This makes sense to me. Consider these assumptions:

1) During peak times, they're already struggling to meet demand. They've figured out that 50% should be achievable without causing too much pain for existing users, and by setting a limit, more users will use it for architecture and then switch back to Opus 4.8 for implementation.

2) They need data on how people are using it at scale to figure out when it should fire up Opus or Haiku agents. They need to be careful not to have a model that kills your usage cap halfway through the week, and they need time to bake in balance.

I think people are reading too much into them not reiterating that they want to offer it in Max subscriptions in the future too. Of course they do, they want to compete!

I think they're trying to avoid setting up expectations right now because of all this backlash, so they didn't restate something that might make people mad. I don't even think they miscalculated - I think we're being hyper-nitpicky.

I would expect that by late July, they'll figure out the balance, and either offer a higher tier Max subscription (40x for $300, perhaps), or have Fable drop to Opus for implementation consistently, and then start offering Fable again in the Max subscriptions. They need to do something to make sure a Fable user doesn't just hit their head on the ceiling all the time.

I have been positively surprised by the current usage limits on the Max 5x plan.

I think it might even be suitable for full time work, if the $200 plan gives four times the usage.

Particularly if they also remove the halving in the future.

I am not holding my breath for them adding an even more subsidized subscription tier.