OK.... hacking away at the problem and i think i'm on to something (or they're secretly serving better models / done secretly training and throttling us). Been going hard on 3 full-time tasks in codex since 7PM last night. 14 hours in and i'm at 66% usage left...WAY better...🧵
1
0
1
55
i think, like IRL, what gets measured gets done. it appears to be MUCH more focused.
2) i tell it NO UNIT TESTS or any CI runs until AFTER it finishes the feature and ensures that it works. The unit test thrashing is gone (yeesh) - and i'm noticing a huge diff in speed...
2) i tell it NO UNIT TESTS or any CI runs until AFTER it finishes the feature and ensures that it works. The unit test thrashing is gone (yeesh) - and i'm noticing a huge diff in speed...
1
0
0
11
OpenAI's plan....
1. Turn 20x into 100x to get devs
2. Drop 20x per week for 5 weeks
3. Announce Sol 6 as "more efficient" and give back like 5x or so (to the 20x-ers)
4. IPO with "we're profitable!"
Unforeseen...
5. Everyone feels bait and switched and goes open source
6. Fail
1. Turn 20x into 100x to get devs
2. Drop 20x per week for 5 weeks
3. Announce Sol 6 as "more efficient" and give back like 5x or so (to the 20x-ers)
4. IPO with "we're profitable!"
Unforeseen...
5. Everyone feels bait and switched and goes open source
6. Fail
1
0
1
34
I have 3 subscriptions to chat gpt and i'm sitting here with DAYS until it resets. This is wild. I can definitively say i got significantly less done this week than prior weeks. Something isn't very wrong @thsottiaux @ChatGPT ... i miss the old days when you could use codex
0
0
1
75
