This is also a reason to have functional LoraWan mesh network

fyi LoRa is a useful building block for building offline (non-internet) mesh networks but LoRaWAN is not: LoRaWAN is for IP-over-LoRa and it requires one of the nodes to have internet access to talk to a centralized server which mediates everything. (tldr: forget about LoRaWAN.)

Sorry I don’t have time to braindump on other things in this space, but I will say https://columba.network/ looks interesting - it is a reticulum client for android which can use LoRa (if you have a LoRa radio) and also BLE (100m range) and WiFi (both locally to other nodes and to the internet to talk to distant nodes). I have not actually tried Columba myself, but i think it looks promising.

edit: looking closer i think the one big missing feature in Columba for this type of application is group chat. Unfortunately Reticulum doesn’t have a decentralized group chat yet; its 1:1 chat protocol (LXMF) is completely decentralized (two people can keep talking if they’re in range of each other even when they can’t see anyone else in the mesh) but the group chat protocol it has now (called RRC, similar to IRC) requires a centralized relay server. You can run an RRC relay yourself, including on a non-internet-connected network, but it only works for people who are able to reach that relay (and it doesn’t keep history so you miss what was said while you couldn’t reach it). I see that the Columba author has actually implemented RRC for Android, but it’s currently in a separate app called eridanus. I don’t know if anyone is trying to build a decentralized group chat protocol on top of reticulum yet but probably someone is. It’s a very complicated problem to solve well, though.

i used https://marshdeer.github.io/xkcd2501-generator/ to remake it:

(posted here in [email protected])

nowhere did I say it was propaganda in this instance

you didn’t say it but you strongly implied it by saying in this thread that i “spread propaganda” without mentioning (until your later comments, thanks) that you don’t actually think what i’m saying here is “propaganda” (by which you mean, i infer, “incorrect”).

I am saying you spread propaganda because it has been proven time and time again that you do.

proven! lol ok, sorry i missed my trial 😂

fwiw depending on the definition of propaganda i could actually agree with you - a lot of what i post is advocacy for something (as is some of what you post).

eg, looking at a few of my recent posts: is this propaganda? or this? or this? i would say all three definitely are, in the sense that they are politically motivated communication, and i stand by them.

but you’re obviously using the term “propaganda” in its pejorative sense because you disagree with (some) of what you’ve seen me advocating for (and/or agree with other things you’ve seen me debunking), which imo is intellectually dishonest. rather than say “i disagree with arthur about XYZ, but in this case the evidence is lacking for the claims being debunked in this thread” you just said that i “blatantly spread propaganda” and drew a spiderman meme equivalency between me and the other person who you now acknowledge (thanks) is “more wrong” in this case.

i assume you’re probably saying i “spread propaganda” because i posted something you disagreed with elsewhere, but do you actually think i’m spreading propaganda here in this thread? i am honestly, out of personal curiosity, trying to determine if there is any truth in this narrative about the yandex datacenter strikes. i think i’ve been clear that i’m not actually absolutely certain that there is not, but so far all of the “evidence” for it i’ve seen presented is falling apart.

The previous post is an ml propaganda account.

imagine thinking this is a propaganda account, lmao 😭

Where did the recent drop in social media engagement of European far-right party members come from? I posted an article yesterday with examples for that, and everyone can go these accounts looking it up.

This post? That is the same article as is linked in this post, and exactly what I did look up to verify in my comment at the top of this very thread.

I admittedly only checked their first example, to be fair, but given that the fascist who’s lack-of-retweets they’re saying is evidence also had just 6 RTs on a tweet before the strikes… I expect to find the similar non-evidence in the other two also.

To be clear: I don’t doubt that those chuds have lots of fake followers, and they probably do buy twitter engagement, and sure they could be buying it from Russia… I’m just saying the narrative about these yandex strikes causing a drop in inauthentic behavior appears to be bogus as far as I can tell.

From the article:

As per the bot analysts at Tom’s Hardware, bot traffic from the Yandex network has dropped to just 0.4% of the maximum number of HTTP requests recorded in the past two weeks. That means activity from automated, Russian-hosted accounts has been virtually eliminated on social media.

That… is not what that means. The HTTP requests graphed by Cloudflare shown in the Tom’s Hardware article are the yandex web crawler. This is traffic from YandexBot, which is their search engine’s equivalent to GoogleBot. That drop in traffic doesn’t indicate anything whatsoever about russian-hosted accounts on social media, it’s about a search engine’s bots reading web pages. 🤦

Note that the TomsHardware article graph comes from here: https://radar.cloudflare.com/bots/as13238 - this is one of Yandex’s AS numbers. In case you doubt that the traffic cloudlfare calls “bot traffic” there is really just the YandexBot, and you really want to believe that maybe Cloudflare is classifying social-media-posting HTTP requests as “bots” in that graph, you can check their graph for YandexBot specifically here https://radar.cloudflare.com/bots/directory/yandex and you’ll see:

Besides grossly mischaracterizing what the Tom’s Hardware article says by implying that the YandexBot web crawler has something to do with the kind of bots they’re talking about, this The London Economic article also cites some slightly more plausible sounding evidence for reduced social media activity:

Take Richard Tice for example. Below are three posts on X from the Reform deputy leader to his almost 400,000 followers on Monday, 11 October. At the time of writing, none of the posts had hit triple figures for reposts or comments, with one not even reaching 10,000 views.

Could British fascists be getting retweets from bot farms in a yandex datacenter? Sure, they’re a cloud provider, they’re no doubt hosting some stuff like that.

But, does this example actually demonstrate it? No. The first yandex datacenter strike was on October 8; here is a tweet from @TiceRichard from October 6 with only 10 replies and 6 retweets:

If there has actually been a measurable drop in “inauthentic” social media posts (or other retweets or likes or whatever) as a result of the yandex datacenter strikes, I still haven’t seen evidence of it.

And we’ll see if Adobe goes after them on the basis that it’s warez or not. I think it’s entirely likely that they might even if there’s no legitimate ground to do that. Companies are well known for bullying open source projects that are viewed as a threat to their profits.

Yes, I agree it’s possible that Adobe goes after them for something even if they didn’t decompile photoshop - and also even if it isn’t really a threat to their profits.

I do not care what level of feature parity it has with Photoshop or GIMP. My only interest is that it works for my use case, and that’s what I tested. It solves a problem for me as an individual by providing me with a tool that has the UX I like and the functionality that I need.

Cool, if it works for you then I wouldn’t tell you not to use it! I’m just skeptical that it actually will; I assume you haven’t used it enough to really find out how well it works in practice yet.

It’s not like I’m forced to look at any of these projects, or use them.

I guess we’re not forced to look at them but to avoid looking at them we’d need to stop reading the internet. I only looked at this one because I was conflicted about if i should delete this post from /c/opensource here - the discussion has been informative so I certainly won’t now, but i’m still conflicted about the fact that by leaving this post here we’re potentially helping this genAI huckster get more paying customers for his token-reselling business which this project is obviously a loss-leader for.

Btw I found the video I mentioned earlier, here is a link to the part comparing PhotoCraft’s Levels adjustment to Photoshop’s: https://youtu.be/Oc51Ty9h-68?t=216 (and later he compares other basic features).

And whether it’s copyright infringement or not is of little interest to me since I view copyrights as being inherently harmful.

But from that perspective it would be much easier and more effective to simply use an LLM to make a crack for pirating the original software than to try to reimplement the whole thing. Or, to actually decompile and recompile it to get something (pirate-)source-available with actual feature parity. (People are also doing both of those things with various proprietary software, of course.)

However, since the implementation is using a different language, that does make the implementation itself a clean room one.

That isn’t what “clean room” means in this context. It is widely accepted legal doctrine that direct ports of software from one language to another constitute derivative works under copyright law.

This project’s claim to be “open source” would be false if it is actually an automated port of the output of a decompiler of proprietary programs; that would make it warez.

And it still takes time to add features using LLMs. It’s not magic, and I’m more amazed how much they managed to add already rather than it not having parity yet.

Indeed. But if they haven’t gotten around to adding more than the most basic features in the Levels adjustment window (probably one of the most commonly used features of Photoshop) I assume they’re missing a lot of other stuff too.

It does appear to work with PSDs fairly well, but again I don’t have really complex ones because I use the tool in an amateur way to begin with. […] So, the feature parity isn’t really the stumbling point for me, it’s the UX.

OK, I’m taking that to mean that you haven’t actually identified any specific features (or PSD compatibilities) that PhotoCraft has over GIMP besides the UX? It seems to me like using an LLM to improve the UX of GIMP, on top of all of its existing (presumably far greater) set of features, would make a lot more sense.

But this is something that’ll sort itself naturally over the coming months. If the developer stays active and continues working on it, then it’ll keep getting better. If they abandon it and move on then it’ll either die or somebody else picks it up. But it is in a fairly usable state as is from my experience, and I’d argue a lot of the hard work has already been done. So, we’ll have to just wait and see what happens with it.

I concur, in six months or a year I can either eat my words or say “I told you so” :)

And if we get to the point where all this stuff can be done using local models, then we really could see Stallman’s vision for GNU become a reality. Regular people, who aren’t developers, could use fully open source stacks, and they could modify software to work exactly the way they want. If we’re lucky this might kill closed source software entirely in the long run.

I wouldn’t hold your breath, but perhaps with a lot more advances and if/when local models can perform coding tasks at reasonable speeds on affordable hardware… maybe? Meanwhile right now we’re drowning in new 1M LOC slop projects overpromising and underdelivering.

people are literally just reverse engineering these applications by having the LLM decompile it and observe what it does

Have you seen the creators of this particular project claim anywhere that they decompiled anything? I’ve seen several commenters claiming that but I think they’re incorrectly assuming that that’s what they did.

For one thing, decompiling and recompiling Adobe or Microsoft code would surely be viewed as copyright infringement if there isn’t a clearly-documented “cleanrooming” process in the middle (having an LLM write a spec (in English) from the decompiled code, and having another session without access to the copyrighted code write new code from that spec - a fun LARP many people are doing lately).

But also, if they were actually doing that then I’d expect more feature parity than I’ve seen in the few youtube videos I’ve skimmed through showing these apps. I’m too lazy to find the link now but I watched one comparing PhotoCraft to Photoshop, and one example was that PhotoCraft’s Levels diagram is very primitive in comparison to Adobe’s (or to GIMP’s for that matter).

I did something similar recently where I had an LLM figure out the proprietary format for the way my Nikon camera stores RAW files. There’s currently no open source implementation for it, and now I can read them using my own library. So, something that used to be a herculean effort before, is something you can now do in a few days using an LLM.

Congratulations, that’s awesome (and I don’t doubt it). But implementing a RAW format is a lot easier than re-implementing all of the Photoshop features which are required to be able to say (honestly, a least) “this has .PSD file format compatibility” without a long list of caveats.

I am using his PhotoCraft and LightCraft, and they work fine for what I need. I can process my photos using completely open source tools now that have a UX I’m used to.

I am curious how much you’ve actually used it. Eg, can you confirm what I saw on youtube showing the Levels adjustment dialog lacking most of the features found in Photoshop’s and GIMP’s? (Is that still the case in the current version?)

And have you noticed it having compatibility with (or at least some support for) any modern PSD file format features which GIMP lacks?

You’re saying this as if NPM ecosystem didn’t exist before LLMs.

yes there was a lot of crap before 😭 but don’t you agree it’s 1000x worse now?

You look at the number of stars a project has, how long it’s been developed, how many users it has, and so on. This gives you an idea of whether it has legs or not.

This project has a lot of stars but only because it just got a ton of clickbait press grossly exaggerating its completeness. I assume the vast majority of the stars are from people who have not tried running it. How long it’s been developed is… two weeks, as of today. How many users? It’s far too soon to say.

I’ll certainly trade having to rely on proprietary software for the problem of having too many open source options.

I agree with you here! But I’m assuming (perhaps incorrectly) that this particular one is not likely to ever become a particularly good option.

for people wondering what “folios” in Linux memory management are: https://lwn.net/Articles/937239/

I don’t think any decompiling was involved here; if it was, Adobe and Microsoft would have DMCA’d it off of github already.

LibreOffice is fine, although compatibility with MS Office is still spotty

Have you seen anyone claiming that this brand-new vibe-coded thing has better compatibility (and/or feature parity) with the proprietary applications it claims to replace than the mature free software solutions currently do?

I would be surprised if it actually does anytime soon.

It’s very obviously better to have open tools than closed ones.

I agree, if it works well and it’s freely licensed it’s a good thing. It isn’t clear how well this actually works though, and the fact that the author is disparaging existing free software solutions to advertise his business selling genAI tokens is not a good indicator.

This whole screeching about slop is both incredibly tiresome and irrational in the extreme. These are new tools that make software development easier […]

It’s made it so easy that there is currently a deluge new projects which often seem plausible at first but mostly turn out to actually be garbage upon closer inspection.

It’s irrational not to be skeptical of a two-week-old project claiming to be ready (or “almost” ready, lmao) to replace multiple 30-year-old suites of software.

Let me know if I’m mistaken, and anyone other than the author tries it and credibly claims that this thing is actually a better (in any way whatsoever) replacement for MS Office than LibreOffice is.

Is OpenOffice still maintained?

The descendant of it called LibreOffice has been the well-maintained thing to use for the last 15 years.

you don’t want to “svimma” in a lake in swedish

TIL svimma in Swedish is to faint, or pass out, and some UK dialects have the related sweem (more commonly swoon in English)

this meme is exaggerating for comedic effect; there are much shorter routes which can also make it good gas. (eg)

Not sure where you get the idea that I posted this as a marketing ploy reselling genAI tokens?

Sorry if it seems like I meant to imply that that was your intention; i did look at your profile before asking and saw that you aren’t a spammer :)

However, it seems like reselling tokens is the primary business model of the company which has presumably just spent a significant amount of dollars worth of tokens to vibe code this suite of software.

Whether they have actually produced a dozen useful apps which many people will actually use, or merely made something which is just barely functional enough to get a bunch of free marketing hype, is not yet clear to me. They have certainly succeeded in getting a lot of press, though.

Should this be allowed here in [email protected]?

It is obviously built using Claude, but it is not (as I first thought) a project that started a month ago - the first commit in the current repo is from a month ago but it refers to their previous (“multi-hundred GB” lol) git repo. So this is actually built on something with history going back to before the vibe coding era. (edit: actually i take it back, i see now photocraft started on september 30 and the commit message on the second commit is “one-shot” 🫠)

I see the project lead’s two most recent posts on the site formerly known as twitter are disparaging LibreOffice and Gimp and Kdenlive.

Until yesterday this part was under a “fair source” (meaning not open source) license but now it has been relicensed MIT + Apache 🤔

Have you actually tried any of these 12 apps yet, OP? Are they any good, or are they just marketing for a service reselling genAI tokens?

On href, I think all 3: style.css, /style.css, ./style.css point to the same place.

Are Anti-AI Linux distributions practical?

It’s not possible to have a 100% no-LLM-coding-involved Linux distro anymore now that the kernel itself is welcoming LLM-assisted contributions, but it’s still possible to say that everything which the distro produces itself will not use AI.

How can AI commits be responsible?

What Debian means is: by reading what it produces and not just blindly accepting it but having humans review it carefully. The quality of LLM-assisted software is highly variable. Experienced people like Debian and Linux kernel developers are much less likely to publish very bad slop than inexperienced developers who are newly able to make a thing work well enough to ship it.

I am concerned about the extent of Debian’s embrace of LLMs but I still have a lot more confidence in them than most other distros (especially small hardcore-anti-AI distro projects lacking sufficient developer capacity to even keep up with security updates).

I expect we’ll be seeing lots more things like this non-upstreamable-because-AI hardware enablement and distros like Nura (formerly postmarketOS) will either either adopt more nuanced LLM policies or be forked.

I am doubtful about if there will be any serious distros with hard LLM bans in a few years from now. But also, perhaps the bubble will have burst by then, so hosted inference won’t be so cheap and the use of LLMs will probably be quite different.

the primary reason is AI is a thief

This is not the primary reason for most people, I don’t think. I think the fact that it increases complexity while reducing human understanding is the biggest problem; it’s possible (easy even) now to produce mountains of good-enough-by-many-metrics source code which the person doing the prompting has very little comprehension of. The fact that it is also a plagiarism machine is far less of a problem than this. (There are a litany of other problems too of course, several others of which I would also rate as worse than the plagiarism aspect.)