Did anyone ever say any of this was "good software"??
Its lowest common denominator crapware. Why rebuild it? Was it really your lifes dream to build that?
SWEs in general have been paid significantly higher salaries for decades in part because of the skill level needed and that they are generally working on capital projects (build new thing) vs operational projects (maintain thing built). It's the difference between needing a Landscaper to come in a completely reshape a bit of land for £10k once, and Gardener to come in a mow/prune once a week for £10. Most salaried SWEs don't necessarily see that every project they work on is a new patch of land to be landscaped; those who've contracted do.
SWEs have been effectively paid this "danger money" because of that inherent volatility in projects being shut down, which has been there since the dot com bubble but is now amplified because of the relative size of the tech sector.
The skills barrier which has been eroded in part by the lack of professional standards (anyone who can edit a PHP script can call themselves an SWE vs a Chartered Accountant which has to be certified), and now explicitly because of AI.
In some ways this just looks like "FOSS for Angry Millennials" without the understanding of the further damage that this will do to them and their industry. These are the forces its trying to get revenge for: (a) high inflationary creating risky environments for businesses to navigate, (b) tax changes which have impacted how organisations pay for SWEs, (c) AI offering "efficiencies" in all parts of the value chain, (d) the demand for ever-growing returns by shareholders as we enter the end-state of neoliberalism.
None of that is improved by this movement. Regardless of what you think of them, the original Occupy movement made it impossible for the people responsible for the crash to not be aware of/somewhat put out by those impacted by the crash. This is Just Another Open Source project. The way for those impacted is: (a) keep their skills sharp by helping charity projects that need the volunteer support; (b) get active politically on a local level; (c) use the time to start their own businesses with others.
There is a firmament that is needed on which all tech grows like Linux which is a good thing. This is not good for the Landscapers or the Gardeners, nor does it fix the issues which have led to the layoffs.
2. People and companies often just use SAAS for convenience. Of course we could host just about everything ourselves, but that's a shit ton of work with variable payoff.
3. Using Github and Discord. LMAO!
I noticed that others are taking a shot at the big boys like Photoshop, but I don't think that makes sense since 1. Jesus Christ imagine trying to rewrite Photoshop, with or without Astra and reverse engineering, let's be real! And 2. GIMP already exists and people don't use it for various reasons an ultra nerd that doesn't have osx or windows on any device in his house (me) could understand. Instead I'm looking for small niche markets that are locked in by the software equivalent of a petty village despot. Decent examples so far include quilting software and funeral home SASS. I'm also looking for anyone stuck on old dead software with no migration path because of no export functionality or nothing supporting whatever given file format anymore. Apparently QuickBooks desktop falls into this category, but the space feels full enough, with e.g. GnuCash, that I don't think I would be much help there.
Investigations so far: https://github.com/508-dev/genairosity/issues
Basically leveraging subsidized tokens to liberate proprietary software. A bit of a petulant table flip maybe, but it kinda feels like the only thing left to us. Either we can kill ourselves scrambling around doing 996 for startups to try to avoid being a permanent underclass... Or we can light it all on fire.
Also inspired by open weight models tanking possible profit margins on frontier LLMs. One form of anti capitalist activism could be to just make it impossible to monetize something. The Manfred Macx form of activism.
Many other people in this thread with similar projects seem to be following the same workflow:
1. LLM agents sent out to find locked in proprietary software without a FOSS alternative
2. Targets selected by a human for deeper research on feasibility and alternatives
3. Clean room plan established
4. Human uses the software and takes notes
5. Feed notes into a token subsidized frontier model for orchestration. Architecture and plan written, beginning normal agentic development flow.
I'm slowed down significantly by step 4. Perhaps I will look into contributing my efforts to a larger project such as the one linked instead.
PhotoCraft is already better than GIMP3 in every aspect. Usability is much better, binaries are 4 times smaller, and it's way faster. GIMP takes 5+ seconds to start, vibe-coded Rust Photoshop starts instantly.
I think the "old" open source community has very low standards for user experience when it comes to GUI software. It's one of the main reasons I could never get myself to use desktop Linux.
With AI, there are no excuses for open source software to be worse in any aspect at all.
Developers should have higher performance standards now that DX matters way less (you're not writing code by hand anyway).
Regarding Photocraft, I've been following it with interest. My impression is that it comes off initially as an incredible accomplishment, 1:1 with Photoshop, however extended usage reveals more and more flaws until it's determined the software isn't really usable yet https://www.youtube.com/watch?v=BGShygDux9o
This is my general impression as well with overall excitement around LLM assisted reverse engineering: very easy to get something that looks like a slam dunk, until you actually try to use it. Then begins the long, long, long slog of the work needed to make the software usability and parity, which apparently can't be done by LLM agent swarms yet.
The video you linked is also three weeks old, which is an eternity in the scope of this project, since each week hundreds of PRs are merged, and several versions are released. At the speed they are going and with the amount of attention and usage they're getting, I expect they'll manage parity in months, but even in the current state, it's still useful enough.
The person testing it also seems to be an artist, so their workflow demands can be significantly different from someone doing image editing. Most artists I know use more specialized software like Clip Studio Paint over Photoshop anyway.
I think to stay competitive, open source software needs to adopt AI-first workflows that mostly automate quality control and review, rather than just allow LLM contributions. Allowing LLM contributions without sufficient infrastructure will just drown maintainers in slop.
> Then begins the long, long, long slog of the work needed to make the software usability and parity, which apparently can't be done by LLM agent swarms yet.
We think it can be. This is just Pareto principle. We're hill climbing the finer curves.