You can't assume more tokens == more value. Turns out bad ideas, amplified, just turn in to bad implementations.
Github says we've 14x the number of commits this year. There is no evidence of anything close to 14x economic value being created. Most of those tokens are purely waste.
This is what hype looks like. Managers were convinced they had secured a source of magic fairy dust that immediately finishes any job it lands on and has no downsides, and they saw their employees weren't using it for reasons they couldn't understand, and so they went about solving the problem in a way that would make sense for someone with that worldview.
Yet another example that stupid fads like "tokenmaxxing" were just designed to drain budgets (Uber [0], Amazon [1] and many others all falling for it) without understanding what exactly should be built any why.
The ones still actively encouraging it were creating the illusion of "productivity" gains with frivolous measurements such as "leaderboards" and dishing out timed discounts like a casino or a betting shop would.
Now do you really think "tokenmaxxing" with a fleet of multiple subagents with loops on auto mode (by default) with the highest model settings (xhigh) running for days 24 / 7 per employee is a good idea?
You can't assume more tokens == more value. Turns out bad ideas, amplified, just turn in to bad implementations.
Github says we've 14x the number of commits this year. There is no evidence of anything close to 14x economic value being created. Most of those tokens are purely waste.
I'm honestly mystified why anyone ever thought it would lead to some kind of positive outcome.
This is what hype looks like. Managers were convinced they had secured a source of magic fairy dust that immediately finishes any job it lands on and has no downsides, and they saw their employees weren't using it for reasons they couldn't understand, and so they went about solving the problem in a way that would make sense for someone with that worldview.
This was hardly a surprise [0].
Yet another example that stupid fads like "tokenmaxxing" were just designed to drain budgets (Uber [0], Amazon [1] and many others all falling for it) without understanding what exactly should be built any why.
The ones still actively encouraging it were creating the illusion of "productivity" gains with frivolous measurements such as "leaderboards" and dishing out timed discounts like a casino or a betting shop would.
Now do you really think "tokenmaxxing" with a fleet of multiple subagents with loops on auto mode (by default) with the highest model settings (xhigh) running for days 24 / 7 per employee is a good idea?
[0] https://news.ycombinator.com/item?id=48277922
[1] https://fortune.com/2026/08/07/uber-ai-spending-tokenmaxxing...
[2] https://www.businessinsider.com/amazon-ai-leaderboard-tokenm...