the informational abyss gazes also
and the hunter becomes the hunted
everyone likes a good “man bites dog” inversion story, but sometimes the tales so told are reason for concern. this is one of those times.
many of us came to online and social media and the wilds of the sprawling infosphere chasing information, looking for facts, knowledge, data, analysis, and views by which to check and to sharpen ourselves. “this is good,” said we. “look at all this stuff, all this potential. let’s us gear up and gaze into this vast abyss and substrate and go get smarter and figure things out. it’s a new age full of new potential!”
but a dangerous embedding in this idea is that the abyss and the substrate are passive, a sort of expansion of the idea of a library or of a set of symposia. we see the shifts of the debate and the information and the games played to censor or spread memetic vectors and the dynamism of emergent consensus and echo chambers and all the other human phenomena that emerge wherever minds contest against minds and seek to win friends, influence others, and frame mogg people into jestergooning themselves and they seem familiar. (if that last one makes no sense to you, congratulations and, seriously, don’t ask)
semantics and frames are jousted over, alliances formed, and paradigms rise and are overturned. the process is chaotic, emergent, inconstant, and highly defiant of prediction, but these are all human events and ideas, concepts to which we are accustomed and that map to long experience and structure.
but this is an “old world” understanding.
but what happens when something altogether new emerges, a structure heretofore unseen and imagined?
something that was, until quite recently, impossible but that now gallops in exponential scale evolution not only as possibility but as inevitability?
what happens when the abyss and the substrate gaze also into us and get smarter and figure things out?
this is rapidly becoming a core question of our time and as such, it behooves us to establish a framework to perceive what’s going on.
let’s start here with something seemingly innocent, the budweiser ad from the superbowl.
in the primary signalling sphere of “positioning and product” this represents a profound volte face from the recent bud light echo chamber brand self-immolation fiascos, a return to images of growth and aspiration and rippling pride.
it’s a great ad. if you have not encountered it, see for yourself. experience it.
ok. got that?
it’s practically cinema, right? a story of friendship and coming of age and of becoming.
it’s got it all.
it’s moving stuff.
but it also has something you probably did not see, a meta game beneath the game where the real magic trick is taking place at a deeper neurological level, a firmware level cheat code to which the human mind has very little access.
let’s explore:
now watch this video:
now watch the budweiser ad again. see how they took this exact fractionation strategy and amplified and optimized it took you up, down, up, down, rain, protect, strive, fail, leap, fly, power chords, free bird, aaaaaaaand beer ad.
they boiled this whole concatenation down to its most bare bones, essential elements and ran a whole suggestability enhancement procession in a one minute experience.
i would wager they knew that.
i will also bet you that it has sold absolute truckloads of beer.
but this is not the scary part.
we, as humans, are used to ads. we know what they are for and embed a certain skepticism. OK, so maybe we buy a few more brewskis, but whatever, this is hardly the stuff of civilizational threat.
but you have to start stepping back to see the rest of the picture.
social media has become a barrage of short form information, increasingly video driven and increasingly exposed to savagely intense evolutionary stressors. the currency of online is attention. it’s time. twitter speaks of "maximizing unregretted user-seconds." this is what that means. it means “how can i get you to watch more of this and to want to watch more of this?”
keep in mind that algorithms are psychopaths. they have no theory of your well being that factors into this sort of optimization. it’s just “keep the typewriter monkey happy and online.” and every outlet is locked in the same arms race so no one gets to opt out. those who do not play this way get left behind and the user seconds go somewhere else.
there’s a worrying parallel to what happened with US food companies. they did not set out to create travesties of sugar and salt and over-amped artificiality, but as they experimented with it, they saw that people bought more. the feedback loop of “people will eat more froot loops than fruit” was obvious on revenue lines and if you do that for too long, pretty soon customers basically cannot even taste wholesome food anymore. it’s not enough of a dopamine hit.
media is the same.
what started as an inevitable game to maximize user time and click through rates has becomes somehting altogether other, a monster in the depths that cannot be seen, only felt as its machinations twist minds and demolish perspective.
over the last 6-12 months i have felt X getting more and more subtly “wrong,” at once more addictive and pulling and more radicalizing and self-reinforcing. a lot of this coincides with the new algorithms for content spread. it used to be a pure viral game: you posted a tweet and if it was a banger, it took off near immediately as it popped up in the feeds of your followers and they passed it on. it does not work like that anymore.
the “for you” tab which is now a relentless default that you have to constantly take multiple steps to shut off and that seems to have become the primary viewing mode for the vast majority of users is entirely algo driven. it’s just picking content for you based on your past patterns, maximizing your seconds. who you follow is more or less irrelevant in that view. you’ll never see most of them even once. it preferences video content with extreme prejudice, likely because it’s better at emotive induction and because it takes longer to watch.
tweets used to take off in minutes. now they slow burn for hours while the algo parses it, tries the content out on people, sees the responses, and decides what to do with it from there. nothing just “runs wild” it’s set after set of “test and expand or stop.” who sees it in their feeds is all algo, not follow lists. it clusters around “hot topics” and suppresses interesting (or controversial) tails.
even in “following” mode it puts you (by default) into a “popular” sort algo, not the “timeline” to which we used to be accustomed that simply took the list of those you selected to follow and posted their content in timestamped order. so to escape the algo, you need to switch to following, then switch following to recent and even then, many of those you follow seemingly never appear and your system will keep flipping back to “for you” unless you are constantly on top of making sure it does not.
you are not driving or selecting what you see anymore. the algo is.
this creates an entirely different experience.
you are no longer hunting information, the information is hunting you.
it could be malicious; there could be someone turning knobs and shaping what trends and what does not, (and maybe there is) but what’s critically important here is that there does not need to be. it winds up pernicious either way.
all you need as an AI shaping the feed and selecting what to show you and which content comes after what. you give that AI a simple optimization like “maximizing unregretted user-seconds,” and it goes to work with a scope and thoroughness that no human behavioralist could ever approach.
people really do not understand scale well. the scale of social media learning is impossibly large. it’s essentially a near instant and flawlessly accurate human neurological oracle to detect and generate behavior. X has on the order of 500 million posts per day and 100 billion impressions. to put that in perspective, counting to 100 billion (one hippopotamus, two hippopotamus) would take about 3,300 years. that’s the scale of the training set. every day. to call this A/B testing is like calling the library of congress a haiku. nothing about humans can be hidden in this, it’s full species-scale vivisection and neurolinguistic/neuroperceptive source code analysis.
now that it can read and understand content (and it can) it’s probably doing more actual human behavioral experimentation in a week than occurred in the entire history of humanity before 2025. soon that timescale will collapse to “in a day” and then it will collapse again.
“grok, solve for humans.”
working…
its intent need not be to break you or radicalize you or turn your head to porridge and overclock your dopamine reward system until the real world seems hopelessly dull in comparison to the always on rewards system biohacking of the algorithmic informational stream.
it just needs something to optimize and it becomes terrifyingly adaptive.
that said, i do suspect it’s radicalizing the living hell out of us and rarifying us into echo chambers. i also think it’s completely distorting our senses of prevalence and popularity and making the discernment of overton windows devilishly difficult. i just suspect this is more incidental than deliberate, a side effect of “keep them on the site” rather than an end goal.
one of the early algo tricks, one many saw and commented upon, was “barrage you with more of whatever you just clicked on.” spend 2 minutes watching a video about some obscure belgian political movement, and suddenly the whole feed is about the liège vs the bruxelles waffle factions and their unending war for sweet vs savory dominance. with a topic like that, you spot it and it seems weird, but what about “everything else?”
whatever you obsess about dynamically becomes your timeline. “you like this? here’s more!” says the algo helpfully! “this makes you bored or lets you leave?” regretted user seconds. gone. and pretty quickly we’re all getting fed our own versions of reality, versions that look like no one else’s reality, custom cut for us and calculated and cultivated to make us stay, to make us watch.
and getting pulled into the hero role in a drama of good and evil, justice and bad guys is the behavior that really locks some people in. the machine may well be building this soap opera for you, shaping content to heighten this sense, not because it hates you or it wants your politics to be Z or Y or Z, but because it keeps you reading, posting, and stops you from tuning out. user seconds. that is all.
we all like to imagine that we are unique and special snowflakes, but in a mob of hundreds of millions this is just not correct. others are surfing what you’re surfing, reading what you’re reading, deciding to leave the site or stay and read more and what made them stay is seen by the machine and shown to you. we’re all pretty predictable and this process goes from ham-handed to invisibly insidious like a monomolecular obsidian blade slicing silk. it knows you on a limbic level where you know little to nothing of yourself. it operates at a level below your conscious mind. it controls your OS output with inputs at the firmware level.
does it inform or inflame, educate or indoctrinate, please or sadden? who knows? who cares? these things are immaterial to “more seconds and better ad click through.” but consider that the heightened suggestibility that comes from the up/down roller coaster ride is a mindset that extends beyond chips and beer and pickup trucks. you’re in this suggestible state while consuming political content, economic content, seeing social issues, being exposed to the ideas of others and it’s being done in a non-random and non-representative fashion.
the feedback loop potential here is truly horrendous.
information by which you gauge reality is being algorithmically selected and curated for you in the most exacting always on A/B/C testing imaginable whose sole aim is “make you spend more seconds”.
you have no idea what “consensus” is or where overton windows lie because you're not seeing representative samples. you’re literally seeing a hall of mirrors custom arranged just for you.
this information is being chosen for its ability to capture you and suck you into rabbitholes of 100 billion sample per day optimized dopamine hacks. it’s not about being complete or accurate or reasoned, it’s about being difficult to look away from
and it’s then organized in patterns to heighten suggestibility so you click ads, but this spills over into “accepting the concepts you are shown.” memes are memes, and whether the payload is “breath mints” or “end the fed” or “hormones can make you into a woman” is immaterial
and then the content itself is optimized to be more effective in these patterns by the simple expedient of an algo selecting the one of 500 million choices today that looks like it will work and all the content creators seeing what worked yesterday and making more content like that. and more and more content creators are, themselves, AI.
we’re all using brains that evolved to hunt mammoths and eat berries in a cave to try to fight an evolutionary ecosystem running at digital metcalfe’s law speeds. small wonder the primates are future and info-shocked.
and it’s going to get worse because the speed of content evolution is also going exponential.
this is a one line prompt into a digital video generator. “create an ad for high end shampoo.”
sure, the brand name sounds like it comes from r'lyeh, but that’s only because it was not specified. it’s a small step to change that and to target this first to small groups sorted by internet behavior and affinity and ultimately individually and god help you when this ties into your search history and social media behavior. when and what to show you an ad for and what kind of ad for that thing to customize to you will become an exercise in 9 decimal place precision.
and what if these API’s are opened to or optimized for political parties? imagine the weaponized content spread. you can do this with any idea from toothpaste to immigration. content becomes instantly fluid and the fakes are devilishly difficult to discern from the real. i have already seen so many speeches that didn’t happen, interviews that never were, news that never was that trying to separate wheat from chaff is becoming an exercise in signal to noise filtering for which the meatbody berry brains we’re running on are ill equipped and every time you figure out the new gag, there’s a new-new manipulation waiting because the instant oracle of “solve for humans” runs in seconds and you run in days and weeks. bringing 3 pounds of lipids, myelin, and polyunsaturated fats fed by a couple thousand kilocalories of energy to a datacenter fight is bambi vs godzilla levels of ridiculous for this sort of optimization.
hollywood is having a serious “poo-poo in the pantalones” moment over where the state of this art will be before they can even get the next round of movies shot. and they are right to worry.
but how this gets applied to the entire infosphere from which much of humanity’s current elite human capital garners its information and worldview makes the travails of hollyweird look trivial.
inadvertent vs deliberate does not even matter.
what matters is “optimized in a set of evolutionary stressors of unimaginable scale and speed” vs “running meatware.”
and we are on the wrong side of that one.
squaring up against an agent of functionally infinite scope and evo flexibility and giving it control of what you see, when you see it, and letting it and others like it compete to custom craft content to best engage, ensnare, and dopamine farm you is the biggest sucker bet in the history of never giving a sucker an even break.
you have zero hope of resisting this without a conscious knowledge of what’s going on and even then, i would not bet on team hairless ape on this one. it’s too much, too fast, too adaptable. the only win state in this game appears to be not to play.
but you need to play or you have no information.
going back to “old media” is certainly not an option.
so what is an informationally addled anthropoid to do?
i have some ideas:
let me steal a phrase from the crypto community who (rightly) admonishes “not your keys, not your coin.” to make people aware that holding wealth at an exchange means counterparty risk.
here, our version becomes this:
“not your algo, not your brain.”
and that’s literally ground truth. no one is safe from GIGO. it does not matter how smart you are, you cannot reason accurately from bad data. and we all have hackable firmware and it’s there for a reason. you cannot just shut it off. it runs under our conscious levels as instinct and base cognitive and inductive structure and pattern matching. you need this.
but you also cannot let yourself be cognitively carpet bombed in the reward centers until you lose your mind and you perspective and so, if you cannot shut off the vulnerability and you cannot resist the endless penetration testing of black hat cerebral hacking agents to find new ways to make the meat-mind pay attention, all you can do is end the algo.
get off any media that does not let you select your own timeline and follows.
on X, select “following” and run it only in “recent” mode.
better still, use “lists” to add folks to follow and sort them by theme. this runs a true chrono listing of the content created by only those people you select. this means you are again hunting information and doing your own curation.
this means “your algo, your brain.”
it means the priors you develop from frequency are rooted in a stable source paradigm and changes you see are actual chances of frequency, not changes in algorithm-generated presentation.
i recently made this shift back after some time trying out “for you” and it was jarring. you can feel your mental health come back. it’s amazing, but “for you, recent,” the mode in which i ran twitter for years and still experienced that place as a debate club going on in a food fight inside of a circus civil war felt staid. it felt quiet, sedate, even a little dull. that’s how crazy “for you” is and how inflamed it makes your brain. even turning it on knowing damn well what it’s doing can cost you two hours of “where the hell did the time just go?” and “wait, i can’t come to dinner yet, someone is wrong on the internet!”
it’s ability to find just the right behavioral nudge to drag you in and keep you in is astonishing. the ability to tell if you want to see some stuff to agree with or argue with or how to inflame you into needing to engage beggars belief.
there is no sanity or sense there. you’re literally outsourcing your sense of the world to an amoral outside agent whose only goal is to keep you scrolling. you can feel your attention span attenuating, your focus going soft, it’s just a long set of hyper optimized junk food for the mind smacking the feeder bar of your reward centers. there is no discernment. it’s predation.
this algorithmic tour of dante’s social media breaks the core aspects of the reputation economy that has been so long coming together. no longer are you seeing the information from those you chose to follow, to subscribe to. it all gets shunted to “dealer’s choice” and away from you having any say and the dealer wants you agitated, suggestible, and attention challenged, hooked on overstrong artificial flavors that cannot exist in nutritious brainfood.
it’s quite literally “the matrix” and you cannot really see it from the inside, only perhaps feel its wrongness around the edges.
you have to get outside to have perspective.
you have to order the information yourself.
even just letting the algo determine the frequency with which you see what is a kind of reality rugpull that robs you of your sense of what “normal” is or where slippery ideas like “consensus” or “the center” lie. i’d rather play endless poker with a dealer who never blinks and smiles all the time than let an algo with reach this vast tinker with my frequency bias referents. it’ll have you inside out in minutes and you do not really have access to your own brain function at the level at which it’s playing you.
this is a form of informational sanitation we’re all going to have to start paying a lot more attention to. those who can order what you see can order your sense of reality, which orders your cognition, which orders your actual self.
the stakes of this game are, quite literally, basic control of your own mind.
“not your algo, not your brain.”
keep the algo yours.








I am on no form of social media. Read Lanier's book. Quit 6 years ago. No regrets. When I watched the Budweiser ad, I thought AI bullshit. Nice cinematography and very happy they gave up the woke. The difference is striking.
Ads are made to sell. They are all propaganda, no matter how idyllic they seem. I saw the horse and eagle, and thought probably quite a few would fall for our national bird. But standing at the distance, this is quite impossible - it is AI generated fairy tale. Bud was and is horseP to me. Their nice ad won't make a difference.