Sarah Jamie Lewis

It just really sucks that so many free software tools from the kernel and compilers up are rapidly reconfiguring themselves in service to slop generators. I am so grateful for the effort people are putting to create human-centred forks. But one thing I do think about a lot is the mental toll of constantly chasing a compromised version of someone else's vision v.s. pursuing new ideas. The whole thing just feels like ddos on the attention of people who should be working on bigger things.

Feeling nostalgic for when browsers didn't need data retention agreements with AI companies in order to render web pages. Also unsurprising how fast Mozilla are pivoting from "totally local / in browser / private AI" to clearly not that. Remember to "opt out". "Privacy protections are built into how Firefox Smart Window works: conversations aren’t saved on Mozilla’s servers by default, and partners like Mistral agree to zero data retention. " (source: mistral.ai/news/mistral-x-mozilla/ )

This evening I spent some time going through the latest firefox HEAD, with the idea being to catalog how many files/lines of code could be removed solely on the basis that they only exist to power the built in llm / chat / local ai stuff. I am currently at 136000 lines of code, a number so absurd that I have forced myself to triple check it - and I'm pretty sure I'm barely half way through this patch. The security auditor in me weeps.

Folks, if your personal outlook is "this is too hard / it will never happen" - that's great, I have that voice in my head too and I also believe it occasionally. But sometimes you have to tell that voice to shut the fuck up, and actually try to do something that seems hard or difficult or even impossible. Especially when the other choice is to do nothing and watch the world rot. You are, of course, free to do that but I am not, and I'd prefer you did it somewhere not in my replies.

Apropos of many ongoing threads discussing the state of modern floss: Main thing I am thinking about: How to design an operating system that, as a consequence of interface, architecture and structure, minimizes both the overall number of required dependencies and the depth of dependency trees of any and all hosted software. Once you have that, I think many other challenges become tractable (and many proposed "solutions" to the catastrophe of modern software complexity lose their appeal)

Mozilla and, by extension, Firefox have laid out the direction they want to go - and have consistently moved in that direction. There is no redemption arc there, they are not going to turn the ship around - and every week and month that goes by, Firefox gains more slop and drifts further from the visions that founded it. Projects like Tor Browser and Waterfox are painstakingly disabling / patching out the worst - but every release becomes more expensive, and things do slip through.

Oh I forgot to mention, If you would like a number which I think epitomizes the scope and scale of these AI features e.g. if you are at all concerned with how complexity of software relates to maintainability or security... The size of the *patch* that I produced (whose sole purpose to to remove AI related features) is 37 megabytes.

An uninformed plan:

• Take Tor's base Browser, the one they use as a branding-neutral base (very clever name) for the privacy-focused soft forks Tor Browser and Mullvad Browser (i know, i know - but they already paid for the work, so some good might as well come out of it)

• Rebase your soft forks on Base Browser, and benefit from that work

• Form a working group to co-manage base browser

• Hard-fork from firefox as you reject future rebases in favour of privacy and humanity.

I understand politically/organizationally why this module exists, and I understand that Mozilla's priorities will never be my priorities. But given the number of conversations/messages I've had in recent days on the topic of "how many lines of code browsers have / how complex they are." I would just like for it to be understood that much of that complexity isn't inherent to the nature of browsers, it doesn't need to be there (or in many cases, anywhere)

For those wondering: I did this by fucking hand, it took me ~5 hours.

As previously stated, I am not the correct person to manage this project - I'm not even sure this should be a project - I just reached a point where I needed to do *something* and this is something. So I deeply encourage anyone who is at all interested in a potential future of a non-slop, human-centric web to dive in and contribute patches/ideas/feedback whatever (or do something else useful, like contribute to Servo) If we want that future we are going to have to build it.

Bootstrapping a project like this is a daunting task. But we are where we are. I am almost certainly not the right person to manage this, but I believe this needs to be done enough to try and get the ball rolling. I've setup https://codeberg.org/basebrowserproject/organization/issues/1 as a place to start to gather notes/issues/ideas and connect people - if you have some skills or any other ability to help, and want to help, then get in touch.

But that plan only works if enough soft forks come together and commit to it - it will take serious combined effort (and a lot of community support and funding) to redirect the future of the web. I still have enough hope and optimism in me to believe that might be possible.

By uninformed I mean that I've been thinking about this path for over a year, and I honestly believe it's the only way forward https://mastodon.social/@sarahjamielewis/114088612358750696

I tried to be as surgical as I could with the removal, so there is likely still a fair amount of vestigial code that I didn't notice or overlooked. I also didn't intentionally remove any of the AI-related code in the android source, in an attempt to minimize the scope. If you are interested in this absurd project and are comfortable with building firefox from source, feel free to test it out and report any issues.

This has been a very useful and cathartic exercise for me, if only in the sense that I now have a much better idea of which parts of Firefox I wish to be different, how big those parts are, and the extent to which they are integrated with the rest of the browser (the fact that we are now over 1.25 million lines still hasn't settled in my brain yet) I have a few more patches I'd like to sketch out, my plan is to keep chipping away at this in whatever spare time I can find.

To make clear my reasoning behind my thoughts, a little story: I made the decision to adopt veganism well over a decade ago. One evening I had a realization that my values were not aligned with my actions, I did not like that feeling, so I did something about it. I've recently realized that I'm feeling that same kind of dissonance again. Firefox, or any modern web browser, has progressed to a point where using it has began to come into conflict with some of the deepest values I hold.

Based on the response to my thread yesterday, and the fact that I have been having conversations along this line for a year I really believe that there is demand for such a thing (and IME with doing non-profit dev work for the better part of the last decade: demand is not uncorrelated with the ability to long term fund a thing). People want free and independent browsers, unencumbered with non-consensual bullshit, built by humans.

A scenario I would love a word for: Firefox has code earmarked for "SponsorProtection" i.e. elements that will have additional protections applied to them so e.g. clicking on a link on a sponsored newtab element won't leak IP address to the sponsor (just to Mozilla's relay) So there is this doubling of functional complexity: first the new code needed to display and manage sponsorship elements AND then the new code needed to (partially) protect the client from the sponsorship functionality.

Because of this I am not overly concerned with appeals to notions like "market share" or "average users" as an end to themselves. I firmly believe that this software, in it's manufacture, in the way that "features" are exposed to the people who use it, in the philosophical direction of it's roadmap is unethical; it is introducing harm into the world. And my first priority and motivation is to minimize or entirely eliminate my participation in those systems in whatever way I practically can.

Soft forks i.e. disabling as much as possible is a stopgap that will get harder over time. Maintaining a hardfork base that ruthlessly cuts out stuff and attempts to hold onto it's own security will take a village I'm not sure we can find (but if we could, things could get incrementally better - maybe?) Servo has a village, but is on the other side of the functionality gap. Would you rather spend your hours funding new human written code, or on the elimination of generated garbage?

A subtractive approach to browser dev (whether soft or hard fork) can only go so far.

Ultimately the root problem, as many noted:

1. "web standards have become obscenely complicated - for various reasons - and developing a browser necessitates larger team efforts which perpetuates more complication."

There are two ways to respond to that: attempt to turn the tide (and try to avoid getting swept up in it) or abandon the modern web in favour of something that allows smaller browser to blossom.

I've started unloading some notes from conversations I've had, and some stuff that was previously in my head. If you are looking for a place to start, how about recommending the "features" that would need to be removed from any good base: https://codeberg.org/basebrowserproject/organization/issues/3

And the Linux kernel is far from unique in that regard, the full embrace of age verification and LLMs by systemd is another example (if one I have way more quibbles about) These attitudes are pervading the core layers of critically important projects, and we need counters.

I don't think that is the most efficient way of bootstrapping this thing - given the amount of effort these projects are already expending purging the same garbage functionality from Firefox release after release I think there is a world where a joint effort results in more-capacity (especially if committing to developing that base independently after a point) But I have no way of proving that, and thus a new project might be the only practical way of bootstrapping this thing.

To respond to a lot of replies at once - the terms "privacy", "security", "censorship", "trust", "encrypted", "anonymity", "jurisdictional flexibility", and "deniability" while all sharing related domains are all technically describing very different properties. My work and research depend on being able to treat these properties separately, and understand each in isolation (and warning that it is important to not conflate them with each other)

It is incredibly important to be specific about what properties apply in situations to be able to make the correct trade offs. An in that respect, VPNs are incapable of providing privacy - under any specific definition of privacy. They can route around censorship, they can provide security, they can permit jurisdictional and trust flexibility. But they can only do those things under very specific assumptions which require sharing additional information (albeit changing who gets what data)

@sashin@veganism.social I don't know if this will ever become anything. I do think it would be a good idea if many firefox forks came together and committed to co-maintaining a base that was outside of Mozilla direct/indirect authority. Consider this project an exploration into the viability of that.

I have the kind of life and personality that really gets into the latter (I have written at least one gemini/mini web browser, and I will almost certainly write another https://mastodon.social/@sarahjamielewis/112928301142039640) (One of the many reasons I am not the person to manage a general purpose browser project) But I do recognize that while much could be replaced with standalone clients (to the despise of many) it would leave a very big gap between what we have and what is needed.

Do I love the existence of an barely documented binary blob in an open source browser project attached to "filtering" functionality? No, I can't say I do. According to contemporary reports and discussion, this filtering functionality was first added so that such sites were not displayed next to "commercial partners". So that's fun.

@d1@autistics.life 100% agreed this is an absurd project, but on the other hand...at some point I think there is an argument to be made that the risk reduction obtained from removing hundreds of thousands of lines of code from a code base outweighs the effort to track other fixes...

I still think the IP assignment situation on the fediverse is in an absurd state. If you are a registered user on a big instance you are required to agree to provide a worldwide licence to your content. But someone with followers on that server, can send content to that instance and have that content be distributed without any such transfer of IP (that's just how activity pub works). Really feels like a vestigial artifact inspired by the legal realities of centralized social media.

@yoasif@mastodon.social if/when this develops further I will attempt to find the resources to provide binaries (it's not quite the intended end-goal for this project, but I agree it's likely a valuable thing)

@dalias@hachyderm.io It's part of the SportsWidget for the newtab/widgety homescreen - another thing I think best ideally handled by (out of tree) extensions.

@danmcd@hostux.social Recently: I think the inclusion of a dependence on Rust was premature, and yes not taking a strong stance against AI/LLM is I think a mistake but neither of those is a strong enough reason by themselves or even together. The development that concerns me more is the explicit rejection of social of political responsibility by core maintainers. To put it another way: a project like the Linux Kernel needs a strong vision, and I don't think they have one anymore.

I have many technical quibbles about recent linux kernel stuff, but I can at least configure or patch pretty much all of that out of anything I use. I'm even somewhat sympathetic to the argument that "statistical models are tools and we shouldn't outright reject a tool" i.e. I think there might be a future where we ethically construct and use large open models for some uses. But the explicit rejection of social of political responsibility by core maintainers cannot be overlooked.

@chrisvest@mastodon.social @dalias@hachyderm.io No joke this is something I am already concerned about.

If you find yourself looking at this absurd project and thinking "I want to believe" then I've thrown together a list of activities that might be good places to start: https://codeberg.org/basebrowserproject/organization/issues/18

@80columns@indieweb.social My position is that the newtab extension should probably just be removed (https://codeberg.org/basebrowserproject/organization/src/branch/main/0002-Remove-the-newtab-default-homepage.patch) (Some other notes here: https://codeberg.org/basebrowserproject/organization/issues/10)