apantel

16 ms·

151.
by apantel 2y ago
> China’s 10th Five-Year Plan Central planning doesn’t work.
152.
by apantel 2y ago
But all my stuff is here.
153.
by apantel 2y ago
R S T L N E
154.
by apantel 2y ago
There’s another video on his channel where he tries to answer why he built it: https://youtu.be/OSGLrzSuCtM It sounds like a bunch of little pet projects that could be run on a very small cluster. I think he just has a lot
155.
by apantel 2y ago
Give it time. The technology is a year old.
156.
by apantel 2y ago
The ads are the real content from Medium’s perspective. The article is actually the medium by which the real content is delivered, like a train carrying dark passengers. The article is not what Medium cares about delivering to your browse
157.
by apantel 2y ago
So many ads… Edit: but it’s so apt. What that valuation is actually showing you is the value of sitting in between people looking for things and people looking to sell things to people looking for things, and turning that position into an
158.
by apantel 2y ago
You’re holding it wrong! You need to hold it so the sun is behind it and can shine through the screen.
159.
by apantel 2y ago
Clickbait. No signs of spiders, no signs of life.
160.
by apantel 2y ago
This is going to be an absolutely enormous market, and it would be really good if an open source option wins it, instead of another big tech walled garden. The good news is that it’s a software solution, not hardware. Open source can only
161.
by apantel 2y ago
They’re expensive to train, not to run. A technical person could produce and run the kind of agent I’m talking about today on pro-sumer hardware. It would take a lot more effort to make something that non-technical people can use.
162.
by apantel 2y ago
I think the advertising model is bound to collapse once GPT agents / assistants progress to the point where anyone can set one up to do a lot of their internet searching for them. The GPT will take your query and provide you an answer
163.
by apantel 2y ago
I think this is missing the point and not seeing where the technology is going. For generative art, right now it's basically prompt in -> image out, with no iterative process. But I can tell you that in coding, which is a creative
164.
by apantel 2y ago
It should be possible to come up with generalizable algorithms to determine a confidence score for any output, something akin to strong or weak connections in the brain. Is the response to the prompt supported by robust connections and str
165.
by apantel 2y ago
Sam didn’t get the black leather jacket memo…
166.
by apantel 2y ago
Caches, precomputed values, lookup tables — it’s all good as long as it’s well-organized and maintainable.
167.
by apantel 2y ago
I absolutely agree that latency is the real thing to optimize for. In my case, I only leave the application to access the db, and my applications tend not to be write-heavy. So in my case latency-per-request == how much work the computer
168.
by apantel 2y ago
I think it depends on what you’re building and who’s building it. We’re all benefitting from the fact that the designers of NGINX made performance a priority. We like using things that were designed to be performant. We like high-FPS gam
169.
by apantel 2y ago
The counter-argument to this is if you are building something that is in the critical path of an application (for example, parsing HTTP in a web server), you need to be performance-minded from the beginning because design decisions lead to
170.
by apantel 2y ago
Profit is revenue - expenses. The margin on one product is not the profit made on that product. That margin is joining all of the other margins, which have to conver the whole cost of running the business. Then once all those expenses ar
171.
by apantel 2y ago
This is the obvious complaint, but ASCII is the only common subset of the various encoding schemes. For some things like programming languages and protocols I think it make sense to have an ASCII constraint. You can express pretty much an
172.
by apantel 2y ago
This is the way to do it. First you want as complete coverage as possible with a unit test showing that the implementation is sound. Then you can try things and refactor for performance, checking as you go that you didn’t break anything.
173.
by apantel 2y ago
Yeah you can’t update these phones. Things just degrade if you do. Happily running iOS 15 on my iPhone 12.
174.
by apantel 2y ago
I agree. The first time you try to make a new type of thing, it takes a lot of trial and error to understand the domain and get things working. But once you’ve done that and understood the domain, you can usually do much much better with
175.
by apantel 2y ago
The issue is LLM’s also generating the web results. Garbage in, garbage out. The real problem is maintaining the ability to find the ground of real knowledge.
176.
by apantel 2y ago
Is this a joke? Of course they’re conscious. Jeez.
177.
by apantel 3y ago
It’s a quality of artisans to make the things they make polished and complete. There’s an inherent satisfaction in doing that if it’s your thing. In other words, applying that polish is an end, not a means.
178.
by apantel 3y ago
Democracy is already obsolete as a form of government. For democracy to work, you need an educated public, i.e. electorate, that is reasonably cohesively informed by a legitimate press. The press has been destroyed by the collapse of the
179.
by apantel 3y ago
The author deals too much in absolutes. Generative AI is a miracle that we’re already desensitized to. Use it for what it’s good for. There are lots of things. That said, the whole enshittifiation paradigm is 100% the Zeitgeist right no
180.
by apantel 3y ago
Base 10 is dominant because we have ten fingers. The fingers are a register that can count up to 10. Once you fill the register you increment a ‘tens’ counter (in your mind) and clear the register. Base 10 is just an elaboration of that.

More ›