conputer dipshit
@davidcrespo.bsky.social
web dev + hot dad. enjoy charts, unions, conputer games, philosophy. chicago crespo.business
in other words, you could make the exact same argument about pre-LLM learning: surely no one will be allowed to take the time to learn. clearly someone was!
November 26, 2025 at 7:32 AM
in other words, you could make the exact same argument about pre-LLM learning: surely no one will be allowed to take the time to learn. clearly someone was!
I don’t really buy the premise that the reward obviously goes to whoever does the work fastest. that is not how it has ever worked in the industry, at least not everywhere. learning while you work makes your work better
November 26, 2025 at 7:30 AM
I don’t really buy the premise that the reward obviously goes to whoever does the work fastest. that is not how it has ever worked in the industry, at least not everywhere. learning while you work makes your work better
so in the case of the social apps, the AI slop content was generated through some provider and the problem is that the feeds are boosting it. the boosting is very distinct from the generation
google is different because it sounds like it’s their own AI summaries that are the main culprit
google is different because it sounds like it’s their own AI summaries that are the main culprit
November 26, 2025 at 7:15 AM
so in the case of the social apps, the AI slop content was generated through some provider and the problem is that the feeds are boosting it. the boosting is very distinct from the generation
google is different because it sounds like it’s their own AI summaries that are the main culprit
google is different because it sounds like it’s their own AI summaries that are the main culprit
the only real overlap there is Google. Meta makes some LLMs but their choices about what to boost through their algo feeds have nothing to do with that. Same is true of Twitter. Pinterest doesn’t make LLMs. OpenAI and Anthropic make LLMs but don’t have algo feeds
November 26, 2025 at 7:11 AM
the only real overlap there is Google. Meta makes some LLMs but their choices about what to boost through their algo feeds have nothing to do with that. Same is true of Twitter. Pinterest doesn’t make LLMs. OpenAI and Anthropic make LLMs but don’t have algo feeds
the paper doesn't show that and the authors said so!
they gave the caveats but here's my version:
- experts on their own open source proj are not representative of most software dev
- measuring time undervalues trading time for effort
- tools are noticeably better than they were 6 months ago
- it really does take months of use to get the hang of it
- experts on their own open source proj are not representative of most software dev
- measuring time undervalues trading time for effort
- tools are noticeably better than they were 6 months ago
- it really does take months of use to get the hang of it
November 26, 2025 at 6:43 AM
the paper doesn't show that and the authors said so!
I had skimmed it but I just read through it now in full. like I said, I think "flooding the internet" is the wrong emphasis — the article is almost entirely about the bad things Google, Meta, and Pinterest are doing to actively boost slop
November 26, 2025 at 6:01 AM
I had skimmed it but I just read through it now in full. like I said, I think "flooding the internet" is the wrong emphasis — the article is almost entirely about the bad things Google, Meta, and Pinterest are doing to actively boost slop
I’m not talking about high quality content. I’m talking about not garbage
November 26, 2025 at 4:30 AM
I’m not talking about high quality content. I’m talking about not garbage
my most heterodox socialist opinion is that everyone should listen to a few episodes of Acquired. I especially liked the Nintendo and IKEA ones
www.acquired.fm/episodes
www.acquired.fm/episodes
Acquired | Episodes
Every company has a story. Acquired goes behind the scenes of the biggest tech IPOs and acquisitions of all time. Hosted by Ben Gilbert and David Rosenthal.
www.acquired.fm
November 26, 2025 at 3:09 AM
my most heterodox socialist opinion is that everyone should listen to a few episodes of Acquired. I especially liked the Nintendo and IKEA ones
www.acquired.fm/episodes
www.acquired.fm/episodes
to elaborate: the number of times I have heard executing well discussed in real life Business contexts as a contingent thing you could fuck up. easily in the hundreds
November 26, 2025 at 3:07 AM
to elaborate: the number of times I have heard executing well discussed in real life Business contexts as a contingent thing you could fuck up. easily in the hundreds
can you build a sustainable trillion dollar revenue industry around it? truly who can say. I certainly don’t feel confident saying you can or can’t
November 26, 2025 at 2:51 AM
can you build a sustainable trillion dollar revenue industry around it? truly who can say. I certainly don’t feel confident saying you can or can’t
my hottest take is that for all the handwringing about hype, LLMs are actually underhyped in the sense that they are significantly more capable *today* than 99% of people think
November 26, 2025 at 2:50 AM
my hottest take is that for all the handwringing about hype, LLMs are actually underhyped in the sense that they are significantly more capable *today* than 99% of people think
totally agree, predictions beyond a few months are very tough
at this point I don't know WHAT is going to happen. a bit of a wild mental place to be in. I feel like any predictions made in the current environment about the upcoming few years are just worthless right now.
November 26, 2025 at 2:47 AM
totally agree, predictions beyond a few months are very tough
I’m honestly not sure whether I never see really bad results, or I do occasionally see them and find them so easy to detect that I just tune them out. I suspect people remember the times when google was good with rose-colored glasses, like we didn’t have to search 10 times
November 26, 2025 at 2:44 AM
I’m honestly not sure whether I never see really bad results, or I do occasionally see them and find them so easy to detect that I just tune them out. I suspect people remember the times when google was good with rose-colored glasses, like we didn’t have to search 10 times
I agree: it is so bad, I do not ever use it. when I search, I use Kagi or an LLM with web search built in. I get good results. I suppose this is why I think the problem is solvable in principle
November 26, 2025 at 2:40 AM
I agree: it is so bad, I do not ever use it. when I search, I use Kagi or an LLM with web search built in. I get good results. I suppose this is why I think the problem is solvable in principle
I see this as proving my point. I think it’s a search problem and you’re complaining about bad search results. I’m not denying it’s bad — I haven’t used Google in a year, I use Kagi — I just think it will get better
November 26, 2025 at 2:34 AM
I see this as proving my point. I think it’s a search problem and you’re complaining about bad search results. I’m not denying it’s bad — I haven’t used Google in a year, I use Kagi — I just think it will get better
in response to this. a perfectly reasonable thought, I am just not persuaded
One of the big tidal forces of the coming years is that the baseline reliability of things you see on a screen is going to decline.
I think this has not been adequately metabolized.
This is one version, but there are a bunch of them.
I think this has not been adequately metabolized.
This is one version, but there are a bunch of them.
NEW: AI “recipe slop” is overrunning search and social. Food creators say Google’s AI Overviews and glossy fake food pics are drowning out real, tested recipes — collapsing traffic and setting home cooks up for disaster, especially this Thanksgiving.
Gift link: www.bloomberg.com/news/article...
Gift link: www.bloomberg.com/news/article...
November 26, 2025 at 2:31 AM
in response to this. a perfectly reasonable thought, I am just not persuaded
I certainly see the problem, I guess I just think of it as a search and curation problem, not fundamentally different from other search and curation problems. it's certainly possible it's especially tough, I just don't see any reason yet to think it is
November 26, 2025 at 2:27 AM
I certainly see the problem, I guess I just think of it as a search and curation problem, not fundamentally different from other search and curation problems. it's certainly possible it's especially tough, I just don't see any reason yet to think it is
which spaces do you mean? not bluesky. twitter is destroying itself by actively promoting slop. same deal with facebook and instagram
November 26, 2025 at 2:16 AM
which spaces do you mean? not bluesky. twitter is destroying itself by actively promoting slop. same deal with facebook and instagram
I am only responding to this claim that we will see baseline reliability decline over the coming years. I don't think we will
One of the big tidal forces of the coming years is that the baseline reliability of things you see on a screen is going to decline.
I think this has not been adequately metabolized.
This is one version, but there are a bunch of them.
I think this has not been adequately metabolized.
This is one version, but there are a bunch of them.
NEW: AI “recipe slop” is overrunning search and social. Food creators say Google’s AI Overviews and glossy fake food pics are drowning out real, tested recipes — collapsing traffic and setting home cooks up for disaster, especially this Thanksgiving.
Gift link: www.bloomberg.com/news/article...
Gift link: www.bloomberg.com/news/article...
November 26, 2025 at 2:04 AM
I am only responding to this claim that we will see baseline reliability decline over the coming years. I don't think we will
I'm not saying it's good or helpful. all I'm saying is I don't think we will see the baseline reliability "decline" "over the coming years", in response to this claim otherwise
bsky.app/profile/alex...
bsky.app/profile/alex...
One of the big tidal forces of the coming years is that the baseline reliability of things you see on a screen is going to decline.
I think this has not been adequately metabolized.
This is one version, but there are a bunch of them.
I think this has not been adequately metabolized.
This is one version, but there are a bunch of them.
NEW: AI “recipe slop” is overrunning search and social. Food creators say Google’s AI Overviews and glossy fake food pics are drowning out real, tested recipes — collapsing traffic and setting home cooks up for disaster, especially this Thanksgiving.
Gift link: www.bloomberg.com/news/article...
Gift link: www.bloomberg.com/news/article...
November 26, 2025 at 2:03 AM
I'm not saying it's good or helpful. all I'm saying is I don't think we will see the baseline reliability "decline" "over the coming years", in response to this claim otherwise
bsky.app/profile/alex...
bsky.app/profile/alex...