Memes as "Toy Models" (Reza Negarestani)

Continuing the discussion from Meme-making as theory-making / critical meme reader #2 contribution WIP:
@Syderas on Reza Negarestani's Intelligence & Spirit:

I really like this way of thinking about Internet memes. I have argued that Internet memes are the ideal paradigm cases for anti-Dawkinsian memetics; Dominique Boullier (2018) and Sal Hagen (2022) make the same point, referring to memes as "the drosophila of our social sciences". Incidentally, Doug Campbell, one of my supervisors, also referred to Internet memes as the drosophila of meme studies. He's a philosopher of biology and is currently working on a software project about simulating evolution, which is of course a "toy model" of evolution.

It would seem that memes certainly have all of the makings of toy models. But what exactly are they modelling? Do they primarily perform a bookkeeping role "in the same sense as Wimsatt’s ‘genetic bookkeeping argument’, in which genetics is most useful as the study of readily identified markers of the evolutionary process"? Or, could they be said to perform a more speculative role as in Negarestani's thesis that humanity is a toy model for posthuman intelligence?

As far as the biological analogy persists, it's not much of a jump to consider genetic engineering as "toy modelling", in situations where in situ investigation is impossible because the phenomenon being developed has never existed in the wild before. It seems to me that we can extend this analogy very far, and fruitfully so. Creating a mailing list or a Discord server for a project is becoming so widespread that PC Gamer put out an article begging people to stop (the article ends with "And for the love of all that is holy: skip the #memes channel."); silos like mailing lists, Telegram channels, and Discord servers facilitate "toy model" experimentation with memes in a way that allows the resultant content to be released into "the wild" afterwards. 4chan threads are a sophisticated and early example:

The use of threads as makeshift contexts with spatial identities led to the boards becoming a collection of agar plates that different strains of memetic techniques could be developed and tested out on again and again. The disproportionately large influence of minor platforms such as 2ch or 4chan on their relatively mainstream counterparts is partly explained by this long-standing tradition of localised and temporary rules as a form of gamified memetics.

If memes are toy models, then making memes is an experimental research methodology. If making memes is an experimental methodology, then we should develop tools and infrastructure to facilitate a standardised and even replicable way to conduct such experiments. We should develop our own laboratories for meme studies!

1 Like

Thanks to @Syderas for bringing this point up and doing a pretty good job at summarising it. I think Negarestani's point as it regards AI is intriguing and thought-provoking, but, more importantly, it stems from a broader and far more profound principle.

I feel like the concept of a 'toy model' is probably pretty familiar for the academically-inclined, but maybe a little elaboration is in order. A toy model is a deliberately-simplified or abstracted model that is used for the purposes of making first-order estimations or calibrating intuitions. The physicist's joke about spherical cows is a humorous exaggeration of this practice (and its shortcomings): tl;dr, if you want to know how aerodynamic a cow is, it's easier to just think of it as approximately resembling a sphere, because then the math becomes a lot simpler. Most physics classrooms ignore air resistance for similar reasons, assuming that everything exists in a vacuum.

In my experience though, toy models are far more prevalent in economics. It's variously accepted (at least, imo, by good schools), that the vaunted 'perfectly competitive market' doesn't exist — it's an ideal type, a set of mathematical relations that are basically never fully manifest in reality, in the same way things that we identify as circles basically never have circumferences of exactly 2πr. All other economic models are similar constructions: contingent on assumptions and hacks and spherical cows, which 'bake-in' limitations and deviations from reality.

As it is said, all models are wrong (but some are useful). Or, in Korzybski's gloss, the map is not the territory, and a 1:1 map is not one of those useful models. There seems to be a fundamental tradeoff between total/absolute specificity and comprehensibility.

(Consequently, all of our physical models can be 'wrong', too. It's easy to see the failings of Newtonian physics, because it breaks down at the level of quantum physics — but this doesn't mean that quantum physics itself has no blind spots.)

I think the received interpretation of these dicta is humbling, and well-understood by the rationalist community over at LessWrong, incidentally, but I also think it somewhat boringly transforms empirical study into a practice of model-refinement and calibration. This can feel pretty grim, clinical, and depressing. I think the profound thing that Negarestani is highlighting here is that even our wrong models can be useful; combined with the right truth-seeking kind of attitude, they give us contrapositive indication of what a 'right' model looks like. We can get to a fuller picture of what 'intelligence' is by proposing (inherently limited) models of human intelligence, and then seeing how artificial intelligences stack up against it. We need to experiment, and 'get it wrong' sometimes, in order to know how to get it right.

Where is the link to memes and meme studies? Well, I sometimes find myself getting frustrated by the oft-abstruse and overcomplicated systematic and formal descriptions of memetic behaviour and practice (sorry, but the Jreg thread makes me a tiny bit irate — it feels like the String Theory of irony). But, my gripes with these kinds of models, if I can articulate them, can actually be important and useful stepping-stones to a more-integrated, easier-to-comprehend schema. As Syderas glosses, our 'views' (models) need to be "open to change and redefinition", and continual refinement. We shouldn't be afraid of putting forward new concepts and models in experimental spirit, but we also need to be ready to 'kill our darlings', or otherwise abandon them, if something else (including a future version) is more fit-for-purpose.

Are memes themselves toy models? Maybe, but I would say only insofar as a meme encodes or represents a specific "model of the world".


You hit the nail on the head with the point that incorrect models are still generative of knowledge, even when they are inaccurate (let alone full-scale simulations).

I think much of it is nonsense too, but that doesn't matter as much in the "applied memetics" or "memetic engineering" context: the models become correct post hoc, by virtue of people making memes that fit the model. It's "toy models" of the strange sort that Negarestani describes.

Models are not the same kind of things as paradigm cases or lab species: the role of a paradigm case or a lab species is not to be a simpler, abstracted representation of the real thing. But when we consider Internet memes as a paradigm case of memes per se (or when we test drugs on lab mice because they are similar enough to humans), it would seem that we are engaged in a kind of modelling. Only, both the models and the modelled object both exist. I'm not sure how defensible this position is, but it seems quite useful for our current discussion.

A relevant example of how the distinction between models and paradigm cases can be blurred in a useful way is @hawaidolphino and the rest of Clusterduck's "Meme Manifesto" project:

Their project seems to embrace the abstruseness and overcomplication of "meta-memes" as part of the medium and the aesthetic, which I think actually works well in the ongoing physical installation.

It's becoming increasingly harder to research memes without thereby engineering new memes in the process; making and posting memes is also becoming more viable as a research practice. I think we all ought to embrace that, and start testing research methodologies based on this new way of thinking about research models.

I can't find the exact quote but I believe it was D&G (and if it wasn't it was some other philosopher discussing them), making the point that one shouldn't necessarily always attempt to find a rhizomatic understanding, because sometimes an arborescent analysis reveals more. Forgive me for butchering the words but it went something like 'We can find rhizomes just looking at the roots of a tree just as we can find arborescent structures in certain views of a [insert something typically considered rhizomatic]'.

Essentially I think this makes the case that our power is not in finding the 'correct' model so much as we have the ability to swap models around as we see fit. Different models tell different stories and thus drive different actions. Considering that, I think it could be fun to apply models to create meme understandings that are deliberately 'bad' to create new understandings, much like Debord (or one of the Situationists, anyway) [I'm so sorry my memory is all over the shop on this thread] 're-reading' Paris by trying navigate using a map of Munich.

Following this train of thought, music theory I was taught in high school uses a macro-model called the 6 concepts of music, which are: duration, pitch, dynamics and expressive techniques, tone colour, texture and structure, which are all different models that can be applied to interpret music (or any sound, really). These 6 concepts aren't perfect as they overlap, but it's in that overlap that they really shine. The way my teacher explained it was that if one concept is like looking through blue glasses, and another is a pair of red glasses, when you take a red and blue lens out of each you can see in 3D.

Anyway when I get around to it I might start a creative thread or something for 'mis-mapping' the 6 concepts of music to visual memes. That would be cool I think.


That's brilliant. It doesn't seem particularly important to always restrict ourselves to a subset of memetic phenomena (e.g. memes in situ, like posts on social media, as opposed to memes that a researcher made themselves) or to a particular ontology of memes (insofar as different ontological beliefs about memes result in different usage of memes by memeculturalists, as well as the researchers).

I think that much of contemporary meme studies researchers hold back on the fact that they know a lot more than researchers are supposed to know about subject matters (in order to maintain the distance appropriate for positivist deference to logic and data). Yet, it's certain that our understanding of memes will influence the direction which memes take. Taking the analysis of any given meme in one direction over another, therefore, seems not so much a matter of picking one correct view, but rather of experimenting with different potential combinations of such vectors of knowledge that can be interpolated to further direct both memes themselves and the research of memes.

1 Like

I tend to side with the Bookkeeping alternative. I think on the same level as memes are a new form of communication that depends on new conditions (the internet and its framing of subject relation - the "wow that is LITERALLY me" effect), they're the resurfacing of what id call communicative "needs" that have existed for ages, culturally. I think the cut-up/remix model of memes for example is not so different from the kinds of instincts or desires that would have lead someone to, say, plunderphonic music, mail-art, relay novels, fanfiction or what have you.

WITH THAT SAID, the idea of memory is a topic i am very interested in on basically all fields i encounter, so that may be just me retrofitting it into broader theories on art lol.

I do find the idea of taking a non-standardized concept like memes into a laboratory-like condition to be very compelling!

1 Like

Yeah, i think that last sentence is a good reason why im on the side of memes being a kind of model-as-bookkeeping. I feel like an interesting aspect of the rise of those gigachad-wojak-virgin-doomer type memes was when people were saying how they're basically old school ragecomics with new faces. What interests me in that comparison is that ragecomics were created as a form to describe a humorous occurence in the meme-maker's life, but in a controlled environment that can be repeated, exaggerated and create its own sort of internal logic that can relate to other rage-comics, but not to the "real life" situations they're meant to describe (see how condensend the kind of characteristics and personalities the "derpina" or "me gusta" characters became, where there was little to no context necessary for us to understand what the punchline would be).

(Sidenote, but i sometime last year wrote a little text on "leftist wall of text" memes that may be of interest to the idea of trying to extrapolate a formula of communication out of memes, and the percieved dryness of the affair. I can make a separate thread for it since its quite a big boy text, so lmk.)

Very well put - which is why i think the kind of contingent "mememaker as the god of its own storytelling" aspect i mentioned above is so interesting.


I like the way that you formulate your position: memes are a new form of communication that depend on new conditions, but resurface latent communicative needs which have motivated earlier modes of communication. I think this true in a useful way, meaning that it enables further inquiry in more promising directions than other theories can. Namely, it preemptively deals with hype-driven metaphysical mistakes: when we theorise about novel entities, it's easy to attribute too much causal power (with regard to associated phenomena) to the novel entities themselves. That leads to the kind of monotheoretical, siloed discourses which @cris.v.gutierrez criticised above.

One of the last papers in the Journal of Memetics (Lissack, 2004) attempts to redefine memes along similar lines:

If, on the other hand, memes are redefined such that the evolutionary selection process is no longer an aspect of the ontology of memes but rather of the environmental niche (cf. Laland & Odling-Smee, 2000; Laland et al., 1999; Odling-Smee et al., 2003) of which the memes are evidence, then the field may have other avenues of advancement and a potential relevance to managers. Such a redefinition would entail recognition of the relationship between a given meme and the context of the social and ideational environment of which it is an affordance and which it demands be attended to. Memes in this casting are a label for successful boundary object indexicals and lose their privileged status as replicators. Instead, the replicator status is ascribed to the environmental niches and the memes are their representatives, symbols, or semantic indexicals. [note 1]

With this definition, memes are repackaged as symbols and their impact on management is not that of a viral contagion but rather as an indicator of success and change in environmental niches. If an environmental niche has an important managerial role, then paying attention to its symbols and affordances can also be important. Memes are stripped of their casual role and instead become semantic tokens capable of evoking ascribed meanings. It is the process of evoking and the efficacy of the meme as the trigger for attention, recall, and repetition of the ascribed meaning that give memes relevance to managers.

When memes are so redefined, as symbols that index their environments and mediate our interaction with the environments by representing them back to us, it's our own latent communicative needs which drive their evolution (and not the other way around, with memes driving our evolution). This is one way in which some classical memeticists modified their theories to escape unverifiable and circular theories of meaning and language (viz. thinking is memetic competition, hence we cannot trust that our own thoughts would tell us anything true about memes either, since they are themselves memes).

Meme scholars working in the 2010s within the memeology tradition offered similar criticisms of memetics and emphasised "agency". Limor Shifman's "Memes in a Digital World" is probably the most influential work in this aspect, but nevertheless ascribes unique traits to memes as cultural replicators, albeit aiming to do so from "a purely social/cultural perspective" (ibid.). Shifman is ultimately a Dawkinsian in this regard, and considers memes to be "the building blocks of complex cultures, intertwining and interacting with each other". And she makes a great Dawkinsian sort of point (2013):

But if we differentiate between content-, form-, and stance- based memes, we might discover that so-called “innovations” are sometimes old ideas or communicative practices in new textual gowns. This framework may therefore allow us to think about the delicate balance the between diffusion of innovation and the diffusion of tradition . Such a trajectory seems to fall in line with Benjamin Peters' (2009) recent suggestion to reconceptualize “new media” as “renewable media”: Rather than just asking if a medium is new or not, it might be more rewarding to trace the various forms that certain technological ideas take in the course of history.[4]

And so, there's a continuity of what we might term latency-based thinking, between classical memetics and the study of Internet memes (which Shifman takes to be instantiations of memes and Dawkins takes to be memes). It neatly loops back to the question of what is being replicated and what is new; we can also consider which latent communicative needs, if any, are new. This suggests to me that it would be productive to examine what epistemic needs are latent to meme studies as well as the nascent meme research networks therein. Not a psychoanalysis of researchers, but a metaepistemology of meme studies that would shed light onto how best to bridge the subdisciplines and avoid reinventing the wheel (and repeating the same attendant mistakes).

1 Like

This is a very limited context, though. The models become correct post hoc, but only for a small set of objects-of-study (i.e. the ones which were made to fit the model in the first place). The number of cases described by the model expands in scope, but it isn't clear that the model begins to describe cases that pre-existed said toy any better. Of course, I don't think you're saying quite this — and I'd have to actually sit down and read Negarestani to comment on how closely we are reproducing his argument — but nevertheless, I take the point that he could be saying more than one thing, or that, at any rate, we can glean more than one thing from him and our discussion:

a) Toy models can help us clarify our assumptions and orient our investigations; and,
b) If we're setting out to create something, it will typically start out in conformity to our models, and hence furnish them with referents.

They aren't the same, but I think they are all intimately tied-up in our practice of modelling (that is, loosely, our habit of creating/building/observing, and then drawing analogies and conclusions therefrom). I'd suggest that paradigm cases are exemplars,1 insofar as either a model ends up being constructed from a 'paradigm case' (i.e. the 'base-case'), or because said case is taken is an instructive/educational instance of the things modelled.

As regards drosophila: it seems to me that tests on 'lab-species' like this presuppose some degree of analogy or 'likeness', which I figure is similarly presupposed when we are engaged in this thing called modelling. It's just that, in these situations, we don't initially set out to 'construct' the fruit-fly, or rodent, or whatever. Analogy is the main operation here, not necessarily abstraction; the latter seems to be its own practice of finding higher avenues across which to analogise.

Doesn't this happen all the time? Maps and territory (ironically, a paradigm case in their own right), essentially have to coexist in order to be useful, and arguably, accurate. I'd also say it's an open question as to how purely conceptual models — which is to say, concepts — exist. Likewise mathematical models and identities. Ultimately these are all metaphysical questions; for my money, I'd attribute at least a kind of virtual existence that allows predication, belief, causation, et cetera. What do you think? Have I totally misinterpreted what you were saying here?

Yeah, I think this is an important point. To my mind there is this vaguely Zen-esque illogic at play, where too-rigidly and dogmatically clinging to "being rhizomatic!!1!" itself becomes strictured and arborescent (although, as you/they point out, this is sometimes helpful). As D&G also suggest, full destratification is fatal:

If you free it with too violent an action, if you blow apart the strata without taking precautions, then instead of drawing the plane you will be killed, plunged into a black hole, or even dragged toward catastrophe. Staying stratified—organized, signified, subjected— is not the worst that can happen; the worst that can happen is if you throw the strata into demented or suicidal collapse, which brings them back down on us heavier than ever.

What a great comparison. In a more practical (boring!) context, it reminds me of a Paul Graham essay2 about the underappreciated value of an outsider's view. We can probably understand this as outsiders bringing new insights by reading situations with their own "maps", which were conditioned on different stimuli. The most productive generalists could probably be argued to have high familiarity with a number of different maps and models, and to have the nous to be able to swap between them as necessary and appropriate — and perhaps also identify similarities and dissimilarities between these maps, which could help highlight potential improvements, and yield and understanding of how to manipulate them as a creative and productive endeavour.

Again, though, they should be careful not to be plunged into a black hole bereft of signification.

1 There's a whole nexus / cluster of related terminology here: exemplars, models, analogies, paradigms, abstractions... It's really difficult to escape!

2 Probably this one? I can't quite remember - he's written a couple of times about the value of outsiders and generalists.