Website: http://invece.org
Blog: http://antirez.com
My sci-fi book (English translation): https://www.ibs.it/wohpe-ebook-inglese-salvatore-sanfilippo/e/9791280845337
Twitter: @antirez.
- antirezYou can't read English like if it was a declarative logical language. It is obviously an hyperbole to say "everyone". It means "a lot of people". So why they didn't say "a lot of people"? Language uses hyperboles to make a point stronger.
- The 500 - 5000 figure, which is correct, is why most folks should instead self publish via KDP. 70% royalties mean you can get 10-50k easily with an average book. If the book is a success, you can switch career to a full time author if you wish. All this with full creative freedom. Many years ago, I canceled my Redis book for a large publisher for similar reasons to the OP: too many "do it this way" requests.
- The difference between Win32 and Linux is that the latter didn't realize an operating system is more than a kernel and a number of libraries and systems glued together, but is, indeed, a stable ABI (even for kernel modules -- so old drivers will be usable forever), a default, unique and stable API for user interface, audio, ..., and so forth. Linux failed completely not technologically, but to understand what an OS is from the POV of a product.
- LoRa diverges too strongly when applied in the real world. There are quirks in the chip (Semtech in the latest revision did a few strange things), and in the way chirp modulation works (the one used by LoRa), that makes it a lot better to test in the field.
- I wanted to fix this with my FreakWAN but I was never able to find a user base willing to validate the ideas of the routing I implemented. All the code is open source and easy to modify.
Write me if you are willing to experiment :)
- For a second I'll not consider the fact that I believe your argument is conceptually wrong. Let's focus only on the cultural part: "net positive by about 100% of people", this is deeply US-centric. For most other people, coming in the US is preparing to being exposed to an amount of AC that causes strong discomfort. Ask every European, for instance. Moreover having a more normal temperature would be something people adapt in just a few months: and this would save an enormous amount of energy. But no, let's blame AI, with our asses freezing at 17 degrees.
- No, because it's not a matter of who is correct or not, in the void of the space. It's a matter of facts, and it is correct who have a position that is grounded on facts (even if such position is different from a different grounded position). Modern AI is already an extremely powerful tool. Modern AI even provided some hints that we will be able to do super-human science in the future, with things like AlphaFolding already happening and a lot more to come potentially. Then we can be preoccupied about jobs (but if workers are replaced, it is just a political issue, things will be done and humanity is sustainable: it's just a matter of avoiding the turbo-capitalist trap; but then, why the US is not already adopting an universal healthcare? There are so many better battles that are not fight with the same energy).
Another sensible worry is to get extinct because AI potentially is very dangerous: this is what Hinton and other experts are also saying, for instance. But this thing about AI being an abuse to society, useless, without potential revolutionary fruits within it, is not supported by facts.
AI potentially may advance medicine so much that a lot of people may suffer less: to deny this path because of some ideological hate against a technology is so closed minded, isn't it? And what about all the persons in the earth that do terrible jobs? AI also has the potential to change this shitty economical system.
- You would expect that voices that have so much weight would be able to evaluate a new and clearly very promising technology with better balance. For instance, Linus Torvalds is positive about AI, while he recognizes that industrially there is too much inflation of companies and money: this is a balanced point of view. But to be so dismissive of modern AI, in the light of what it is capable of doing, and what it could do in the future, is something that frankly leaves me with the feeling that in certain circles (and especially in the US) something very odd is happening with AI: this extreme polarization that recently we see again and again on topics that can create social tension, but multiplied ten times. This is not what we need to understand and shape the future. We need to return to the Greek philosophers' ability to go deep on things that are unknown (AI is for the most part unknown, both in its working and in future developments). That kind of take is pretty brutal and not very sophisticated. We need better than this.
About energy: keep in mind that US air conditioners alone have at least 3x energy usage compared to all the data centers (for AI and for other uses: AI should be like 10% of the whole) in the world. Apparently nobody cares to set a reasonable temperature of 22 instead of 18 degrees, but clearly energy used by AI is different for many.
- C has no problems splitting programs in N files, to be honest.
The reason FB (and myself, for what it is worth) often write single file large programs (Redis was split after N years of being a single file) is because with enough programming experience you know one very simple thing: complexity is not about how many files you have, but about the internal structure and conceptually separated modules boundaries.
At some point you mainly split for compilation time and to better orient yourself into the file, instead of having to seek a very large mega-file. Pointing the finger to some program that is well written because it's a single file, strlongly correlates to being not a very expert programmer.
- If this had been available in 2010, Redis scripting would have been JavaScript and not Lua. Lua was chosen based on the implementation requirements, not on the language ones... (small, fast, ANSI-C). I appreciate certain ideas in Lua, and people love it, but I was never able to like Lua, because it departs from a more Algol-like syntax and semantics without good reasons, for my taste. This creates friction for newcomers. I love friction when it opens new useful ideas and abstractions that are worth it, if you learn SmallTalk or FORTH and for some time you are lost, it's part of how the languages are different. But I think for Lua this is not true enough: it feels like it departs from what people know without good reasons.
- > Text generated by an LM is not grounded in communicative intent
This means exactly that no representation should exist in the activation states about what the model wants to tell, and there must be only a single token probabilistic inference at play.
Also their model requires the contrary, too: that the model does not know, semantically, what the query really means.
Stochastic Parrot has a scientific meaning, and just only observing the function of the models, it is quite evident that they were very wrong, but now we have stong evidence (via probing) that also the sentence you quoted is not correct, since the model knows the idea to express also in general terms, and features about things it is going to say much later activates a lot of tokens earlier, including conceptual features that are relevant later in the sentence / concept expressed.
You are doing the big error that is common to do in this context of extending the stochastic parrot to a non scientifically isolated model that can be made large enough to accomodate any evidence arriving from new generations of models. The stochastic parrot does not understand the query nor is trying to reply to you in any way, it just exploits a probabilistic link among the context window and the next word. This link can be more complex than a Markov chain but must be of the same kind: lacking understanding whatsoever and communication intent (no representation of the concept / sentences that are required to reply correctly). How it is possible to believe in this, today? And, check yourself what the top AI scientists today believe about the correctness of the stochastic parrot hypothesis.
- What happened recently is that all the serious AI researches that were in the stochastic parrot side changed point of view but, incredibly, people without a deep understanding on such matters, previously exposed to such arguments, are lagging behind and still repeat arguments that the people who popularized them would not repeat again.
Today there is no top AI scientist that will tell you LLMs are just stochastic parrots.
- I'm not involved in business decisions and while I'm very AI positive I believe Redis as a company should focus on Redis fundamentals: so my piece has zero alignment on what I hope for the company.
- This means your difficulty is not programming per se, but that you are working on a very suboptimal industry / company / system. With all due respect, you use programming at work, but true programming is the act of creating a system that you or your team designed and want to make alive. Confusing the reality of writing code for a living in some company with what Programming with capitalized P is, produces a lot of misunderstanding.
- I hope the best for Zig, Loris. But even if Zig will survive and prosper (I hope for both), still I believe this is not a sounding decision and not the right attitude. I hope I'm wrong, but I wanted to share with you my reasoning. Here you are moving away from the open source marketplace AND from your main revenue stream. It's not similar to not posting anymore to Twitter. A better parallel would be not posting anymore on Hacker News anything Zig related, in terms of potential outcome.
- Wow. I think that's a serious mistake. Maybe GitHub is no longer so great and snappy but nowhere to justify moving something that needs: 1. Money, 2. Exposition, to something obscure just because it's a bit better. It's Git with an UI anyway, there isn't such large difference. I don't care about the fact the post is harsh: it's the content that it is broken from my POV because. It is absolutely legit to do something like that, in theory, but when you are handling a project that - at this point - is also the chosen language of a non trivial amount of folks, you need to act not just following what you like, but what is better for the project in the long time, and it is very hard to see how going away from GitHub (the fucking big market of open source software in the main city plaza -- let's use the same post tones) is better for Zig. What I think it is better is, of course, not absolutely better, but let's zoom on this issue root cause. It is the classical developer intolerance for tool that are not "as they wish/think", which is very common among technical people, but is a POV, I mean this "tool oriented" workflow, where this little feature/customization matters so much in your life (instead of adapting a bit and do not care), that I believe is a problem in our industry, and also has effects on the design philosophy of many programmers, that are too details oriented. Coders spend the majority of their life in the terminal, not on in GitHub. To check issues / PR there is not this Stranger Things Upside Down nightmare.
Another problem with that is that you know what you are leaving, but you don't really know what you find in the new place. GitHub used to go down often in the early days. Now they may not be snappy and unfortunately like 99% of the web felt for this Javascript framework craziness. But the site is always up, I bet has disaster recovery and serious backup policy, and so forth. Can you find this so obviously in other smaller places?
- Thanks, yep I know ControlXYZ, it's cool but far from practical, I have the feeling their contribution would be a lot more impactful if working with OrcaSlicer.
- It was an attempt at using less plastic, in vase mode there is this absurd thing that the first few solid layers are accountable often from 30% or more of the whole object weight! However, yes, it would be more robust that way.
I have a long love affair with vase mode and abusing it, and always saving time/weight is my main goal, like in this other case: https://makerworld.com/it/models/840291-vase-mode-gear-phone...
Your blog post made me thing that we would almost need a specialized vase mode site for models of that kind :D Moreover, there is no reason why the top surface could not be closed with bridging. The slicers have a lot of odd limitations in the context of vase mode.
- Thank you for your comment! The model can be also found on Printables. I have both an MK4 and a Bambulab A1 :D Thanks.
- Some time ago based on the same set of ideas I made this:
https://makerworld.com/it/models/99219-olivetti-style-vase-m...
You can create fast to print objects consuming very little filament, however to have some kind of texture on the surfaces is absolutely needed for strength.