My Keynote at SeaGL and LLM-gen-AI-assisted Reverse Engineering

by Denver Gingerich on November 5, 2026

Over the past few months, there have been an increasing number of reports of people successfully gaining access to their own devices that were designed to prevent them from installing their own updates and modifications. Nearly every one of these reports notes that an LLM-gen-AI was used. While there are obviously a huge number of issues with LLM-gen-AIs (as we note in our LLM-gen-AI recommendations), it is hard to ignore the benefits to people's digital autonomy that reverse engineering like this brings.

As an example, one device that was previously difficult to install updates on was recently freed by OpenWrt (part of SFC). Specifically, the Zyxel XS1930 (a 10- or 12-port network switch, with dual-core 1.0 GHz MIPS CPU), which ships with a fully proprietary OS called ZyNOS, has been upgraded to OpenWrt using the ZyNOS web UI, which only recently became feasible, thanks to the efforts of developers using LLM-gen-AI. Previous efforts to install OpenWrt or other Linux-based systems on the XS1930 required hardware modifications, but now people can update to OpenWrt in a couple of minutes, ensuring they will be able to maintain the security of their device long after Zyxel stops supporting it, not to mention all the new (OpenWrt) features they can now enable if they so choose.

There are many other recent examples of such successful reverse engineering efforts. I covered several of these in the keynote at SeaGL a couple weeks ago (including a guitar amp, the Meta Quest bootloader, a variety of music gear, and the Nintendo Switch 2 - see Slide 16 of the slides for details). Since then there have been even more, including this article on the sub-$100 FM-1, which was covered more thoroughly by an electronic music news site just last week.

Some journalists, recognizing how many of these projects are developed, have been raising the alarm about where this will go if we're not careful. Jason Koebler of 404 Media notes that Inevitably, poorly coded, barely-functional, insecure software is going to be installed on critical hardware, which will lead to devices and networks being hacked. He's right, and this is why we need FOSS communities more than ever.

In the example of the Zyxel XS1930, freed by OpenWrt, LLM-gen-AI was used only to the extent necessary to understand the binary software image format the stock firmware expects. Beyond that, the device runs all the upstream projects distributed by OpenWrt, which have been carefully reviewed and developed by humans over the past several decades. This contrasts with many of the other reverse engineering projects mentioned above, where the software that replaced the proprietary code that previously ran on the device was itself generated by LLM-gen-AI with little or no human review. Especially with the LLM-gen-AI available today, this can be very dangerous, as Koebler correctly noted.

We at SFC, with our long history of finding ways for users to take control of their devices (using a variety of mechanisms, like lawsuits when necessary, our ongoing work toward software freedom on 3-D Printers also building our own community-focused hardware), have reached out to these reverse engineering projects to see if we at SFC can help, and to make connections with our member projects, where it makes sense and does not increase the risk of project burnout.

A related endeavour that I've personally been pushing since my first year at SFC over 15 years ago is to encourage companies to intentionally make their devices open to software modifications. While there have been hints about big companies doing this recently, which people have good reason to be skeptical about, there are smaller examples like several makers of electronic music gear mentioned earlier, that explicitly encourage people to run their own firmwares on the equipment they sell, when they recognized this would be a huge boost to sales and reputation with the community. If you are part of a company or know a company who wants to promote software modifications on the devices the company sells, please reach out.

Beyond the efforts described above, SFC will continue to look for new ways to engage with people who are working to bring the best of FOSS, including to the new world of increasingly modifiable devices that is now, whether Big Tech likes it or not, truly upon us.

[permalink]

Tags: conservancy, resources

My Introductory Comments for Mark Wielaard's 2026 Distinguished Service Award in Software Freedom

by Bradley M. Kühn on August 12, 2026

I had the honor of giving the introduction for the 2026 Distinguished Service Award in Software Freedom. The video of ceremony will be available in the coming weeks, but I share below that introduction now. Congratulations to Mark on this well-deserved award.

I'm up here now to announce the second annual winner of the Distinguished Service Award for Software Freedom.

In preparation, I'll tell you the story of one of the dozens of incidents in my FOSS career that convinced me that FOSS events like this one are essential to advance communities that spend 99% of its time collaborating over the Internet.

16 years ago — almost to the day — I stood in line to register for GUADEC 2010 in The Hague in the Netherlands. This was my first GNOME conference,and frankly as a command-line geek who only uses a GUI to make more xterms and Emacs frames on a screen, I was feeling nervous about attending.

Behind me in registration line, I heard exclamations from a young Dutch man with rosy cheeks and huge smile. He couldn't contain his excitement in blurting out: “You are … You must be … Bradley Kuhn!!!!”

I turned to face him fully and said: “wait, you, you are mjw, aren't you? You are the guy who wrote Classpath!”

Mark answered with what I'd soon learn was his true humility: “Well, you know, I didn't write Classpath, I mean, YOU are the one who named it Classpath … and well, there was already a lot of code when I took over.”

I already knew Mark Wielaard to be a dedicated, kind, and humble FOSS developer from our email exchanges in the early 2000s. But spending time with him at GUADEC 2010 — really getting to know the whole person who is “mjw” — was one of the best experiences of my career.

I boarded the flight back home thinking: “What a joy it is to have such people in the FOSS community! What great work we will all do together if all of us are like mjw!”

Mark didn't rest on his laurels after his Classpath work. He worked on the GCC front-end for Java. Then, when Sun (and later Oracle) released some of their Java code, he lead the IcedTea project that integrated the Sun OpenJDK code-dumps with existing Classpath work.

Today, he continues work in FOSS by maintaining the low-level key infrastructural project elfutils, and the Valgrind dynamic software analysis tool. Even with all that funded work, he still finds the time to serve the community as an essential volunteer maintaining the oldest and most prestigious Free Software code-hosting site, Sourceware.org.

For me, the only sad part about this moment for is that there is only one Mark, not a thousand. No, I am not trying cloning experiments in my basement to make more Marks! But I had hoped that day — 16 years ago on that plane — that the FOSS community would be filled with people like Mark. I have come to realize that his optimism, dedication, focus, and care are EXCEEDINGLY rare traits in our community. We should all aspire to approach FOSS in the manner that Mark does, and I hope all of you will take time at this event to learn from Mark how to improve as a FOSS contributor! He has a wealth of knowledge and wisdom to share — and he will share it with us willingly with kindness.

It's for that reason that I nominated — and easily received wholehearted, universal agreement from everyone at Software Freedom Conservancy — that Mark receive this second annual Distinguished Service award.

In addition to this lucite award, this honor includes a ten billion binary US Dollar cash prize. (For those that prefer decimal, that's USD $1,024. Mark's contributions are so significant that I deeply wish we could wave a magic wand and dub him a DECIMAL billionaire, because people as wonderful as Mark really should rule the world.

Mark, please come up on stage!

I am proud and honored to present you with this Second Annual FOSS Distinguished Service Award.

[permalink]

GitHub & other LLM-gen-AI Companies Urge You to Misdirect California Legislature on FOSS Licensing!

by Bradley M. Kühn on July 3, 2026

Another Reason to Give Up GitHub!

Last week, Microsoft's GitHub¹ announced they'd joined a coalition of other LLM-gen-AI for-profit entities who oppose updates to the California Artificial Intelligence Transparency Act (“Cal. Bus. & Prof. Code § 22757”). In their statement, GitHub mistakenly claims that the licensing termination requirements of §22757 run contrary to Free and Open Source Software (“FOSS”) licensing principles. This article explains why these statements made on behalf of Microsoft & GitHub incorrectly represents how FOSS licenses work, and why the entire point is moot since the LLM-gen-AI systems in question are not FOSS themselves anyway!

California's Policy Goals

Cal. Bus. & Prof. Code §22757 (and its updates currently under debate in the current session's SB 1000) are — like most legislation we see around new technologies in USA State and Commonwealth legislatures — well meaning and seeks reasonable goals, but remains confused about some details that are obvious to those who deeply understand the technology. Regardless, the law's aim is good policy. Definitely read through its interesting terms found in the amended §§22757.{1,2,3(a)} proposed in SB 1000. I suspect anyone who uses a for-profit, proprietary LLM-gen-AIs (perhaps by choice, or perhaps under mandate from their employer) would strongly prefer LLM-gen-AI vendors to provide the tools and information that §22757 mandates.

These policy goals — mandating transparency and allowing users to “trust but verify” these systems — are precisely the requirements that we want deployed widely for any LLM-gen-AI. SFC's own recommendations on LLM-gen-AI, in fact, correlate with and encourage “after market” implementation of some of §22757's requirements. While this law is of course not written the way technology policy wonks would likely write it, on balance, the law is a good one.

Where GitHub (et al.) Hung their Opposition

Of course, GitHub, Mozilla, Hugging Face, Black Forest Labs, (and likely other companies) hate this law. It requires them to do work to treat their users better. No one in the business of proprietary technology wants to do any more of that than is absolutely necessary to keep the customer. After all, it cuts into profits if companies do anything nice for their users that the customers have not directly demanded through a contractual requirement.

These for-profit corporations use misdirection and disinformation to convince the public that this law is bad for FOSS. Cal. Bus. & Prof. Code §22757 (both as on-the-books now, and as proposed for amendment) is not bad for FOSS, and, is generally good for the software right to repair.

Complaints regarding SB 1000 from Big Tech (and their cronies) focus on a narrow point in the amendments (found in §22757.3(b)(1-3)). Herein, I analyze this clause, and refute these companies' falsehoods about how it impacts FOSS. The key portion found in the proposed-amended §22757.3(b)(1-3) reads:

[22757.3](b)

1. If a covered provider licenses its GenAI system to a third party, the covered provider shall require both of the following as terms of the license:
   1. That the system remains in compliance with this chapter, to the extent it is technically feasible.
   2. That the covered provider may revoke, suspend, or terminate the licensee’s authorization to use the GenAI system if the licensee modifies the GenAI system such that it no longer complies with this chapter.
   1. If a covered provider knows that an identifiable third-party licensee modified a licensed GenAI system such that it no longer complies with this chapter, the covered provider shall terminate the licensee’s authorization to use the GenAI system within 72 hours of discovering the licensee’s action.
   2. A third-party licensee shall cease using or making available a licensed GenAI system, including a copy or modified version of the GenAI system, after the licensee’s authorization to use the GenAI system has been terminated by the covered provider pursuant to paragraph (2).
   3. This subdivision does not require a covered provider to monitor, investigate, or otherwise inquire into a third-party licensee’s use or modification of a licensed GenAI system.

On the surface, their argument that claims that text contradicts the irrevocability of FOSS licenses has truthiness. Nevertheless, their argument is sophistry. Here's why:

These LLM-Gen-AI System Are Not FOSS

SFC called our statement seeking a FOSS-friendly LLM-gen-AI system aspirational precisely because none of the current publicly deployed LLM-gen-AIs in wide use are anywhere near FOSS. The portions installed on the users' computers are of course, proprietary. Those on-system UIs are thin layers that access a service (via an API) — which lives on some server running trade-secret software. Unsurprisingly, GitHub's statement gives not a single example of a specific LLM-gen-AI whose distribution is already thwarted by the existing Cal. Bus. & Prof. Code §22757, nor one that was not thwarted by the existing law but would be thwarted if the amendments in SB 1000 are adopted. They didn't name one because there isn't one!

Non-Copyleft licenses Allow Additional Terms

While non-copyleft licenses such as the MIT and 3-Clause-BSD licenses are indeed irrevocable, they also permit redistributors of the software to impose additional terms that are not accounted for in the upstream license text. Therefore, even if Microsoft's GitHub were to release all of Copilot (model, server-side code, on-system UI) under the MIT license, they could do so with an additional term that complied with either formulation of Cal. Bus. & Prof. Code § 22757.

Copyleft Licenses Do Not Outright Conflict

I highly doubt that Microsoft, GitHub or the other entities (who deploy these systems in California) would ever willingly release one of their LLM-gen-AIs under a copyleft license. However, even if they did, the GPL Agreements already account for such situations.

First, nothing in SB 1000's amended § 2757.{1,2,3(a)}) conflicts or contradicts anything in any version of the GPL Agreements. Any entity that might someday release a GPL'd LLM-gen-AI can both simultaneously comply with GPL's requirements and meet all the requirements in §2757.{1,2,3(a)}). Nothing in GPL prohibits a redistributor from doing extra nice things for their customers and users beyond copyleft — as long as they don't directly contradict a requirement already in the GPL.

SB 1000 2757.3(b) requires some complex analysis, but causes no problem. All versions of the GPL Agreements had to consider that software patent licensing might impose additional restrictions outside of the GPL. Specifically, the GPL has no real mechanism to force a third party² who has never copied, modified, distributed, installed, and/or deployed the GPL'd software to issue a GPL-compatible patent license. All versions of the GPL Agreements therefore have a catch-all clause to deal with situations where external conditions make it impossible to comply with the license. Here's that clause from GPLv3:

12. … If conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot convey a covered work so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not convey it at all. For example, if you agree to terms that obligate you to collect a royalty for further conveying from those to whom you convey the Program, the only way you could satisfy both those terms and this License would be to refrain entirely from conveying the Program.

This clause was specifically designed to catch situations like we have with Cal. Bus. & Prof. Code § 2757.3(c) (and SB 1000 §2757.3(b)'s amendments). Imagine that this series of exceedingly unlikely events come to pass. Some Big Tech company:

1. uses a copyleft license for an LLM-gen-AI system,
2. deploys that system in a way that is accessible to the general public of California,
3. has licensed that system to a third-party under the same copyleft license, and
4. that company further fails to comply with 2757.{1,2,3(a}) with regard to the publicly accessible instance of that system.

If all that were to happen (unlikely), there is at that moment a condition imposed on this company that contradict[s] the conditions of [the GPLv3]. This company now cannot convey [the] covered work so as to satisfy simultaneously [their] objections under [GPLv3] and … other pertinent obligations imposed by §2757.3. The GPL Agreements planned for this possibility: in the end, GPLv3 requires everyone in the distribution chain from the original company down to all its licensees to stop conveying the work. (NB: GPLv2§7 is nearly identical to GPLv3§12.)

Furthermore, note that the GPL Agreements terminate on their own when any entity in the distribution chain violates any term — including GPLv3§12 / GPLv2§7. So, yes, GPL Agreements are irrevocable, but only as long as everyone remains in compliance with the license. Termination for reasons of non-compliance is fully accounted for, and thus the SB 1000 changes to Cal Bus. & Prof. Code § 22757.3, despite GitHub's misdirection, actually make the law more copyleft-compatible by changing the term ”revoke” to ”terminate“ — which tracks GPL's language exactly!

Copyleft Can Easily Be Adjusted If a Court Finds the above Analysis Incorrect

Even if the analysis above fails, and a Court rules that there is an irreconcilable incompatibility in California between the GPL Agreements and Cal. Bus. & Prof. Code § 22757 (as it currently stands, or as amended by SB 1000), a simple addition to GPLv3§7 can account for this. That section, entitled Additional Terms is designed to explicitly incorporate pro-user policy terms that might not be explicitly available in GPL Agreements. Regardless, there is no need to rush to update the GPL Agreements on this front, as the above analysis likely holds.

Contact the CA Legislature Now

Below we include our own template letter that we urge you to download, modify into your own words, and send it to Senator Becker. It's particularly important to do this if you live or work in California, and be sure to Cc your own senator as well.

The letter is available in LaTeX, Markdown, and PDF.

---

¹ In June 2018, Microsoft acquired Github for US$7.5 billion — at which time Github became a wholly owned subsidary (effectively, just a separate division within Microsoft). Many Github users did not realize that even before acquisition, Github was on a sustained anti-copyleft campaign, and the merger solidified that work with Microsoft's 20 year sustained anti-copyleft campaign. Thus, the announcement last week — wherein Microsoft's Github appears to be standing up for the irrevocable nature of copyleft (given that non-copyleft always allowed further restrictions that took away the irrevocability anyway) — is even more disingenuous.

² Various terms of the GPL Agreements can bind patent holders in various ways to keep the code safe from patent infringement claims for downstream users when the patent holder has actually engaged with the software in some way. However, it is common (particularly in a non-practicing-entity situation) that the patent holder demanding licensing fees has never even made a single copy of the copylefted software in question, much less installed, distributed, and/or deployed it.

[permalink]

Tags: conservancy, GPL, law, Git

Ethical and Moral Considerations in Proprietary Software Usage

by Bradley M. Kühn on June 2, 2026

In this philosophical essay, I explore the question: “When (if at all) is it ethically and morally acceptable to use proprietary software in the production and/or improvement of urgently needed copylefted FOSS?”

The question presents a complex conundrum. I attempt herein to rigoriously examine it through both a priori ethical analysis and a posteriori (and folksy) consideration of my personal experience and the shared experiences of the early software freedom movement.

I surprised myself at the outcome of my analysis. I conclude that under some circumstances (of which we have already witnessed in key historical examples), use of proprietary software by FOSS contributors to create/improve FOSS becomes a moral imperative. And, that imperitive often supersedes the moral imperative to avoid using that proprietary software.

A Parable of Competing Moral Imperatives

I grew up lower middle class in the USA — which is quite privileged by global standards. As such, I never went hungry, but most meals did not include second helpings. My family were early adopters of “extreme couponing”. My earliest childhood memories are climbing through (on Monday mornings) the gigantic bin of recycled Sunday newspapers behind the grocery store. My job in this endeavor was “Sunday insert extraction”. Back then, more than half of all households received the Sunday paper. That Sunday insert was the goldmine. The insert paid for the paper subscription (and much more) through its colorful 30-40-page advertisements filled with coupons. Get your hands on 10–20 more of those inserts freely from the recycle bin, and you could get “Extreeeme!”.

The late 1970s and early 1980s were the wild west of extreme couponing. There were nearly no restrictions on how many identical coupons one shopper could use. A shopper could use 50¢-off coupons for 49¢ trial-sized products. I remember once my mother filled the entire checkout belt with 30+ trial size dishwashing liquids — for which my mother paid ≈65¢ (i.e., just the sales tax).

My young indoctrination to extreme couponing now feels as if it’s part of my DNA. If you’re ever on on mute during a voice call with me and I don’t reply quickly, it might be because I’m currently standing in a grocery store aisle — comparing value per volume of different sizes and cross-referencing that with a coupon’s fine print.

I actually miss having my hands covered in newspaper ink every Monday — as that analog way was superior to digital. For much of my life, I enjoyed “extreme couponing” as one of the few activities that put aside my worries about the future of users’ software freedom, rights, and privacy. Sadly, about ten years ago, everything started to change. Now, using any coupon is a constant battle with proprietary, bait-and-switch spyware.

Major grocery chains in the USA provide proprietary apps — now the primary place providing coupons. Most chains have parallel websites that allow account holders to “clip” them digitally. Printed ads still exist, but are rife with phrases like “Digital Deal only”. Shoppers now collide in the grocery aisles while navigating the Kafkaesque apps and websites to figure out how to (virtually) clip a good coupon.

The once relaxing process of watching television on a Sunday afternoon while clipping paper coupons slowly morphed into a two-hour ordeal of figuring out the minimum proprietary Javascript required, then building a text cross-list noting the discounts, screenshotting the bigger discounts, and then using the list and screenshots to prove to the grocery manager that I really was supposed to get $1-off of 4+ avocados this week. And, yes, I clipped it in the web browser. And, no, I can’t install the app on this device to “check it in the app”. 🙄

Most of my life, I couponed due to absolute financial necessity. I can fortunately now afford to pay full price, but frugality is part of my moral code. While the correct moral action for these grocery chains is to simply reduce prices for all shoppers automatically, they behave unethically. They prefer to use proprietary software to track you, to attempt to manipulate you into buying items you don’t need, and all the other obvious anti-features. In response to their unethical affront to the public, I feel the moral imperative to exploit that system by not falling for the manipulation tactics, maximizing my grocery budget, and acting as a watchdog when the digital coupons just seem to “not work”.

Is This Essay About FOSS, or?

My colleague Karen Sandler and I have keynoted three times regarding our own ethical and moral analysis of software freedom in today’s digitized world. Our primary thesis: one must constantly compromise some software freedom purity to merely participate as a normal citizen. My lifelong grocery shopping experience exemplifies one situation of hundreds where tasks that once included no software interaction now mandate proprietary software usage. I lament this reality; it annoys me. But annoyance isn’t immorality, and software freedom activism was founded on pragmatic idealism. I fear that recent hard-core FOSS rhetoric has forgotten what that means.

As a fan of Kantian moral frameworks, my formulation of pragmatic idealism focuses on ranking the competing moral imperatives. While “use no proprietary software myself” is a noble moral tenet, its value remains paltry when viewed comparatively.
The moral imperative to avoid writing proprietary software towers over the moral imperative to avoid using it.

The conclusion is similar in a Utilitarian ethical analysis: provided one does not encourage others to join them, using proprietary software harms only the self. Writing proprietary software harms everyone who ever uses it — perhaps into the distant future.

The reverse situation yields similar analysis. Often, an individual, by using proprietary software today, can assure that many others will ultimately use less proprietary software in the future. In such situations, use of proprietary software by the individual is clearly a higher-ranked moral imperative (because it prevents many others from facing similar moral dilemmas). This also serves the greatest good for the greatest number: one person uses proprietary software so that many others won’t.

SFC lives this truth every day we operate. As part of SFC’s fiscal sponsorship of our member projects, we use a lot of annoying proprietary and/or trade-secret software (including banking systems, corporate donation and purchase order systems, third-party donation services, etc.). Yet, if SFC didn’t use those systems on our member projects’ behalf, at least one individual (possibly many) from each project would use them, instead.

Application of pragmatic idealism shows that using proprietary software is occasionally the right ethical choice for at least three reasons: (a) serving a higher moral imperative outside the scope of FOSS, (b) limiting the number of people who use proprietary software (now or in the future), and (c) increasing the efficiency and/or efficacy of urgently needed FOSS advancement.

Historical Context of These Competing Moral Imperatives

While my opening example of using proprietary software to increase efficiency for FOSS’ advancement focused on the non-FOSS imperative of frugality, the principle expands well beyond that. In fact, the earliest FOSS contributors clearly believed that other imperatives sometimes justify proprietary software use.

By the time Richard M. Stallman (RMS)¹ began to formulate software freedom as an essential right in the mid-1980s, proprietary software was the norm. Most computer users were using a Microsoft, Apple, or proprietary Unix system. RMS chose to write a copylefted Unix-like system (GNU) to replace proprietary Unix on its existing hardware.

But consider how RMS and the other early GNU contributors approached that work. If the goal was to minimize the developers’ use of proprietary software, the obvious approach would have been as follows: write a small bootable kernel on the existing ITS computers for the new hardware. Then, one could painstakingly write the basics of a C library in assembler (because, no FOSS cross-compilers existed in those days). After probably a decade of fulltime work by 3–5 people, there might have then been enough of a C library, a C compiler, a shell, and a text editor that others might also have software freedom.

I’m quite glad that the early GNU contributors did not take this approach. In fact, RMS started by writing the key program he knew best how to write quickly: another Emacs implementation — designed for existing proprietary Unix systems. And, GNU Emacs was working nicely in less than one year!² Next, the GNU contributors tackled GCC — but not by bootstrapping in assembler. Instead, they used existing, proprietary C compilers to bootstrap. That too brought software freedom to C programming much more quickly than the “pure” approach.

Nearly all readers already know how the story ends: these early contributors quickly brought most of a working POSIX system to users — all licensed under copyleft licenses — except for the kernel. That story is often told, yet so rarely is the most important point the focus: the entire FOSS community used part-FOSS/part-proprietary systems for at least 15 years. Our (correct) focus was to deliver as much software freedom to other users as we could — as quickly as possible. Our overarching moral imperative was “create as much copylefted software as possible” — even if we used proprietary software as part of FOSS’ production.

The users’ software freedom matters much more than the developers’ software freedom. If FOSS developers find a path that accelerates software freedom for their users, then that higher moral imperative demands that path.

Using the Tools of the Oppressor Against the Oppressor

We dream of a world without proprietary software — where every byte of code on every piece of hardware provides complete, Corresponding Source to its users. Despite our best efforts, we drift further from that world every day. We race against the wealthy people whose wealth grows faster when they produce more proprietary software. We will not beat them in our lifetimes, but we can make progress. Along the way, we must constantly reconsider: Should we use some proprietary software to accelerate urgent improvements to FOSS? Willingness to sometimes answer that with “yes” (pursuant to rigorious criteria) epitomizes pragmatic idealism!

Every day, I find that I’m using more proprietary software than I did in the early 2000s. I’ve been on that unfortunate trajectory since the mid-2010s. I was one of the last hold-outs who using only X200 and T500 laptops — the last laptops ever made that could run all FOSS code from the up-top to the down-deep. Nevertheless, in November 2025, I switched to a Novacustom V54 that requires so many blobs for the internal devices that 77.5% of the packages contained in the firmware distribution are proprietary.

Today, my personal software freedom on my own laptop (at boot-time) is measurably 77.5% less than it was when I lugged around the T500. However, everything I do for SFC now happens quicker. I spend substantially less time waiting for beancount to load SFC’s books. I can attach three additional external displays (instead of just one). I stopped injuring myself every time I travel from the T500’s weight (as my X200 became too slow many years before). These and so many other efficiency improvements from my Novacustom laptop made my work at SFC (at least) 50% more efficient.

I gladly take that trade-off all day long. My job is to preserve, protect, and promote users’ software right to repair. That moral imperative outranks my moral squeamishness every time the blob on my wireless card processes a network packet.

Frankly, the software freedom and rights of computer users of the future deserve priority over activists’ and developers’ preference to use only FOSS. We should reluctantly but doggedly use any proprietary software that assuredly accelerates creation of and improvements to essential and urgently-needed copylefted FOSS.

---

Footnotes

¹ RMS deserves substantial credit for formulating the first ethical framework for software freedom, his invention of copyleft, and his authorship of the earliest GNU programs. I am however not comfortable mentioning RMS and his work without noting his bad behavior that I frequently witnessed and his ongoing refusals to curtail or apologize for that behavior.

² Free Software, Free Society: Selected Essays of Richard M. Stallman, 3ʳᵈ edition (2015). GNU Press, Boston, MA. Page 21.

[permalink]

Tags: conservancy, GPL