly’ significant agreement in their preferences, a mean preference rating was computed for each painting in both experiments. As predicted, the correlation of liking with order was insignificant for the chronological group. On the other hand, it was strongly negative — much more so than when the paintings were shown in a random order-for the reverse-chronological group (Page 341)
    • Good discussion on methods


Page 343

  • A really interesting element to consider is how the transformation from owl to cat differs from “morphing” to latent-space interpolation. Morphing is a dissolve that includes control points for an animated transition that typically involves input from users (e.g. the animator). Latent space interpolation lacks human involvement and creates images from that latent space that would probably be outside of human cognition:
    • BigGan
    • Humans seem to slowly generalise from a less familiar object until it resembles something more familiar, where detail is added up to a point, where it then varies.
  • Should we conclude from the changes in primordial content that the later drawings were made in more primordial states of mind? This is a possible but not necessary conclusion. The most parsimonious explanation would probably be that the later drawings are less realistic because of the cumulation of simplifying memory effects and lack of skill. It is reasonabler though, to expect variations in the degree to which thought is primordial even in a laboratory situation. Primordial cognition does not mean only extreme states such as dreams and delirium, which are in fact not conducive to artistic production. Our thoughts are always varying to a slight degree along the conceptual-primordial axis (Klinger 1978; Singer 1978). It is exactly such slight variations that I have been talking about whenever I have claimed. that artists are creating in an altered state of consciousness. (Page 344)
  • The series of responses to the sentence stem “A table is like ___ ” shows remarkable parallelism with what has occurred in French poetry since 1800. In order, the responses were:
    1. the sea, quiet
    2. a horizontal wall
    3. a formicaed [sic] bed
    4. the platonic form
    5. a dead tree
    6. a listening board
    7. versatile friendship
    8. vanquished forest
    9. a seasoned man
    10. two chairs
  • The first three responses are based upon the flatness of the surface of a table. The fourth response mediates between these and the next two, which refer to the material composition of tables. All of these similes are clearly “appropriate,” suggesting a high level of stylistic elaboration: irrelevant responses were being filtered out. This filtering is discarded with the seventh response. The ninth and tenth responses clearly indicate a new, less elaborated “style.” The ninth response may be mediated by the word “seasoned,” which is transferred from wood to man: man and wood share a common attribute, although in a completely different sense; therefore they are compared. The tenth response is based upon the close associative connection between table and chair. Associative contiguity overrides objective dissimilarity–exactly as happened in twentieth-century French poetry. Presumably, subjects in the later trials could continue to increase originality only by lowering level of elaboration, by applying less stringent stylistic rules to their responses. (Page 347)
  • The task of a scientist is to produce ideas. That wouldn’t be hard if it weren’t for the constraints. One of the most severe of these is that the ideas have to be true. More exactly, the ideas cannot be contradicted by empirical evidence. Furthermore, this constraint cannot be evaded by producing ideas that cannot in principle be tested against reality. Scientific ideas have to be falsifiable (Popper 1959). They have to be susceptible of being shown to be incorrect. This selection pressure is common to all scientific disciplines. Scientists are in the position of poets before the twentieth century, who had to produce realistic similes. Why don’t scientists discard this constraint and produce surrealistic science? In fact, this is just what mathematics has done. In its early stages, mathematics was an empirical science. The constraint for realism has long since been abandoned, so that mathematicians produce “theories” that do not necessarily correspond to any empirical reality. (Page 350)
  • Can we make an analogous argument for science? I think that we can, but the pressure to increase arousal potential in science must lead in an opposite direction: toward less rather than more primordial thought. Broadly speaking, a poet’s task is to create ideas of the form “x is like y” where x and y are coordinate terms or concepts on the same level. Habituation of arousal potential forces successive poets to draw x and y from ever more distant domains. They seem to do this by engaging in ever more primordial thought over time. Scientists, on the other hand, have to produce statements of the form “x is related toy, ” where x and y are on different levels. (Page 354)
  • These people formulate the general laws and develop the basic methods that define the paradigm. After the paradigm is established, normal science begins. Normal science is carried out by the members of what Crane calls an invisible college-a group of people who interact with each other and share the same goals and values. In the late stages of a paradigm, most major problems have been solved. This leads scientists to specialize on increasingly specific problems. This is necessary in order to extract the remaining ideas. In this stage, anomalies may also appear. In the final stage, there is exhaustion or crisis. The paradigm offers few possibilities and many problems or anomalies. As a consequence, members defect and are not replaced by new adherents. Very much the same thing happens in art and literature. A style gathers recruits who tend to know and interact with each other. There is excitement at first; but eventually the possibilities of the style are exhausted, and it becomes stale and decadent. Crane ( 1987) has analyzed the recent history of American painting in a manner analogous to her earlier analysis of science. (Page 358)
  • A paradigm can die for two reasons. There can be exhaustion without anomaly: that is, the paradigm has succeeded too well, and there are no problems left to be solved, as happened to Euclidean geometry. After the time of Euclid, there was nothing left to be done. The fit with reality had become perfect. Although Crane does not put it in these terms, we could say that arousal potential fell to zero, and the field elicited no further interest at least in the sense of scientists wanting to work in the area. In the case of exhaustion with anomaly, a different set of affairs exists. The exhaustion is not perceived as such. What is perceived is that a fit with reality can be obtained only with very complex theories. Further, new ideas can be generated only with considerable effort. In addition, there are undeniable anomalies or incongruities. Complexity, effort, and incongruity are likely to cause negative affect. In contrast, a new paradigm- if it is to be successful- will be more attractive: the fit with reality is not perfect, but neither are anomalies present as in an old paradigm. Anticipated payoff in relation to effort is much higher.
    The new paradigm usually wins by default. The old paradigm does not die. Its adherents die. Because they have not been able to recruit new disciples, the paradigm dies with them. Once this happens, the new paradigm becomes dominant, and the process begins anew. (Page 358)

Similar neural responses predict friendship

Similar neural responses predict friendship

Authors and related work


A detailed, lay overview has been written up in the New York Times: You Share Everything With Your Bestie. Even Brain Waves.

The study took a cohort (N = 279) of graduate students in a graduate program. Students were asked to list who their friends were, from which a social network was constructed. A subset (N = 42) of these students were then asked to watch a series of videos while their brains were being monitored by an fMRI machine. The timings of brain activations across 80 regions of the brain were compared to see if there were similarities that correlated with social distance. Statistically significant similarities exist such that friends could be identified by firing patterns and timing. Particularly, individuals with one degree of separation were strongly resonant(?), while individuals with three or more degrees of separation could not be discriminated by fMRI.

My more theoretical thoughts:

This is more support for the idea that groups of people “flock” in latent belief space. If everyone fired in the same way to the videos, then the environmental influence would have been dominant – a video of a sloth or a volcano is “objectively” interpreted across a population. Instead, the interpretation of the videos is clustered around individuals with high levels of social connection. Humans spontaneously form groups of preferred sizes organized in a geometrical series approximating 3–5, 9–15, 30–45, etc. This is remarkably similar to the numbers found in social organizations such as flocks of starlings (seven). As we’ve seen in multiple studies, a certain amount of social cohesion is beneficial as away of finding resources in a noisy environment (Grunbaum), so this implies that belief space is noisy, but that beneficial beliefs can be found using similar means.  Grunbaum also finds that excessive social cohesion (stampedes) decrease the ability to find resources. Determining the balance of explore/exploit with respect to depending on your neighbors/friends is uncomputable, but exploration is computationally more expensive than exploitation, so the pressure is always towards some level of stampede.

This means that in physical and belief spaces, the density and stiffness of connections controls the behavior of the social network. By adjusting the dial on the similarity aspect (increasing/decreasing stiffness of the links) should result in nomadic, flocking and stampeding behavior in belief space.


  • Research has borne out this intuition: social ties are forged at a higher-than expected rate between individuals of the same age, gender, ethnicity, and other demographic categories. This assortativity in friendship networks is referred to as homophily and has been demonstrated across diverse contexts and geographic locations, including online social networks [2345(Page 2)
  • When humans do forge ties with individuals who are dissimilar from themselves, these relationships tend to be instrumental, task-oriented (e.g., professional collaborations involving people with complementary skill sets [7]), and short-lived, often dissolving after the individuals involved have achieved their shared goal. Thus, human social networks tend to be overwhelmingly homophilous [8]. (Page 2)
    • This means that groups can be more efficient, but prone to belief stampede
  • Remarkably, social network proximity is as important as genetic relatedness and more important than geographic proximity in predicting the similarity of two individuals’ cooperative behavioral tendencies [4] (Page 2)
  • how individuals interpret and respond to their environment increases the predictability of one another’s thoughts and actions during social interactions [14], since knowledge about oneself is a more valid source of information about similar others than about dissimilar others. (Page 2)
    • There is a second layer on top of this which may be more important. How individuals respond to social cues (which can have significant survival value in a social animal) may be more important than day-to-day reactions to the physical environment.
  • Here we tested the proposition that neural responses to naturalistic audiovisual stimuli are more similar among friends than among individuals who are farther removed from one another in a real-world social network. Measuring neural activity while people view naturalistic stimuli, such as movie clips, offers an unobtrusive window into individuals’ unconstrained thought processes as they unfold [16(page 2)
  • Social network proximity appears to be significantly associated with neural response similarity in brain regions involved in attentional allocation, narrative interpretation, and affective responding (Page 2)
  • We first characterized the social network of an entire cohort of students in a graduate program. All students (N = 279) in the graduate program completed an online survey in which they indicated the individuals in the program with whom they were friends (see Methods for further details). Given that a mutually reported tie is a stronger indicator of the presence of a friendship than an unreciprocated tie, a graph consisting only of reciprocal (i.e., mutually reported) social ties was used to estimate social distances between individuals. (Page 2)
    • I wonder if this changes as people age. Are there gender differences?
  • The videos presented in the fMRI study covered a range of topics and genres (e.g., comedy clips, documentaries, and debates) that were selected so that they would likely be unfamiliar to subjects, effectively constrain subjects’ thoughts and attention to the experiment (to minimize mind wandering), and evoke meaningful variability in responses across subjects (because different subjects attend to different aspects of them, have different emotional reactions to them, or interpret the content differently, for example). (Page 3)
    • I think this might make the influence more environmental than social. It would be interesting to see how a strongly aligned group would deal with a polarizing topic, even something like sports.
  • Mean response time series spanning the course of the entire experiment were extracted from 80 anatomical regions of interest (ROIs) for each of the 42 fMRI study subjects (page 3)
    • 80 possible dimensions. It would be interesting to see this in latent space. That being said, there is no dialog here, so no consensus building, which implies no dimension reduction.
  • To test for a relationship between fMRI response similarity and social distance, a dyad-level regression model was used. Models were specified either as ordered logistic regressions with categorical social distance as the dependent variable or as logistic regression with a binary indicator of reciprocated friendship as the dependent variable. We account for the dependence structure of the dyadic data (i.e., the fact that each fMRI subject is involved in multiple dyads), which would otherwise underestimate the standard errors and increase the risk of type 1 error [20], by clustering simultaneously on both members of each dyad [2122].
  • For the purpose of testing the general hypothesis that social network proximity is associated with more similar neural responses to naturalistic stimuli, our main predictor variable of interest, neural response similarity within each student dyad, was summarized as a single variable. Specifically, for each dyad, a weighted average of normalized neural response similarities was computed, with the contribution of each brain region weighted by its average volume in our sample of fMRI subjects. (Page 3)
  • To account for demographic differences that might impact social network structure, our model also included binary predictor variables indicating whether subjects in each dyad were of the same or different nationalities, ethnicities, and genders, as well as a variable indicating the age difference between members of each dyad. In addition, a binary variable was included indicating whether subjects were the same or different in terms of handedness, given that this may be related to differences in brain functional organization [23]. (page 3)
  • Logistic regressions that combined all non-friends into a single category, regardless of social distance, yielded similar results, such that neural similarity was associated with a dramatically increased likelihood of friendship, even after accounting for similarities in observed demographic variables. More specifically, a one SD increase in overall neural similarity was associated with a 47% increase in the likelihood of friendship(logistic regression: ß = 0.388; SE = 0.109; p = 0.0004; N = 861 dyads)Again, neural similarity improved the model’s predictive power above and beyond observed demographic similarities, χ2(1) = 7.36, p = 0.006. (Page 4)
  • To gain insight into what brain regions may be driving the relationship between social distance and overall neural similarity, we performed ordered logistic regression analyses analogous to those described above independently for each of the 80 ROIs, again using cluster-robust standard errors to account for dyadic dependencies in the data. This approach is analogous to common fMRI analysis approaches in which regressions are carried out independently at each voxel in the brain, followed by correction for multiple comparisons across voxels. We employed false discovery rate (FDR) correction to correct for multiple comparisons across brain regions. This analysis indicated that neural similarity was associated with social network proximity in regions of the ventral and dorsal striatum … Regression coefficients for each ROI are shown in Fig. 6, and further details for ROIs that met the significance threshold of p < 0.05, FDR-corrected (two tailed) are provided in Table 2. (Page 4)
    • So the latent space that matters involves something on the order of 7 – 9 regions? I wonder if the actions across regions are similar enough to reduce further. I need to look up what each region does.
  • Table 2Figure6
  • Results indicated that average overall (weighted average) neural similarities were significantly higher among distance 1 dyads than dyads belonging to other social distance categories … distance 4 dyads were not significantly different in overall neural response similarity from dyads in the other social distance categories. All reported p-values are two-tailed. (Page 4)
  • Within the training data set for each data fold, a grid search procedure [24] was used to select the C parameter of a linear support vector machine (SVM) learning algorithm that would best separate dyads according to social distance. (Page 5)
  • As shown in Fig. 8, the classifier tended to predict the correct social distances for dyads in all distance categories at rates above the accuracy level that would be expected based on chance alone (i.e., 25% correct), with an overall classification accuracy of 41.25%. Classification accuracies for distance 1, 2, 3, and 4 dyads were 48%, 39%, 31%, and 47% correct, respectively. (Page 6)
  • where the classifier assigned the incorrect social distance label to a dyad, it tended to be only one level of social distance away from the correct answer: when friends were misclassified, they were misclassified most often as distance 2 dyads; when distance 2 dyads were misclassified, they were misclassified most often as distance 1 or 3 dyads, and so on. (Page 6)
  • The results reported here are consistent with neural homophily: people tend to be friends with individuals who see the world in a similar way. (Page 7)
  • Brain areas where response similarity was associated with social network proximity included subcortical areas implicated in motivation, learning, affective processing, and integrating information into memory, such as the nucleus accumbens, amygdala, putamen, and caudate nucleus [27, 28, 29]. Social network proximity was also associated with neural response similarity within areas involved in attentional allocation, such as the right superior parietal cortex [30,31], and regions in the inferior parietal lobe, such as the bilateral supramarginal gyri and left inferior parietal cortex (which includes the angular gyrus in the parcellation scheme used [32]), that have been implicated in bottom-up attentional control, discerning others’ mental states, processing language and the narrative content of stories, and sense-making more generally [3334, 35]. (Page 7)
  • However, the current results suggest that social network proximity may be associated with similarities in how individuals attend to, interpret, and emotionally react to the world around them. (Page 7)
    • Both the environmental and social world
  • A second, not mutually exclusive, possibility pertains to the “three degrees of influence rule” that governs the spread of a wide range of phenomena in human social networks [43]. Data from large-scale observational studies as well as lab-based experiments suggest that wide-ranging phenomena (e.g., obesity, cooperation, smoking, and depression) spread only up to three degrees of geodesic distance in social networks, perhaps due to social influence effects decaying with social distance to the extent that the they are undetectable at social distances exceeding three, or to the relative instability of long chains of social ties [43]. Although we make no claims regarding the causal mechanisms behind our findings, our results show a similar pattern. (Page 8)
    • Does this change with the level of similarity in the group?
  • pre-existing similarities in how individuals tend to perceive, interpret, and respond to their environment can enhance social interactions and increase the probability of developing a friendship via positive affective processes and by increasing the ease and clarity of communication [1415]. (Page 8)

The Radio in Fascist Italy

The Radio in Fascist Italy

  • Philip Cannistraro
  • Journal of European Studies
  • scholars have generally agreed that the control of the mass media by the state is a fundamental prerequisite for the establishment and maintenance of totalitarian dictatorships (pg 127)
  • It is not so widely acknowledged, however, that contemporary totalitarian governments have been largely responsible for the initial growth of the mass media-particularly films and the radio-in their respective countries. (pg 127)
  • In their efforts to expose entire populations to official propaganda, totalitarian regimes encouraged and sponsored the development of the mass media and made them available to every· citizen on a large scale basis. (pg 127)
  • Marconi shrewdly reminded Mussolini that it would be politically wise to place control of the radio in the hands of the state, pointing out the radio’s great potential for propaganda purposes (pg 128)
  • “How many hearts recently beat with emotion when hearing the very voice of the Duce! All this means but one thing: the radio must be extended and extended rapidly. It will contribute much to the general culture of the people” (pg 129)
  • … to insure that EIAR’s programmes conformed to the requirements of the regime’s cultural and political policies. The High Commission included government representatives from each major area of culture: literature, journalism., the fine arts, music, poetry, theatre, and films. The programmes Commission screened the transcripts and plans of all and censored the content of all broadcasts. (pg 130)
  • His broadcast, ‘The Bombardment of Adrianople’, was awaited by the public with great interest and was heralded by critics as the most significant cultural event of the Italian radio.ts Marinetti’s colourful language and emotion-packed presentation blasted un expected life into the Italian radio. His flam.boyant style introduced the concept of the ‘radio personality’ in Fascist Italy, and the success of his talk encouraged those who, like Marinetti himself, hoped to make the radio a new art form. Broadcasts by Marinetti, most of which were lectures on Futurism, continued to be heard on Italian radio each month for more than a decade. (pg 131)
  • The regime quickly recognized the effectiveness of this technique in· arousing listener interest, and it was an easy matter to transfer microphones to mass rallies from where the enthusiastic cheers of the spectators could be heard by radio audiences. (pg 132)
  • The popular announcer Cesare Ferri created the characters ‘Nonno Radio’ (Grandfather Radio) and ‘Zia Radio’ (Aunt Radio), speaking to Italian youth with unprecedented familiarity in terms they easily understood. (pg 132)
  • In order to popular arouse interest in its EIAR sought to stimulate indirect audience participation through public contests for short stories, poems, songs, In and children’s fairy tales. addition, surveys were conducted among listeners to discover trends in popular taste. (pg 133)
  • The radio had an important task to fulfil in the totalitarian state, that of binding the Italians together into one nation through common ideals and a common cultural experience inspired by Fascism. (pg 134)
  • Mussolini proclaimed Radio Rurale a great achievement of the Fascist revolution, for contemporary observers saw it as a new instrument with which to integrate rural existence into the mainstream. of national life. (pg 135)
  • The measures taken by the regime to overcome cultural and political provincialism by creating a mass radio audience in the countryside met with qualified success. (pg 137)
  • Regarded by many as an important step towards the creation of a truly popular culture, Radio Btdilla’s purpose was to give the working classes of the city and the countryside the means of acquiring a radio at a modest cost. Through the radio art, instruction, music, poetry-all the cultural masterworks–cease to become the privilege and unjust monopoly of a few elitist groups’. (pg 139)
  • ‘The ministry, in carrying out its delicate functions of vigilance over radio broadcasting, must guide itself by criteria that are essentially of a political and cultural nature.’ (pg 140)
  • Once the radio had been integrated into the structure of the Ministry of Popular Culture, the Fascists began to develop m.ore effective ways of using broadcasting as a cultural medium. While the number and variety of programmes had begun to increase by the beginning of the decade, it was only after 1934 that they became politically sophisticated. (pg 141)
  • Fascist racial doctrines became a major theme of radio propaganda during World War II. An Italo-German accord signed in 1940 to co-ordinate radio propaganda between the two countries included measures to ‘intensify anti-Jewish propaganda’ on the Italian radio as well as in foreign broadcasts.78 The Inspectorate for Radio Broadcasting organized an important series of anti-Semitic that centred around the ‘Protocols of Zion’, and talks such as ‘Judaism. versus Western Culture’, the ‘Jewish International’, and ‘Judaism. Wanted this War’, were broadcast from 1941 to 1943. (pg 143)
  • information received from the Vatican radio during World War II was generally regarded more accurate than the obvious propaganda broadcasts of the Allies (pg 147)
  • On the radio he astutely employed direct, forceful language, shouting short and vivid sentences to create a sense of drama and arouse emotional reactions. ‘This ‘maniera forte’ that characterized Appelius’ radio talks had a great appeal for many Italians, especially for the ‘little man’ who wanted to be talked to on his own level in terms he could readily understand.121 In his broadcasts Appelius screamed insults and ranted and raved at the foul enemies of Fascism. with a powerful barrage of verbal abuse, inciting his audiences to unmitigated hatred and scorn against the evil ‘anglo-sassoni’ and their allies. (pg 150)
  • In the broad context of Fascist cultural aspirations, all the media aimed at similar goals: the diffusion of standard images and themes that reflected the ideological values of Fascism.; the creation of a mass culture that conformed the needs of the Fascist state in its capacity as a totalitarian to government. (pg 154)

Sick echo chambers

Over the past year, I’ve been building a model that lets me look at how opinions evolve in belief space, much in the manner that flocks, herds and schools emerge in the wild.


Recently, I was Listening to BBC Business Daily this morning on Facebook vs Democracy:

  • Presenter Ed Butler hears a range of voices raising concern about the existential threat that social media could pose to democracy, including Ukrainian government official Dmytro Shymkiv, journalist Berit Anderson, tech investor Roger McNamee and internet pioneer Larry Smarr.

Roger McNamee and Larry Smarr in particular note how social media can be used to increase polarization based on emergent poles. In other words, “normal” opposing views can be amplified by attentive bad actors [page 24] with an eye towards causing generalized societal disruption.

My model explores emergent group interactions and I wondered if this adversarial herding in information space as it might work in my model.

These are the rough rules I started with:

  • Herders can teleport, since they are not emotionally invested in their belief space position and orientation
  • Herders appear like multiple individuals that may seem close and trustworthy, but they are actually a distant monolithic entity that is aware of a much larger belief space.
  • Herders amplify arbitrary pre-existing positions. The insight is that they are not herding in a direction, but to increase polarization
  • To add this to the model, I needed to do the following:
    • Make the size of the agent a function of the weight so we can see what’s going on
    • When in ‘herding mode’ the overall heading of the population is calculated, and the agent that is closest to that heading is selected to be amplified by our trolls/bot army.
    • The weight is increased to X, and the radius is increased to Y.
      • X represents AMPLIFICATION BY trolls, bots, etc.
      • A large Y means that the bots can swamp other, normally closer signals. This models the effect of a monolithic entity controlling thousands of bots across the belief space

Here’s a screenshot of the running simulation. There is an additional set of controls at the upper left that allow herding to be enables, and the weight of the influence to be set. In this case, the herding weight is 10. Though the screenshot shows one large agent shape, the amplified shape flits from agent to agent, always keeping closest to the average heading.


The results are kind of scary. If I set the weight of the herder to 15, I can change the change the flocking behavior of the default to echo chamber.

  • Normal: No Herding
  • Herding weight set to 15, other options the same: HerdingWeight15

I did some additional tweaking to see if having highly-weighted herders ignore each other (they would be coordinated through C&C) would have any effect. It doesn’t. There is enough interaction through the regular populations to keep the alignment space reduced.

It looks like there is a ‘sick echo chamber’ pattern. If the borders are reflective, and the herding weight + influence radius is great enough, then a wall-hugging pattern will emerge.

The influence weight is sort of a credibility score. An agent that has a lot of followers, or says a lot of the things that I agree with has a lot of influence weight The range weight is reach.

Since a troll farm or botnet can be regarded as a single organization,  interacting with any one of the agents is really interacting with the root entity.  So a herding agent has high influence and high reach. The high reach explains the border hugging behavior.

It’s like there’s someone at the back of the stampede yelling YOUR’E GOING THE RIGHT WAY! KEEP AT IT! And they never go off the cliff because they are a swarm Or, it never goes of the cliff, because it manifests as a swarm.

A loud, distributed voice pointing in a bad direction means wall hugging. Note that there is some kind of floating point error that lets wall huggers creep off the edge.Edgecrawling

With a respawn border, we get the situation where the overall heading of the flock doesn’t change even as it gets destroyed as it goes over the border. Again, since the herding algorithm is looking at the overall population, it never crosses the border but influences all the respawned agents to head towards the same edge: DirectionPreserving

Who’d have thought that there could be something worse than runaway polarization?

The Gamergate model of press relations

This is good enough that I’m reposting the entire thing. The original is here.

PRESSTHINK, a project of the Arthur L. Carter Journalism Institute at New York University, is written by Jay Rosen

I remember the first time I heard about Gamergate. A random follower on Twitter asked me if I had been following the story, which he said was “about ethics in games journalism.” No, I had not been following the story. In all innocence, I clicked on the link he sent me and tried to make sense of what I read. I failed. The events it described were impenetrable to me. (Disclosure: I am not a gamer.)

Eventually I learned what Gamergate really was. The more I learned, the more depressed I felt. The people who promoted Gamergate said they were concerned about journalism ethics. As a professor of journalism with a social media bent, I felt obligated to examine their claims. When I did I discovered nasty troll behavior with a hard edge of misogyny. “It’s about ethics in games journalism” became an internet joke. Deservedly so.

Recently Ben Smith, the editor-in-chief of Buzzfeed’s news operation, wrote: “The big story of 2014 was Gamergate, the misogynistic movement championed by Breitbart and covered primarily by new media. That turned out to be a better predictor of the presidential election than any rubber chicken dinner in Iowa (or poll by a once-reputable pollster).”

Ben is right. The Gamergate model in press relations posits that high-risk tactics should not be ruled out of consideration. It says that rejection and ridicule by the mainstream media can be a massive plus, because events like these activate — and motivate  — your most committed supporters: your trolls. The Gamergate model proposes that transgressing the norms of American democracy is not some crippling defect, as previously believed, but a distinct advantage because the excitement around the transgression recruits new players to the fight, and guarantees the spread of your content.

The Gamergate model anticipates that the mainstream press will freak out. Full stop. And it seeks to profit from this reaction. What the traditional press considers negative publicity is, from the Gamergate point of view, a kind of gift to The Leader. Trump and his advisors have absorbed these lessons. Gamergate is thus one possible template for the future of White House-press corps relations. Those who have not studied it carefully will be at a distinct disadvantage.

Filter bubbles, echo chambers, and online news consumption

Filter bubbles, echo chambers, and online news consumption

  • Seth R. Flaxman – I am currently undertaking a postdoc with Yee Whye Teh at Oxford in the computational statistics and machine learning group in the Department of Statistics. My research is on scalable methods and flexible models for spatiotemporal statistics and Bayesian machine learning, applied to public policy and social science areas including crime, emotion, and public health. I helped make a very accessible animation answering the question, What is Machine Learning?
  • Sharad Goel – I’m an Assistant Professor at Stanford in the Department of Management Science & Engineering (in the School of Engineering). I also have courtesy appointments in Sociology and Computer Science. My primary area of research is computational social science, an emerging discipline at the intersection of computer science, statistics, and the social sciences. I’m particularly interested in applying modern computational and statistical techniques to understand and improve public policy.
  • Justin M. Rao – I am a Senior Researcher at Microsoft Research. A member of our New York City lab, an interdisciplinary research group combining social science with computational and theoretical methods, I am currently located at company HQ in the Seattle area, where I am also an Affiliate Professor of Economics at the University of Washington.
  • Spearman’s Rank-Order Correlation
  • Goel, Mason, and Watts (2010) show that a substantial fraction of ties in online social networks are between individuals on opposite sides of the political spectrum, opening up the possibility for diverse content discovery. [p 299]
    • I think this helps in areas where flocking can occur. Changing heading is hardest when opinions are moving in opposite directions. Finding a variety of perspectives may change the dynamic.
  • Specifically, users who predominately visit left-leaning news outlets only very
    rarely read substantive news articles from conservative sites, and vice versa
    for right-leaning readers, an effect that is even more pronounced for opinion

    • Is the range of information available from left or right-leaning sites different? Is there another way to look at the populations? I think it’s very easy to get polarized left or right, but seeking diversity is different, and may have a pattern of seeking less polarized voices?
  • Interestingly, exposure to opposing perspectives is higher for the
    channels associated with the highest segregation, search, and social. Thus,
    counterintuitively, we find evidence that recent technological changes both
    increase and decrease various aspects of the partisan divide.

    • To me this follows, because anti belief helps in the polarization process.
  • We select an initial universe of news outlets (i.e., web domains) via the Open Directory Project (ODP,, a collective of tens of thousands of editors who hand-label websites into a classification hierarchy. This gives 7,923 distinct domains labeled as news, politics/news, politics/media, and regional/news. Since the vast majority of these news sites receive relatively little traffic,
    •  Still a good option for mapping. Though I’d like to compare with
  • Specifically, our primary analysis is based on the subset of users who have read at least ten substantive news articles and at least two opinion pieces in the three-month time frame we consider. This first requirement reduces our initial sample of 1.2 million individuals to 173,450 (14 percent of the total); the second requirement further reduces the sample to 50,383 (4 percent of the total). These numbers are generally lower than past estimates, likely because of our focus on substantive news and opinion (which excludes sports, entertainment, and other soft news), and our explicit activity measures (as opposed to self-reports).
    • Good indicator of explore-exploit in the user population at least in the context of news.
  • We now define the polarity of an individual to be the typical polarity of the news outlet that he or she visits. We then define segregation to be the expected distance between the polarity scores of two randomly selected users. This definition of segregation, which is in line with past work (Dandekar, Goel, and Lee 2013), intuitively captures the idea that segregated populations are those in which pairs of individuals are, on average, far apart.
    • This fits nicely with my notion of belief space
  • ideological-segregation-across-channels
    • This is interesting. Figure 3 shows that aggregators and direct (which have some level of external curation, are substantially less polarized than the social and search-based channels. That’s a good indicator that the visible information horizon makes a difference in what is accessed.
  • our findings do suggest that the relatively recent ability to instantly query large corpora of news articles—vastly expanding users’ choice sets—contributes to increased ideological segregation
    • The frictionlessness of being able to find exactly what you want to see, without being exposed to things that you disagree with.
  • In particular, that level of segregation corresponds to the ideological distance between Fox News and Daily Kos, which represents meaningful differences in coverage (Baum and Groeling 2008) but is within the mainstream political spectrum. Consequently, though the predicted filter bubble and echo chamber mechanisms do appear to increase online segregation, their overall effects at this time are somewhat limited.
    • But this depends on how opinion is moving. We are always redefining normal. It would also be good to look at the news producers using this approach…?
  • This finding of within-user ideological concentration is driven in part by the fact that individuals often simply turn to a single news source for information: 78 percent of users get the majority of their news from a single publication, and 94 percent get a majority from at most two sources. …even when individuals visit a variety of news outlets, they are, by and large, frequenting publications with similar ideological perspectives.
  • opposingpartisanexposure
    • Although I think focussing on ‘opposing’ rather than ‘diverse’ biases these results, this still shows that populations of users behave differently, and that the channel has a distinct effect.
  • …relatively high within-user variation is a product of reading a variety of centrist and right-leaning outlets, and not exposure to truly ideologically diverse content.
    • So left leaning is more diverse across ideology
  • the outlets that dominate partisan news coverage are still relatively mainstream, ranging from the New York Times on the left to Fox News on the right; the more extreme ideological sites (e.g., Breitbart), which presumably benefited from the rise of online publishing, do not appear to qualitatively impact the dynamics of news consumption.