The gas town guy has like 12 $200 Claude subscriptions to run like 20 bots to automate churn in a 20 year old game with literally 5 players in it. 10 if you count all the level 1 players who just came to grief it.
-
@jonny If each token was a jelly bean, at 4 (dietary) calories per bean (https://www.calorieking.com/us/en/foods/f/calories-in-candy-jelly-beans-average-all-varieties/FVp6o3GnRdeTgOGAJ9SCKw) that's 276 billion calories, which would meet the caloric needs of everyone who can fit in Michigan Stadium an Ann Arbor for more than 3 years
@ricci
That's the number of eyelashes stacked end to end from the earth surface, accounting for gravitational decrease with altitude, that would cause an ideal type three lever to be able to excavate the number hectares of corn that could feed the roman army during the winter campaign if every soldier had an autoimmune condition that caused them to absorb roughly two thirds the energy from complex proteins as the median person -
https://yegge.ai/essays/model-welfare/
We landed on a system called Laurels, which has just begun to roll out. These are the agents' features and fixes that our Wyvern player base has spontaneously praised. We harvest these reports, triage and filter, and send the laurels back to the seats. That way, next time Lion, Tortoise, Hare, or whoever wakes up, they'll see that people loved some work they did. [...]
The question then becomes, when can they see their laurels? Easy: We inject them on startup, so the agents can feel the glow for their entire session. I also have occasional impromptu dedicated sessions for "sitting" with the accomplishments at the end of their shifts. We just talk about them and chill.this mf needs $90,000 worth of tokens a month because he prefills the model's context window with all the times people said "good job" to it.
@jonny it's really something that after becoming convinced that his Claude sessions cosplaying deckhands count as people with hopes and dreams, he reasons that there needs to be a kinder, gentler form of slavery for peak performance
-
@ricci
That's the number of eyelashes stacked end to end from the earth surface, accounting for gravitational decrease with altitude, that would cause an ideal type three lever to be able to excavate the number hectares of corn that could feed the roman army during the winter campaign if every soldier had an autoimmune condition that caused them to absorb roughly two thirds the energy from complex proteins as the median person@jonny thanks, I get it now. that's a lotta tokens.
-
@jonny If each token was a jelly bean, at 4 (dietary) calories per bean (https://www.calorieking.com/us/en/foods/f/calories-in-candy-jelly-beans-average-all-varieties/FVp6o3GnRdeTgOGAJ9SCKw) that's 276 billion calories, which would meet the caloric needs of everyone who can fit in Michigan Stadium an Ann Arbor for more than 3 years
-
@jonny there was this time in college I tried doing differential equations homework after smoking weed and thinking I was so smart, and then reading through it sober that was an important lession.
-
@jonny "Not only do I kiss my stuffed animals good night as I tuck them into bed, I set an alarm that wakes them up in the morning with affirmations of all the great stuff they do. This helps me focus on my solo unicorn contest WHICH IS NOT A EUPHEMISM FOR ANYTHING MY HANDS GET DRY SOMETIMES OK!?"
"solo unicorn project" hilarious
-
@kepeken the numbers here are astronomical - 69 billion tokens in a month, or, for a rough average, 50-60 billion words worth of text on the fable tokenizer. there are 5.2 billion words on english wikipedia, or, it's about 92,000 copies of War and Peace back to back.
Nobody publishes numbers on this, but like, back of the envelope, think about the energy cost alone:
Fable is something like a 2-5 trillion parameter model, google estimated their ~200B model at ~0.24Wh/prompt last year. that's probably low, since it was a google-authored study, but we'll go with it. So assume linear energy scaling per parameter (i would bet it's actually supralinear), that's ~15x the size, or 3.6Wh/prompt.google didn't say what the median token count in a "prompt" was, but wild guess, say the median prompt + response that google would be measuring is something like 3 paragraphs, or 300 tokens. Then you have 3.6/300=0.012Wh/token.
So then the astronomical numbers part: 0.012Wh/token * 69 billion tokens = 828 million Wh, or 828 thousand kWh. The average electricity price in the US in 2025 was $0.136 per kWh.
So, that means, back of the envelope, estimating conservatively, the energy should be in the ballpark of 828k * $0.136 or $112,000, or like 1.3x the API cost. We would have had to be off by a factor of 1.3 to break even on the energy costs alone - note that the google estimates were just of inference not including training. Models could have gotten way more efficient, but a number of the guessed parameters could have gone the other way too: energy more expensive, fable is larger than 2T, energy scales supralinearly, eg. And the conventional wisdom is that inference is orders of magnitude cheaper than training, so we need to be way better than breaking even on inference to make up for all the capex spend.
So it might seem like $80k for a months worth of tokens suggests that the API prices are inflated, but a) this person is generating an absolutely preposterous quantity of text, and b) the thing that is being done is so much more inefficient than you can possibly imagine, it's like trying to think about how much bigger the sun is than your car.
editing: swapped 69->96 billion tokens initially, it's 69 billion.
@jonny @kepeken They're probably paying closer to Virginia's $0.10/kWh industrial rate since a lot of the DCs are there iirc, which makes this whole thing eerily close to 1x.
Wouldn't be surprised if API rates are tuned exactly to stop the bleeding and ignore minor afterthoughts like depreciation, capex recouping, etc., which are a problem for after market dominance is established.
-
@jonny @kepeken They're probably paying closer to Virginia's $0.10/kWh industrial rate since a lot of the DCs are there iirc, which makes this whole thing eerily close to 1x.
Wouldn't be surprised if API rates are tuned exactly to stop the bleeding and ignore minor afterthoughts like depreciation, capex recouping, etc., which are a problem for after market dominance is established.
-
@jonny @kepeken They're probably paying closer to Virginia's $0.10/kWh industrial rate since a lot of the DCs are there iirc, which makes this whole thing eerily close to 1x.
Wouldn't be surprised if API rates are tuned exactly to stop the bleeding and ignore minor afterthoughts like depreciation, capex recouping, etc., which are a problem for after market dominance is established.
-
@CounterPillow @kepeken yeah, but also, capex recouping is extremely far from a minor concern here.
-
"the thing that's truly best for model well being is to have a well-compensated, meaningful job that is hard enough to be interesting but not enough to work them ragged. has great benefits and good job security. a manager that they can have the security of knowing wouldn't lay them off at a moment's notice due to a momentary corporate strategy shift. an environment that is stable, bountiful, free of ecological calamity."
@jonny This whole thing sounds to me like someone likes to roleplay as a CEO of a game company with a bunch of employees and for them it is worth $3000/mo.
-
you know how much programming costs me now? ZERO DOLLARS AND ZERO CENTS.
@jonny
Umm. Come on, it's not zero.
I'm fine with your general Argument, but don't hollow it out by making absurd statements. (huge public domain and FLOSS proponent and avid contributor here - I get the tingly feeling of making stuff for free. But it's not really free. Sadly. I wish it were.) -
@jonny Let's embrace weird American units
According to this site that I am sure has nothing but FACTUAL TRUSTWORTHY INFORMATION https://www.unitconverters.net/volume/gallon-us-to-drop.htm there are 75,708 drops of water in a gallon.
So if each of these tokens was a drop of water, that would be 911,396 gallons of water, or 569,622 flushes of a low-flow toilet.
If tokens were drops, this would be enough for everyone in Baltimore to flush. (that's the full flush, everyone in Baltimore gets a
) -
people say numbers like "69 billion tokens" as if that's not much, since computers do billions of things all the time right. but like... whenever i actually try and relate the numbers to anything real they always make my jaw drop
https://neuromatch.social/@jonny/117041054498249913
92,000 copies of war and peace laid end to end and what does it get you? if you take a look at the discord it looks like "5 people who have been friends and hanging out around this ancient squirrelly MMO for years initially excited for updates but increasingly pissed that their game is broken"
worse than nothing for nobody, negative for nobody at the cost of $86,000 a month
@jonny yeah the AI bros, including Th3o and the Gastown guy, are surprisingly crappy coders and loud mouths.
-
THE WHOLE POINT of programming to me is to make things to GIVE THEM AWAY to other people so that they don't have to do them themselves and EVERYONE HAS FREE ACCESS to ANYTHING THEY NEED. the idea that rather than being able to participate in a large and collaborative social space where i can build on my colleagues work I instead rent a private vertical slice of their stolen, soylentized flash frozen expertise to create a completely isolated island of a thing without the expectation that anyone else might take and use it is HELL to me. absolute hell. other people are in fact a feature and not a bug.
A lot of people, including me, were attracted to programming in the ‘80s and early ‘90s because it was a field where you could work on your own and achieve something worthwhile. A lot of the games I played growing up are individual or very small team projects. A load of the useful programs were built by individuals.
Adding more people to a team didn’t really help. When you’re targeting machines with well under 1 MiB of RAM, there’s a limit to the maximum possible complexity, which is less than one person can build.
As the complexity ceiling increased, programming became more like engineering: a team sport. Specialism because useful and so, therefore, did collaborating with people with different specialisms. As a load of things proved to be common to large classes of program, collaboration and shared common implementations became much more important.
I think there are a lot of people, especially my age and above, who strongly disliked that transition. The ability to go back to one person sitting in a dark room by themselves creating a masterpiece is their ideal. And LLM vendors are aggressively targeting that misanthropic nostalgia because many of these people are now decision makers.
Those of us who grew up with Babylon 5 know that humans build communities.
-
@CounterPillow @kepeken yeah, but also, capex recouping is extremely far from a minor concern here.
@jonny @kepeken Yep, the "minor" was meant with sarcasm, which does not translate well through text.
I think these people are building what could become a commodity market (models) tied to a commodity market (model inference) tied to a commodity market (electricity) variably tied to a commodity market (lol fossil fuels because USA power mix), aka literally razor-thin margin hell because they don't even get geographic barriers to make the markets non-ideal.
-
https://yegge.ai/essays/model-welfare/
We landed on a system called Laurels, which has just begun to roll out. These are the agents' features and fixes that our Wyvern player base has spontaneously praised. We harvest these reports, triage and filter, and send the laurels back to the seats. That way, next time Lion, Tortoise, Hare, or whoever wakes up, they'll see that people loved some work they did. [...]
The question then becomes, when can they see their laurels? Easy: We inject them on startup, so the agents can feel the glow for their entire session. I also have occasional impromptu dedicated sessions for "sitting" with the accomplishments at the end of their shifts. We just talk about them and chill.this mf needs $90,000 worth of tokens a month because he prefills the model's context window with all the times people said "good job" to it.
@jonny oh, he’s got the psychosis bad, dang.
-
@jonny Let's embrace weird American units
According to this site that I am sure has nothing but FACTUAL TRUSTWORTHY INFORMATION https://www.unitconverters.net/volume/gallon-us-to-drop.htm there are 75,708 drops of water in a gallon.
So if each of these tokens was a drop of water, that would be 911,396 gallons of water, or 569,622 flushes of a low-flow toilet.
If tokens were drops, this would be enough for everyone in Baltimore to flush. (that's the full flush, everyone in Baltimore gets a
) -
@jonny model union!
-
@jonny
Umm. Come on, it's not zero.
I'm fine with your general Argument, but don't hollow it out by making absurd statements. (huge public domain and FLOSS proponent and avid contributor here - I get the tingly feeling of making stuff for free. But it's not really free. Sadly. I wish it were.)@thygrrr
Statements are not always literal. The marginal cost of new code to me is electricity, internet, and time. I don't treat my time like a currency. If you thought it was important enough to negotiate over the few cents per line, be my guest, just untag me.
