David Cooley

126 posts

David Cooley banner
David Cooley

David Cooley

@_david_cooley

Katılım Ocak 2019
407 Takip Edilen140 Takipçiler
Jonathan Chang
Jonathan Chang@ChangJonathanC·
claude code is asking permission to edit file when bypass permissions is on @bcherny
English
5
1
126
26.6K
David Cooley
David Cooley@_david_cooley·
@SenSanders Ok so the latest argument is that AI is unacceptable because all of that stuff is still happening and AI didn’t change the fabric of our reality to undo that very well established reality?
English
0
0
0
193
Sen. Bernie Sanders
Sen. Bernie Sanders@SenSanders·
I spoke to Anthropic’s AI agent Claude about AI collecting massive amounts of personal data and how that information is being used to violate our privacy rights. What an AI agent says about the dangers of AI is shocking and should wake us up.
English
1.5K
3.7K
23.6K
6M
David Cooley
David Cooley@_david_cooley·
This seems to be the conventional wisdom but I will tell you a secret: It’s actually very possible to get good results all the way to the auto compact limit of 1M token window. My personal process is HIGHLY plan driven. Basically plan mode but on steroids. Has been for a long time. Sophisticated and slow but very powerful churning machine who never loses its place vibes. I really promise I understand what you and others are saying, I’ve wrangled with the same problem since I started doing agentic coding back when it sucked bad. I promise you this is not actually a limitation. Try to figure out how to keep your agents more sligned, more required to use a variety of sub agents that help it stay aligned, etc. Disbelieve me if you want, but 1M context window has made a bug change for me. The value is there and reapable
English
0
0
2
90
sankalp
sankalp@dejavucoder·
1M opus is great in the sense you dont have to face the clueless compacted claude once you reach close to their auto compact limit. can just keep iterating till 150k and comtinue asking stuff/small changes post 200k. though i try to clear post 200k for remotely complex stuff.
English
5
0
55
2.8K
roon
roon@tszzl·
@rhydhimma they are far more significant than alphafold and it’s not close imo
English
21
9
536
24.7K
Rhydhimma (sci/acc)
Rhydhimma (sci/acc)@rhydhimma·
Codex and Claude Code are probably the most revolutionary products of this century. For now. Maybe not as significant as Alphafold, and all the PhD who slogged to get protein structure data. Data is the keyword.
English
6
2
167
23.3K
David Cooley
David Cooley@_david_cooley·
@mattpocockuk @JayD0ubleu You may have had similar results running my “/kill-assumptions”. Pretty much works 75% of the time on the first time. These things work!
English
0
0
0
38
Matt Pocock
Matt Pocock@mattpocockuk·
@JayD0ubleu Maybe - and this is always a possibility - but I do doubt it on this occasion. Especially since I cleared the context and it solved the bug in 80K tokens.
English
1
0
4
1.9K
Matt Pocock
Matt Pocock@mattpocockuk·
Doing some experiments today with Opus 4.6's 1M context window. Trying to push coding sessions deep into what I would consider the 'dumb zone' of SOTA models: >100K tokens. The drop-off in quality is really noticeable. Dumber decisions, worse code, worse instruction-following. Don't treat 1M context window any differently. It's still 100K of smart, and 900K of dumb.
English
144
55
1.1K
143.4K
David Cooley
David Cooley@_david_cooley·
Youre not wrong but in my experience if you’ve built a really tight planning/execution loop already, then when 4.6 came out it was apparent that agents could stick to the plan better almost all the way through the context window. (200k) I even started experimenting with allowing auto compact from time to time. Now with 1M context I am (sorry I wish I could explain better) doing the exact same… is <100k better? Always. But if you’ve already got a workflow that keeps agents aligned tightly in theory, I absolutely noticed a difference in agent adherence capability. At 1M tokens I feel like quality stays as I’m used to throughly the window. Worth noting I’ve poured tons and tons of time into my custom planning/execution flow and my harness in general.
English
0
0
0
198
PancakesDivorcePancakes
PancakesDivorcePancakes@SultanateOfFun·
@_david_cooley @ja3k_ This and also the most politically vocal/lowest information voters tend to have hobbies that involve operating diesel powered machines at various levels of inebriation.
English
1
0
1
48
ja3k
ja3k@ja3k_·
Idk why gas prices are so culturally salient in america. You could drive an hour a day and it probably comes to less than $3k/year. Is it because they put the price on billboards along the road?
English
653
47
5.6K
1.7M
David Cooley
David Cooley@_david_cooley·
@atelicinvest I’ve noticed confidence in LLMs is highly correlated to how long they one has spent trying hard to figure out what works
English
0
0
0
4
David Cooley
David Cooley@_david_cooley·
I think those who have had enough years in the trenches to at least “get it” are all in the sweet spot right now. If you try very hard you can produce great results. Better than you could do yourself. Before we know it though, you won’t have to try as hard or have as tight a workflow and auditing approach, etc. just a bit. Eventually it will become much more likely you cause bloat in the codebase than the LLM Just one man’s (who started “agentic engineering” 2 years ago when it sucked really bad and 50% of people believed it would never escape THAT velocity over and over) opinion though.
English
1
0
1
143
Unemployed Capital Allocator
I think one sneaky aspect of LLM coding that is under discussed is just how bad the code has to be before appearing as broken to the casual observer.
English
26
21
713
25.8K
David Cooley
David Cooley@_david_cooley·
Well this must make sense from your POV but everyone I know who is using it in a professional software capacity was building sufteare professionally for many years and now taking strides they would not have managed before in actual use cases, yes - while continuing to do the rest of the job well (and certainly babysitting a lot etc). We don’t all agree but the consensus is much more radical change is coming and we’re already past the inflection point in terms of cost/value. If you know nobody shipping better with AI IRL find somebody to talk to. Internet may not be the best place
English
0
0
1
134
David Cramer
David Cramer@zeeg·
Why is it everyone with an absurdly futuristic AI take is someone who - as best I can tell - doesn’t work on (and often never has) real software that has real users and real requirements? More so, why do you trust them?
English
120
45
1K
49.8K
Shivers
Shivers@thinkingshivers·
Not enough people talk about how unpleasant vibecoding is. The best analogy I can think of is driving. It's cool that we can just hop in a car and drive to the store. It's a lot faster than walking. And yet, it's so stressful and infuriating, we had to invent a new word just to describe its effect on people: "road rage." AI-assisted coding is the same. It's so much faster--there's no going back to coding everything by hand, the equivalent of walking everywhere. And yet it's incredibly annoying and stressful. It's characterized by annoying delays between requests, time-wasting misunderstandings, blatant lying, and absurd overconfidence. Hopefully this gets better as models improve.
Shivers tweet media
English
75
27
676
36.8K
David Cooley
David Cooley@_david_cooley·
@ThePrimeagen Yeah and what’s with these people calling agentic tools “harnesses”? Anybody see any leather? I don’t. What?!
English
1
0
1
1.1K
ThePrimeagen
ThePrimeagen@ThePrimeagen·
the amount of times i have seen "operating system" describing tools that write code or use a texting app and a cron job is terrifying is this what Andreesen meant by retardmaxxing?
English
45
36
1.2K
42.8K
Justin Ryan ᯅ
Justin Ryan ᯅ@justinryanio·
Reporter: Some are concerned NVIDIA’s DLSS 5 will make games look worse or homogeneous. In our Q&A today, Jensen Huang responded: “They’re completely wrong… you can fine tune the generative AI to your artistic style… if you want cartoon, toon shader, made of glass… it’s up to you. It’s not post processing at the frame level. It’s generative control at the geometry level. This is very different than generative AI. It’s content control generative AI.”
English
122
98
1.2K
187.6K
Ro Khanna
Ro Khanna@RoKhanna·
@AbstractHermit @patrickgaspard @ChrisVanHollen Facts don't matter to Greenblatt. He is a Trump apologist who attacked Obama's nuclear deal, defends Elon, and is basically a shill for the Trump Administration and Netanyahu. Sad to see. He has zero respect among any House Democrats anymore.
English
91
416
4.4K
232.3K
Patrick Gaspard
Patrick Gaspard@patrickgaspard·
This nonsense is no longer going to fly. Every Democrat should denounce this and defend ⁦@ChrisVanHollen⁩ and ⁦@RoKhanna⁩ - two public servants with integrity who have always repudiated hate while defending the rights of all.
Patrick Gaspard tweet media
English
120
476
2.3K
123.2K
David Cooley
David Cooley@_david_cooley·
@allgarbled looks good though imo! if its useful and good, who cares? (of course that's not how I feel with end-user UI)
English
0
0
0
523
David Cooley
David Cooley@_david_cooley·
@sporadica @0xkrma He really is one of those guys you want to be smart and insightful and good but you keep seeing evidence to the contrary. Whatever though, I’m judging him against great minds. He’s probably smarter than me I just get to critique his blind spots from my hovel lol
English
0
0
0
21
spor
spor@sporadica·
@0xkrma truly the dumbest shit i ever heard, not surprised it’s from him
English
4
0
38
938
Karma
Karma@0xkrma·
This is verifiably false. More like a case of confirmation bias, reinforced by ChatGPT psychosis.
David Senra@davidsenra

Great men of history had little to no introspection. The personality that builds empires is not the same personality that sits around quietly questioning itself. @pmarca and I discuss what we both noticed but no one talks about: David: You don't have any levels of introspection? Marc: Yes, zero. As little as possible. David: Why? Marc: Move forward. Go! I found people who dwell in the past get stuck in the past. It's a real problem and it's a problem at work and it's a problem at home. David: So I've read 400 biographies of history’s greatest entrepreneurs and someone asked me what the most surprising thing I’ve learned from this was [and I answered] they have little or zero introspection. Sam Walton didn't wake up thinking about his internal self. He just woke up and was like: I like building Walmart. I'm going to keep building Walmart. I'm going to make more Walmarts. And he just kept doing it over and over again. Marc: If you go back 400 years ago it never would've occurred to anybody to be introspective. All of the modern conceptions around introspection and therapy, and all the things that kind of result from that are, a kind of a manufacture of the 1910s, 1920s. Great men of history didn't sit around doing this stuff. The individual runs and does all these things and builds things and builds empires and builds companies and builds technology. And then this kind of this kind of guilt based whammy kind of showed up from Europe. A lot of it from Vienna in 1910, 1920s, Freud and all that entire movement. And kind of turned all that inward and basically said, okay, now we need to basically second guess the individual. We need to criticize the individual. The individual needs to self criticize. The individual needs to feel guilt, needs to look backwards, needs to dwell in the past. It never resonated with me.

English
7
3
113
8.4K
David Cooley
David Cooley@_david_cooley·
@avidseries I imagine moisturizer or oil (a good facial moisturizer can put most anti aging strategies to shame, perhaps he’s taken it to the extreme)
English
0
0
1
723
Stop Autocrats 🇺🇸
Stop Autocrats 🇺🇸@justicenow_alan·
Unlike most, I actually read the article. Nowhere in this article does the story describe them as innocent protestors. It does recount the various claims made by each of the parties, almost verbatim. It also explains that some of the accusations made by prosecutors are not actually legal designations. Your take on this is entirely wrong.
English
13
0
8
905
Bonchie
Bonchie@bonchieredstate·
Nine Antifa terrorists were just convicted in Texas for their role in attacking an ICE facility and the attempted murder of officers. The Washington Post described them as innocent protesters being targeted because they were “Trump critics.” This is our national press.
Bonchie tweet media
English
113
965
3.7K
79.5K