Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Tokenmaxxing is dead, long live tokenmaxxing

12gramsofcarbon.com

Post reply on HN
Print view View original

191–200 of 315 posts

Re: Tokenmaxxing is dead, long live tokenmaxxing

#191

Post by TeMPOraL »

TeriyakiBomb wrote: post #186
cyphar wrote: post #50

Earlier quoted context omitted.

Why would a carpentry shop buy hundreds of thousands of dollars of power tools without consulting with their employees to see what they actually need to get their job done more effectively? The logic of buying the tools then forcing the employees to use them "or else" is completely backwards in any sane world. (Of course, we've all had bosses that went to some marketing seminar and come back having been tricked^Wsold…

To use your analogy again, it's kind of like the shop boss buying everyone a table saw and then saying "The best way to use the table saw is just experiment with it, it's the fastest and most accurate straight cut we can get - the future is table saw."

Yes, this is, in fact, how adoption of table saws and other such tools looked like, while they were still new tools. The basic form and function was established and its utility proven in both testing and early adopters, but as new kind of general tool on the market, every user from "early majority" was still writing the operating playbook for their specific shop conditions and kind of work they're doing.

So yes, it's a great analogy. We're right now well in the stage where bosses say, "evidence is in and conclusively shows this is useful for us, now the job is figuring out exactly how to work it into our particular business".

Re: Tokenmaxxing is dead, long live tokenmaxxing

#193

Post by TeMPOraL »

oblio wrote: post #188
woah wrote: post #105

Earlier quoted context omitted.

> Why would a carpentry shop buy hundreds of thousands of dollars of power tools without consulting with their employees to see what they actually need to get their job done more effectively? Are you suggesting that changes to new production technologies are always driven bottom up by line workers? I'm guessing that historically that's rare.

Historically that rarely happens because industrial equipment is/was generally too expensive for the average worker to purchase on their own, plus workers usually have a budget of roughly 0 to buy extra tools, especially expensive ones. But to give you an example, also roughly 0 companies made developers use Linux and still many developers choose it, so bottom up improvements happen in a decent chunk of cases. Nobody…

> so bottom up improvements happen in a decent chunk of cases. Nobody paid for PostgreSQL promotion. Or Python, etc.

It does, but for better or worse, it's an anomaly. Even now, maybe nobody was paid for PostgreSQL or Python promotion, but modern OSS tools and programming languages usually have a business backing it. Linux, too, wasn't commercially promoted until it was; RedHat isn't exactly a charity after all.

Conversely, no one paid for initial AI promotion either - ChatGPT exploded organically after release, and for the first year or two, companies had a problem because a good chunk of their staff, including especially non-engineers, discovered just how useful it was and wanted to use it at work, casually violating every internal policy, bylaw and even regulatory policies about data sharing. The massive spend on promotion - including first-party spend - came later, but at that point it was already obvious ~everyone is going to be buying it.

I suppose bottom-up vs. top-down may be in part about how mature a technology and industry is.

Re: Tokenmaxxing is dead, long live tokenmaxxing

#194

Post by surgical_fire »

SR2Z wrote: post #46
baconmania wrote: post #8

The implication that tokenmaxxing was an intentional and thoughtfully considered approach rather than blind hype-following by an overpaid manager class who are too far removed from value to understand the downsides of LLMs is hysterical beyond belief.

I really don't understand this take. If you're a carpentry shop that just bought power tools for the first time and you're worried that your employees are sticking with hand tools because that's what they know, then you look for sawdust. The goal isn't to have people work at converting wood into sawdust, the point is that if you wanna see if the tools are working you wanna see proof they're actually being used. I'm s…

Or... If you are a carpentry shop owner, you should understand what exactly the power tools you acquired are good for, how and if they can actually be used by your employees for them to do their job.

This, obviously, presumes that the person managing this hypothetical carpentry shop knows what they are doing. It's almost laughable.

In truth the carpentry shop owner manages on vibes, has no idea what employees do and also doen't trust them, and tells employees he wants to see a lot of sawdust in the workshop floor.

Re: Tokenmaxxing is dead, long live tokenmaxxing

#195

Post by koiueo »

behnamoh wrote: post #9

Tokenmaxxing was never a thing to begin with. Just because a few companies did it doesn't mean it was a widespread phenomenon.

Maxxing is just a catchy and imprecise name.

In my current company nobody forces you to use more tokens, but you're encouraged to write a 300 lines markdown skill.md which takes 8 minutes and costs 5 bucks to execute. That, instead of writing a 200 lines bash script doing all the same thing, but in a deterministic fashion, completing in under 5 seconds and costing 0 if you're not careful with rounding.

Re: Tokenmaxxing is dead, long live tokenmaxxing

#196

Post by IanCal »

rendaw wrote: post #163
aurareturn wrote: post #2

Tokenmaxxing was just a way to force employees to start leveraging AI in a meaningful way. For companies that have measured performance based on token spend, they can now dial it back. Employees have learned to leverage AI for things they wouldn’t have prior. Now they know what’s possible and what’s not. No one is stupid enough to always measure performance based on token spend and have unlimited budget. It was alway…

This is obviously wrong. Management has never cared about how engineers do their job. There's never been a push for any other productivity boosting technology: better languages, better editors, automated refactorings, paid code intelligence tools, etc. But suddenly AI comes along and the CEO says "all developers need to write code entirely with LLMs".

This is absolute nonsense. Management in many places cares enormously about productivity. What’s been a bit different here is a huge claim of improvement other companies are seeing (so you’re going to be left behind), alongside some developers going off and doing this anyway sending proprietary code hither and thither, alongside some devs railing against the very concept of using it. It’s also a wildly powerful tool and how to use it hasn’t been as clear (where does it provide value, where does it not, what can and can’t it do) so experimentation is really important.

Re: Tokenmaxxing is dead, long live tokenmaxxing

#197

Post by IanCal »

pier25 wrote: post #84
aurareturn wrote: post #2

Tokenmaxxing was just a way to force employees to start leveraging AI in a meaningful way. For companies that have measured performance based on token spend, they can now dial it back. Employees have learned to leverage AI for things they wouldn’t have prior. Now they know what’s possible and what’s not. No one is stupid enough to always measure performance based on token spend and have unlimited budget. It was alway…

It seems really absurd that anyone would encourage or even force employees to burn more money to see if maybe something works.

Why? This is literally what experimentation and prototyping is!

You spend money for a potential benefit. In this case it’s also a one off cost to find things that can save money over time.

Re: Tokenmaxxing is dead, long live tokenmaxxing

#198

Post by IanCal »

throwatdem12311 wrote: post #28
aurareturn wrote: post #2

Tokenmaxxing was just a way to force employees to start leveraging AI in a meaningful way. For companies that have measured performance based on token spend, they can now dial it back. Employees have learned to leverage AI for things they wouldn’t have prior. Now they know what’s possible and what’s not. No one is stupid enough to always measure performance based on token spend and have unlimited budget. It was alway…

This is an insane level of cope. The whole tokenmaxxing thing started because Jensen Huang said insane things like having a single engineer spend 250k in tokens or he’d fire him; and that OpenClaw was basically AGI. > No one is stupid enough to always measure performance based on token spend and have unlimited budget. Yes the people forcing these mandates absolutely are this stupid because that’s what people like Jen…

> like having a single engineer spend 250k in tokens or he’d fire him;

That’s not quite what was said there, he’s budgeting half a devs salary as token spend in a podcast and that if he had a 500k engineer who spent 5k on it at the end of the year he’d go ape shit.

Now you can say that’s wild, and sure, but this is not a standard c suite exec talking it’s the ceo of Nvidia.

Even ignoring other hiring costs this is essentially an argument that Nvidia top engineers should get more than a 50% performance improvement with extremely heavy AI usage. To me, that doesn’t seem like such an enormous statement. For the head of a multi trillion dollar company entirely driven by AI sales arguing it gives a useful benefit to engineers isn’t that odd and betting on a 50% improvement within Nvidia seems kinda normal.

> Seriously have you ever actually talked to an average C-Level about AI?

Yes. Single digit percentage improvements over time would normally excite them, the idea of cappable cost performance improvements that last which your devs actually want to experiment with and a cultural and customer expectation that you’re doing this is pretty enticing. Particularly during a time when tokens were heavily subsidised - isn’t that the perfect time to do it? Now that has ended there’s a huge focus on roi.

Re: Tokenmaxxing is dead, long live tokenmaxxing

#199

Post by JoshTriplett »

TeMPOraL wrote: post #170
JoshTriplett wrote: post #164

Earlier quoted context omitted.

Perhaps don't start out with the conclusion that it's obviously better and anyone rejecting it is just "stubborn".

Nobody started out with that conclusions. The tests and experiments and workshops and consulting with select employees you propose, were all done 2-3 years ago. Results are in, but a chunk of population decides to ignore them and obstinately continue believing and claiming that it doesn't work, which is a conclusion they started out with.

Results like these?

https://www.faros.ai/blog/ai-acceleration-whiplash-takeaways

Or these?

https://www.forbes.com/councils/forbestechcouncil/2026/03/16...

Or these?

https://poll.qu.edu/poll-release?releaseid=3955

Yes, "results are in". They're all over the map, about productivity, about stress and churn, about trust, about public sentiment, etc.

But sure, if you want to tell people their productivity will be measured by token usage, they will certainly respond to that incentive by setting your checkbook on fire while they work on a job search.

If a company wants to provide AI accounts for people, along with guidance for usage and non-usage, that might well make sense for some jobs. It certainly makes sense for some uses. If they start measuring token usage, that's even worse than when companies tried to measure lines of code written.

Re: Tokenmaxxing is dead, long live tokenmaxxing

#200

Post by IanCal »

ryandrake wrote: post #97
aurareturn wrote: post #2

Tokenmaxxing was just a way to force employees to start leveraging AI in a meaningful way. For companies that have measured performance based on token spend, they can now dial it back. Employees have learned to leverage AI for things they wouldn’t have prior. Now they know what’s possible and what’s not. No one is stupid enough to always measure performance based on token spend and have unlimited budget. It was alway…

> Management felt like employees weren't leveraging AI fast enough. If my productivity is in line with their expectations, I don’t understand why management cares what tools I’m using to do it. No employer ever told me to use emacs instead of vi, even though I’m 10x more productive in one vs the other. So why all of a sudden does management need to micromanage my tools?

Expectations shift, and tools do matter.

Imagine you had a direct report. They were doing just fine, slightly better than a typical report. Then you found out they were writing all their code in notepad - no linting, no automated tests or live updates, no refactoring tools, no highlighting or any code search. They didn’t have any cross code searches and didn’t have any documentation. When they hit a problem, they’d churn away at it and never reach for docs, google or so.

Still, their performance is in line with what you’d expect from someone in their position.

Would getting them to try emacs, vi, linters, etc be micromanaging them? Do you think they’d perform better with them? They are performing in line with expectations for the role, so why bother with something you think would make them more efficient?

I’ve made this obviously over the top, and can hear already replies from other bemoaning my comparison while missing the point — tools do matter and if you genuinely believe that a developer could be more efficient working in a different way it makes sense to not only want them to try it but to actively fund that change. Hell, this is literally what we argue for in training! Spend money to make someone better at their job!

If you think AI tools make you worse or don’t and can’t help, then that’s one thing. But it makes sense for management if they think it might to spend money on it and to get you to try.

Not only this, but wasn’t everyone here shouting about how tokens were subsidised and it couldn’t last? If so, wasn’t the first half of this year a really excellent cheap time to do the maxxing?

Post reply on HN