Your session limit is still 5 hours. We've just changed the definition of what an hour is
https://x.com/trq212/status/2037254607001559305
(h/t @davidgerard)
Your session limit is still 5 hours. We've just changed the definition of what an hour is
https://x.com/trq212/status/2037254607001559305
(h/t @davidgerard)
@jalefkowit @davidgerard
à la vrai mode Rene, mon ami Claude tient à déclarer : « Ceci n'est pas une heure ! »
Brough to you by Lorentz of Arabia
I thought you were joking with your summary! That is wild
@jalefkowit @davidgerard @adrianh somehow the consensus I am hearing for Claude Code is simultaneously it is by far the best and most advanced agent for writing code ever and it’s a complete rip off that needs hundreds of dollars a month to approach adequate limits which are constantly being reduced arbitrarily
how is that a product
@leon @jalefkowit @davidgerard @adrianh
At its current price, it's one of the few "AI" products which actually earns its fees for increased productivity in terms of code production. With enough cajoling, weird instructions, limits, and adequate testing frameworks, I think it does, ultimately, have a quite defensibly positive impact on code production. As a strong skeptic, this is one application of LLMs that, again, at current prices, is an actual legitimate productivity booster.
It is of course being offered at what's currently a 90% VC subsidy, and at full price that net cost gain almost certainly disappears. There's a small chance they can actually engineer something that's almost as effective and less expensive to run. But I'm not holding my breath.
@MichaelTBacon @leon @jalefkowit @davidgerard @adrianh You're not a "strong skeptic" if you're burbling inanely about how the stochastic garbage generator is good at "code production"—i.e. it outputs trash faster than an unaided human being can, and that is therefore more "productive" by mere maximization of numbers.
If you'd like to be a REAL "strong skeptic", and not a performative fake one, you might start by asking yourself why you would ever trust a delicate task such as writing code (or any other difficult task) to an unintelligent device, designed to be unable to tell apart false information from truthful information (because that would slow the thing down), which is fraudulently being marketed as intelligent.
If you're such a skeptic why isn't your first concern with the fraud? But it's not, is it, Mikey? It's with mere numbers going upwards...i.e. "increased productivity", according to the braindead way in which tech professionals assess "productivity".
This has the same energy as a note on a menu I saw once saying "in order to keep our prices low and comply with the new minimum wage, we will be adding a 8% surcharge to your bill"