Local inference is definitely going to make more and more sense. Modern CPUs have all this amazing hardware well-optimized for inference purposes. I use a lot of web tools and see AI baked in and it feels weird. I want the smartness localized for speed and data security. I think and hope the industry points towards smart ai agents operating as locally as possible.
After the Vercel hosting compromise and having to rotate a ton of keys recently, we are definitely implementing automated rotation of short lived keys. That was super painful.
You have to install their MDM device management software on your computer. Basically that computer is theirs now. So don't plan on just handing over your laptop temporarily unless you don't mind some company completely owning your box. Still might be a validate use for people with slightly old laptops lying around, but beware trying to share this computer with your daily activities if you e.g. use a bank on a browser on this computer regularly. MDM means they can swap out your SSL certs level of computer access, please correct me if I'm wrong.
Those tools aren't floating in the ether: someone has to go download it and run it in some way, automated or otherwise. I think the suggestion is to make that a step before publication as the post suggests.
IMHO, this is one of the coolest aspects of Clojure: the data structures were designed to be immutable as efficiently as possible. This allows for some of the most critical aspects of the language to function. If you're whole program passes around values that are immutable, you don't ever have to worry about which part of the code owns what. Rust is such a different language but the borrow-checker is solving a nearly identical problem: who owns this memory and who is allowed to change it? In Rust, one owner ever gets to write to a piece of memory and you can move that around. It catches lots of bugs and is extremely efficient, but it is definitely a difficult aspect of the language to master. Clojure just says everyone gets to read and share everything, because every mutation is a copy. One piece of your codebase can never stomp on or free some memory that another part is using. Big tradeoffs with garbage collection but geeze does it make reasoning about Clojure programs SO much easier. And the cornerstone of that idea is fast, immutable data structures by default. Maps are implemented similarly in that they are fast to make mutated copies with structural sharing to save memory.
Rich Hickey, Alex Miller, Stu Holloway, Nathan Marz, and the other people in this video are all very impressive, intelligent people. But never undercount how much if their success is due to consistent effort. They worked hard at something they cared about with depth, and also longevity. Anyone can learn Clojure with a bit of skill and tenacity. Anyone can contribute to the community in a meaningful way. It takes effort though. Please join us and contribute back to the community in ways that help you scratch an itch, give it some polish, and then share it. Or join the forums on Clojurians on Zulip. Come to Clojure/Conj, it was fantastic last year.
The only reason the deep state or anyone has any power is because most people don't care. If people cared, we could change. Modern politics is all about distracting everyone with some crazy as often as possible to keep shifting attention and basically disabling any progress.
I think the effect is actually backwards: there may only be 2 windows instead of 4 but the total amount of time window per year should theoretically go up significantly. The 2 removed reports should make both of those quarters less subject to insider trading and therefore more tradeable.
I feel like it's not higher than most consumers, if I have a problem that is serious enough then that's the benefit and direct trade-off of renting - it shouldn't be my problem and my landlord should take it seriously. If everyone self-selects out we are just making the rental market even more hostile.
I feel like you are being a bit contradictory: the suggestion is to dissuade AI content - isn't that "design[ing] a site to deter group(s) of people that they don't want"? I personally don't want to vibe check every HN comment if I can avoid it, I don't even think you can quantify that in any meaningful way. We can engender a site like that at least in spirit. It may be equally as difficult but it's still worth fighting for.
I think we can be a little more nuanced than calling this sentiment outright stupid. A top HN article is about Scientific publications being overwhelmed with LLM trash. LLMs do pose a very real challenge to modern discourse. 10 years ago we could know that if we read something that sounded intelligible that at least some minimum effort had been put forth by a huma to be coherent. That bar is now completely gone. Now all internet users have to become adept AI-sniffers to know if some random bot isn't wandering themnoff a mental cliff with perfect formatting and eloquent prose. Having visceral reactions to that aren't unfounded in my opinion. We've lost real signal and having a forum like this be polluted will be a big casualty if we aren't careful and deliberate about our reaction to AI.
It matters to me because I'm reading it now and feel more informed about this problem. Throwing the towel in and saying it's all pointless isn't helpful.
In theory, it's a legal requirement based on GDPR and CCPA as well as many other new digital rights laws across Europe and many states in the USA. SoundCloud is probably big enough to do that correctly otherwise e.g. the GDPR penalty is a highish percentage of the company's total revenue which gives the laws a good amount of "teeth".
And what special sauce does the web preview use? At some point, someone has to actually parse and process the data. I feel like on a tech site like Hacker News, speculating that Google has somehow done a perfect job of preventing malicious PDFs beckons the question: how do you actually do that and prove that it's safe? And is that even possible in perpetuity?
If you're implementing immutable DB semantics maybe you should consider Datomic or alternatives because then you get that for free, for everything, and you also get time travel which is an amazing feature on top. It lets you be able to see the full, coherent state of the DB at any moment!
It's definitely an interesting approach but the problem is now you have to change all your queries and undeleting get more complicated. There are strong trade-offs with almost all the approaches I've heard of.
Long term you couldn't kill more than existed, asymptotically the maximum number of ODs per unit time would be exactly equal to the number created, impossible to be more.
Both can be true. I have specifically heard that shoe glue needs to be repeatedly compressed to be maintained or it will disintegrate. Anecdotally as well, I had a pair of quality, lightly used, but old, hiking boots that had the sole completely separate after a day of heavy use. The runber sole completely detached from the leather shell inner shoe. The crazy thing was that BOTH boots failed within 20 minutes of each other.