Rodney Howell
57 posts
Rodney Howell
@RodAHowell
I help adults escape the pain of unhealthy lifestyle patterns through habits that align with their natural drives.
Katılım Ekim 2024
102 Takip Edilen9 Takipçiler
@pmarca Or maybe civilization advances when systems evolve enough for more people to actually participate in understanding together.
The gap isn’t just intelligence. It’s translation, trust, and access.
#CivicBranchRise
English
Counterpoint: Pretty much the entire history of civilization is smart people trying to explain things to dumb people and failing.
Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭@elder_plinius
@pmarca for now… the supersuaders cometh
English
@MatthewBerman I downgraded to the $100 plan and may drop to the $20 plan depending on how GPT Prop goes (so far so good). 4.7 was a disaster for me.
English
@mattshumer_ This was true 4-6 weeks ago. Claude degraded, ChatGPT and Codex updated and got much better. People switched. He's a bit late.
English
@adonis_singh My experience is that GPT 5.5 Pro crushes both Opus 4.6 & 4.7 which I have been a heavy user of.
English
Been using Claude Max Plan (200) for a couple of months. I tried GTP 5.5 Pro (pro is key) and now I can change the speed/thinking level message by message and the extended pro level is slow but better than Claude (4.6 or 4.7). This feels like a major shift.
English
@MatthewBerman @sama @thsottiaux I downgraded my Claude 200Max to 100 and am now using GPT 5.5 as my primary. Opus 4.7 is the reason - it's unusable.
English
wow y'all love 5.5
we should think of something nice to do to celebrate!
English
I'm done using Opus 4.7 - too many mistakes. This is not a condemnation of 4.7 - it's great for some people. This is just a data point.
English
@MoeCanDoIt I had the same experience. I'm getting tired of people saying it's all in the prompt as if there is no real variance in his a model performs aside for your prompting.
English
@robinebers I’m glad you’re having good luck. It has done poorly here. Shortcuts galore.
English
you're sleeping on Opus 4.7 because everyone whines about it
meanwhile, here's what I did last night
→ ran a Cursor Claude task for hours
→ new feature built end to end
→ PR babysat into mergeable
→ GPT-5.4 High reviewed it this morning
perfect score, zero notes
first time I've ever seen a Claude do anything and come back clean after review
caveat: I run an automated process where BugBot + @cubic_dev_ run code reviews, so there's obviously GPT in the mix - but that's the point. this is more fun, cleaner code, and the outcome is perfect
Opus 4.6 would eventually give up or take shortcuts
I haven't seen 4.7 do this yet, so I'm actually loving it
English
My biggest struggle with AI is content and context management. Too many docs and doc interdependencies. And if you consolidate you take a context hit every time a model loads your doc. I'm trying @AiClairvoyance. We'll see how it goes.
English
I know Opus' quality varies widely from person to person. My experience with 4.7 has not been good. Many foolish mistakes. I am rolling back to 4.6. I'm not making any claims on model quality - only sharing my own experience.
English
@MatthewBerman @theo I love that you just bluntly ask the same questions I have!
English
If you want to test Opus 4.7 without the lobotomized system prompt, you can try it out in T3 Chat
English
English
ChatGPT has ~900M users
OpenAI new strategy:
>abandon side quests
>pivot to enterprise
>bet on Codex to beat Claude Code
is Sam Altman a visionary?
OAI Investor:
NIK@ns123abc
*🚨BREAKING: OpenAI’s own investors just called them out on record to the FT Early backer: >“It’s a deeply unfocused company” Another investor on TBPN acquisition: >“I don’t get it frankly. It’s a distraction and it irks me” Iconiq Capital: >“We picked. We invested a lot into Anthropic” Investor who backed both OAI and Anthropic: >“OpenAI risks being left in no man’s land” Sapphire Ventures president called it: >“The Netscape of AI” OpenAI investors say to justify $852B you’d need a $1.2T IPO — hard to justify when Anthropic is $380B and growing faster Secondary markets are now pricing Anthropic at a premium over OpenAI INVESTOR SENTIMENT FLIPPING
English
@draginol Whatever you did made it work in Chrome as well now. Nice!
English
@RodAHowell Ok I'm updating the server as we speak with a tweak that might help. Try again at say 7:42pm EST.
English
The biggest problem I see with AI posts is that so many of them are about "We built thing" without a use case.
The beauty of Clairvoyance is that we have already been using it in production for real work, on real products, for a long time.
Example: Art tools described by the team, assigned to AI staff to come up with ideas and shared.
English
@draginol I've tried chrome Version 146.0.7680.178 (Official Build) (64-bit)
And Edge Version 146.0.3856.109 (Official build) (64-bit)
English
@draginol The page loads, then when I click "Download for Windows" I get this error message:
This site can’t provide a secure connection
xgksnrhkevbnpggpbsil.supabase.co sent an invalid response.
Try running Windows Network Diagnostics.
ERR_SSL_PROTOCOL_ERROR
English