The binary would not be nearly as useful as the source code. And even if the AI read the binary and copied it, the story is still it reverse engineered and re-wrote a c compiler all on its own. Which is still pretty impressive to me and has real world use cases.
Maybe Anthropic could release the logs to show how the AI was accessing the files.
They definitely have access to the source code, look at the binutils entry [1], Claude was debugging compilation of "preprocessor #define inside #ifdef"'s in binutils's code.
I was ok with it opening a new story at the end of season 2 because it’s a fascinating dilemma and I’m happy they wanted to explore a deeper question about the morality of having two souls in one body and how that affects their humanity.
However, it should end next season. This idea it’s going to be a long term project is going to ruin everything and now I am sad.
Ignoring what I think of the case itself, I hate how many headlines now are just the talking points for one side or the other in a dispute. At least try to pretend you’re being a neutral reporter rather than regurgitating what someone with an agenda says.
Another pro tip is to not pay at restaurants. If you can leave the restaurant fast enough before they give you the bill, they must have forgotten to charge you and sucks for them! The trick is not to bring bags so you can fake a trip to the toilet!
if you're not joking, actions like these are why we can't have nice things in society, it's cancerous behavior and just because you can, doesn't mean you should.
I think the two comments above yours are poking fun at the guy who is committing a felony by lying to federal agents. They're just making it obvious what he's doing is really shitty, anti-social behavior.
You are grossly misinformed and making an assumption.
You're thinking of being interviewed by a Federal agent. At no point are you being interviewed at a TSA checkpoint. Generally, they have two agents present for that so they can act as witnesses for each other. The FBI specifically uses the 302 for such an interview. Can you cite the relavant US Code here? I can.
Further, you're assuming I'm lying.
As someone who was present (in the room) as DHS was being formed and witnessed the negotiations around the TSA, the "really shitty, anti-social behavior" is sharing misinformation.
This is a scam that the GOP has convinced many of, that taking from the government commons is the right thing to do. But the GOP is the embodiment of a low trust society. I'd rather live in a high trust society.
I'd also rather live in a high trust society, but that's impossible with the government that we have (and it's not just Trump, although he has certainly turned our slow creep towards authoritarianism into a speedrun).
I realized that the GOP has been taking advantage of weaknesses in high trust society. This is an easy thing for fascism to do. So while I want to live in one, they aren't stable and must be protected.
You forgot when autocorrect fixes your word the second time, in 50% of cases you hit send too fast and have to send a follow up correction or edit the message
One that would've helped is "Founders at Work: Stories of Startups' Early Days" which is a collection of interviews with startup founders by Jessica Livingston. Even though this was a non-profit it effectively sounds like it was a tech startup (he was building an app that was based on, at the time (late 2000s), cutting edge machine learning technology). I think by hearing other founders' stories building other tech products, he would've learned about how they structured their organizations and led their teams. I liked the book personally, but the interviews will be hit or miss depending on the participant.
Why not a book about making co-operatives work, or managing direction in non-hierarchical orgs, management structure in public interest companies/charities, or establishing scientific organisations?
It sounds like he learnt lessons about the need for inclusion of users, need for strong information flow, what doesn't work (from his perspective) in hierarchy.
SV tech startups are an incredibly niche form of business, nobody should base their thinking on general management around them alone. They function within an ecosystem designed for them, like plants that only grow in a rainforest.
Most management books that most people recommend are not based on scientific evidence, studies, academic literature, etc. They're mostly memes; relatively recent books written by some kind of famous person, benefiting from the heuristics that make people favor the famous, successful, or high status [regardless of the fact that their lessons are usually from one source, type of company, culture, etc]. Compare that to practical management books based on evidence and studies; they're boring and old, or simply not catchy or sexy, so nobody recommends them.
There's also books that some people know about, and have a good track record, yet nobody follows. Deming's books should be mandatory reading for anyone in management, and anyone who cites Toyota as a model should absolutely have read them. But good luck finding anyone who actually follows the advice (same for Ackoff, Goldratt, Senge, Jacques, etc). Likely they are just too complicated and most people are not smart enough to manage this way.
Ok so walk me through how _only_ focusing on the browser core will make them money other than continuing to be dependent on Google. How can they diversify their revenue streams?
I agree they should make the browser core good, but right now they are entirely dependent on their biggest competitor.
Falling behind on the browser core makes users leave Firefox because the one non-negotiable requirement users have is that the browser works on the websites they need it to work on. Literally nothing else Mozilla tries to do to diversify their revenue streams will matter if the browser they're trying to build it on top of is not sound and functional.
If they weren't slipping on the core browser as much as they have been, there wouldn't be nearly as many shouts as there are today when they instead spend resources on chasing the latest fad.
- Start accepting small individual donations solely for the Firefox team (rather than generalized Mozilla stuff that goes on anything but Firefox).
- Start crowdfunding for features.
- Go RedHat route, offer an enterprise version with centrally managed profiles and DLP feature. Not exactly free-as-in-freedom stuff but still better than adtech.
- Get some EU bureaucrats thru a FUD session against Chrome (does it counts as FUD if it's true?), then apply for some EU funding program. Dirty and messy, still better than adtech.
None of this is particularly lucrative or clean, but I don't see how AI would bring them any __revenue__ (do not confuse with investment) at all. There are too many players there already and many of them are more established - and what Mozilla have?
- Their engineering team? Maybe.
- The browser engine? Completely irrelevant (and that's exactly the problem).
- Their userbase? The userbase they have left seems extremely averse to value-added features in general, and the AI kind in particular.
Then assume they focus on integrating AI into the browser, how do they monetize it next? Sell data? Then there is no reason to choose them over Google. Charge for interference? No chance to compete against established hyperscalers there and would go against their local-first selling point.
The sad truth about platform-crucial software like a web browser is that monetizing it in any way inherently reduces it's value for users. And in case of Firefox it's a pretty small margin that keeps it competitive.
> - Start accepting small individual donations solely for the Firefox team (rather than generalized Mozilla stuff that goes on anything but Firefox).
>
> - Start crowdfunding for features.
Just these two things would make me happy (assuming the crowdfunding goes to the Firefox team as well).
I don't know any of the Mozilla execs but from the outside it looks an awful lot like some grifters were attracted to the free Google money and took money from the people doing the actual work.
If I'm wrong, my apologies. There just seems to be a lot of high salaries and a lot of developer layoffs.
Just to be clear, they are NOT deactivating IntelliSense which suggests classes and functions.
This is an AI inline code suggestion tool using local LLMs.
Not great but may or may not impact your workflow. I love using agents, but Intellijs inline code suggestions (also based on a local LLMs) are usually useless to me.
Very interesting that Valve and Framework seem to be throwing their eggs in the Arch basket over Debian/Ubuntu. When I got my first computer, I installed Ubuntu because it was dominant. Maybe it still is for the average Linux downloader, but why are the Hardware companies more into Arch?
We also sponsor Debian. We are distro-agnostic and pick our sponsorships largely based on what we see Framework Laptop owners using in our post-purchase surveys and community polls.
Great to see this! I have two Framework Desktops running with CachyOS, for AI workloads, software development and light gaming. Great machines!
I wish the header for the power button and LEDs was rotated 180 degrees so the NanoKVM doesn’t need extra cable adapters.
You need a minimal base OS to have the flexibility to build your own stuff on top of it, and you don't want to be at the behest of another corporation. That rules out Fedora, Suse, and Ubuntu. You also need it to be popular and have good hardware support. So the only two realistic options are Arch and Debian.
My guess is that Arch is easier to build on top of because they have a stronger culture of leaving packages as unmodified as possible relative to their upstream sources, whereas Debian maintainers seem to have the opposite culture. A Debian system has a lot of Debian-isms in it overall, whereas the Arch-isms tend to be more like generic sensible defaults rather than OS idosyncrasies.
And over Fedora/RHEL. If I had to guess, it could be that new entrants find it easier to submit changes to Arch Linux packages [1]. ChromeOS also steered away from Debian-based distributions, choosing a Gentoo base.
I'd think it's because they're introducing updates to address issues w/ the hardware quickly and want a rolling-release distro so users can get the updates faster.
Debian testing is about as stable as it gets while also being a rolling distribution. The promotion of package updates from unstable to testing does not take that long either depending on the severity. I would venture a guess that there is more to it.
Personally I'd also think it would be a better engineering choice for Valve to base SteamOS on Fedora Atomic, as it supports the immutable OS paradigm a lot better imo. Especially now with progress in bootc/oci/ostree.
> Arch Linux is an independently developed, x86-64 general-purpose GNU/Linux distribution that strives to provide the latest stable versions of most software by following a rolling release model.
> This page complements the Installation guide with instructions specific to Apple Macs. The Arch installation image supports Apple Macs with Intel processors, but neither PowerPC nor Apple Silicon processors.
(FWIW, I understand that there is benefit to good coverage of a narrower scope, but I do wish Arch would fold https://archlinuxarm.org/ into the main project and be officially multi-arch, but that is not the world we live in.)
Arch package manager here, there is ongoing work behind the scenes to support multiple architectures (aarch64, riscv, etc), but as our volunteers (myself included) are doing this in our free time, progress is up in the air.
I use Linux for 20 years and I study programming for 10 years, bought 100 programming books, so a linux distribution is basically a programming language container. Slackware was for lisp. Now instead of kiss its simple ain’t easy with clojure. Debian is very tightly tied to perl with both communities bent on reproducibility. (Tho rust is replacing perl). Red hat and ibm is a Java shop. Centos is a scala platform at cern. Ubuntu and Python is a data science platform. Sure is a better Debian like ruby is a better Perl. And here we come to arch, when 10 years ago after a brief stunt with Perl basics I started learning c# the first thing I did is try to run the excercises on the raspberry pi. But because of some hard float soft float something they didn’t work. So I had to jailbreak the raspberry pi and run this new distro on it, the arch Linux. Where it just worked. You see ethics of ai and maybe like data science require the system to be fsf endorsed free system that’s what Debian gnu linux reason detre is. And Debian as Steve Jobs with Java were like against mono, you shall not pass. But for a gaming platform that’s a little bit different. Ex red had ceo now works at unity this mono fork. Ms bought blizzard, they want into this gaming thing badly. So that’s why steam os is arch now, less strict than Debian on the Libre side of things. The rest is history. :D
They have a “throw everything at the wall and see what sticks.” Sure it has a lot of crap but they also have major hits like Squid Games, Stranger Things, (both became cultural phenomena) and Daredevil.
Maybe Anthropic could release the logs to show how the AI was accessing the files.