JamesSwift

5 ms·

31.
by JamesSwift 3mo ago
Yep, total agreement here. But they really nail the "it just works" of playback, so all things considered I stick with it.
32.
by JamesSwift 3mo ago
The way they do their list rendering is infuriating for large collections. Instead of showing titles + shimmer, its just big blobs of loading placeholders. So to find where you are in your list of movies you have to stop scrolling and let t
33.
by JamesSwift 3mo ago
I tried to get involved (in very minor ways) a while back but got basically no engagement from the project so I just let it be. I moved to infuse instead as my ios client and its been good overall. I really hate its list performance on larg
34.
by JamesSwift 3mo ago
The server is simple, theres plenty of docker examples online. Cross play between bedrock and java edition, and connecting to custom servers on eg switch is the harder part. And using linux as a client is more advanced.
35.
by JamesSwift 3mo ago
I agree. The issues arent super nuanced (and are pretty "blatant") But I do think the nuance of "who/what should we point our finger at" is important. Because like we see in this thread, the finger is being pointed
36.
by JamesSwift 3mo ago
Well, I didnt bring it up but the other element of qualified immunity is that its purely for civil suits. So it would only show up if an officer was basically sued personally. It doesnt apply to criminal prosecution. Thats another can of wo
37.
by JamesSwift 3mo ago
Qualified immunity is more nuanced. It allows the first offense to be absolved but it works like legal precedent where future offenses by _any law enforcement officer_ is not covered. Now there’s plenty of loopholes where you can craft “uni
38.
by JamesSwift 3mo ago
This got me thinking… why does AI use em-dashes nearly exclusively? I think “in the wild” examples overwhelmingly weight toward en-dashes instead.
39.
by JamesSwift 3mo ago
https://www.youtube.com/watch?v=Hx7Vv5pqpHg
40.
by JamesSwift 3mo ago
Self-directed memories have been awful. That said, I have had decent success in self-feedback-loop skills. I encapsulate a long process into a skill and the last step is to make changes as needed to the skill itself for things it ran into.
41.
by JamesSwift 3mo ago
Sure you can avoid it. Require unprintable tokens on messages, strip non-ascii from input. Structure your AI systems to clearly indicate what is user-generated content.
42.
by JamesSwift 3mo ago
No, AI dials up the OPs axiom to 11. Heavily leveraging AI for dev is going to accelerate you as a solo dev but you will hit serious friction trying to scale to more than the individual.
43.
by JamesSwift 3mo ago
Thats quantized
44.
by JamesSwift 4mo ago
Now Im even more confused : D That would also explain the issue I mention in my other comment. And would also reinforce how much output would degrade without this. Opus 4.5 was a step above previous models in my experience. At some point it
45.
by JamesSwift 4mo ago
Thats really surprising, I stand corrected. I have had a lot of issues with hallucinations I attributed to adaptive thinking, but I wonder if those were actually due to this behavior instead. I also wonder if they actually do a hybrid of &q
46.
by JamesSwift 4mo ago
> all major vendors throw out the reasoning tokens between turns That would be surprising to me. The reasoning _is_ the model intelligence in a lot of respects, and so dropping those from the context would affect its output pretty signif
47.
by JamesSwift 4mo ago
Another aspect Ive been talking about lately with the huge burst of AI productivity recently, is that its much simpler to point AI at the things that have been hanging out in the back of your mind. The things you started and got stuck. The
48.
by JamesSwift 4mo ago
151 commits, 1826 files changed and claude found 0 smoking guns? I call bullshit. It found 3 in one run-of-the-mill query the other day
49.
by JamesSwift 4mo ago
Ive said many times in my day-to-day convos around AI that Ive come to realize more and more that AI is a solo endeavor. Its very difficult to scale things to more people if you are trying to use AI for more than just speeding yourself up.
50.
by JamesSwift 4mo ago
How about Shellac Cache
51.
by JamesSwift 4mo ago
> And yes, before you ask, there is a 25gbase-t standard. Maximum distance: 30m (100ft). 100ft from panel to panel in a house? Oof Not to mention termination and interference considerations, etc. When I looked at it, I decided anything o
52.
by JamesSwift 4mo ago
Most nix users still use brew for casks on mac. Using brew bundle as your interface to it makes things almost seamless, so its not too icky overall. eg I manage my Brewfile declaratively with home-manager, and then run this on file change
53.
by JamesSwift 4mo ago
Well nix and devenv are also dropping intel mac support due to apple cutting off support : (
54.
by JamesSwift 4mo ago
Look... its clear you didnt read the article. The point of contention is that google _hallucinated_ facts. As in, website 1 said company A exists and website 2 said company B committed a crime and now googles AI summary says company A commi
55.
by JamesSwift 4mo ago
> was google search responsible for surfacing correct answers No, as the article lays out, google hallucinated facts that werent present in source material and so at that point the court found them liable for that. Previous protections e
56.
by JamesSwift 4mo ago
They are generally distributed in such a way that values created at nearly the same time dont cluster together which is less efficient for DBs
57.
by JamesSwift 4mo ago
UUIDs also have a nice benefit of it being impossible to query the wrong table with one if you mixup what an FK goes to
58.
by JamesSwift 4mo ago
OK well the initial wording seems like you are presenting this as an inherent limitation. "Has a personality that I dont agree with" is a different critique than "fundamentally does not understand" > If the AI had mor
59.
by JamesSwift 4mo ago
Hmm thats a pretty interesting variant. I dont think it would apply directly to our team, but I like the potential playground it gives for both coming up with solutions as well as being a kickstarter for convos on alternatives.
60.
by JamesSwift 4mo ago
An AI doing a bad job is not the same as it wasnt able to do a good job. I would bet if you asked it if its a good name it would figure it out, and give a logical argument on why to change it. Im not going to ascribe that to "intellige

More ›