10 KiB
FOSDEM 2025 Keynote
Handed off by bkuhn to Karen and Denver due to someone (very possibly Henry Poole) giving bkuhn COVID-19.
Title: Slide 0
Put it up as people are arriving so they know I'm not there. My name is striked out with apologies.
Kant: Slide 1
Kant slide, (First slide (after title slide). Please read this prepared statement from me:
Bradley is heartbroken to not speak today — he last keynoted six years ago, and was excited to keynote again. FOSDEM is a unique event in the world, as its organizers focus on community, not for-profit interests. Those who saw Bradley yesterday know that he work a tight fitting N-95 mask the entire day, as he did nearly the entire time since leaving his home for the trip to FOSDEM — he took his mask off only to eat, drink, and speak on stage. Nevertheless, he tested positive for COVID-19 this morning for the first time in his life. He actually feels reasonably good and since he has received absolutely every recommend vaccine, his symptoms are quite mild.
Bradley asks the entire FOSDEM community to reconsider their safety protocols. Bradley believes deeply in the moral imperatives of Kantian ethics: that, as this quote says more formally, humans have a moral imperative to voluntarily take every action so that it maximize the rights and welfare of all humanity. As such, he has quarantined today for our safety, and while he does not call for a return to mask mandates, he would like FOSDEM to return to “masks strongly recommended” policy indefinitely for the years to come, and for all of you to wear masks voluntarily, especially when in the audience of crowded DevRooms.
The science shows that masking works best in disease prevention if everyone participates. Bradley noticed anecdotally that less then approximately 2% of attendees wore masks yesterday and and at earlier fringe events. We should all commit to voluntarily to making that 98% because it's the right thing to do.
I do ask that you read the statement above exactly as written it. What continues is just what I would probably say. Once I hand this off to you (probably around 13:30 local on 2025-02-02), please feel free to turn it as much into “your talk” as you'd like. I just wrote all this out to help you understand where I was going to go.
Clemens: Slide 2
Does anyone, for a FOSS project they contribute to, use any of the following systems as a substantial part of their contributions on that project. Please keep your hands up after we say each one. Please don't be shy to raise your hand; we aren't judging you and we don't blame you for using these products we're about to list.
-
Microsoft Teams.
-
Zoom.
-
Sourceforge.
-
Jira.
-
Confluence.
-
BitBucket.
-
Any of the many non-FOSS continuous integration systems.
-
Slack — after hands go up, say: which, BTW, is now a Salesforce product.
-
GitHub — after hands go up, say: which, BTW, is now a Microsoft product.
-
gitlab.com's proprietary GitLab instance (i.e., the gitlab instance of your project not self hosted).
Of course, once we said GitHub, the most hands went up, but now at least N% of the room has their hands up.
Note that early on, we said Sourceforge. This is an interesting piece of history that most don't know: [ switch slides ]
VA Linux / Sourceforge Debacle: Slide 3
Sourceforge is a very interesting case. Most younger developers may not know that in the late 1990s, Sourceforge (and forge software in general) was a revolution in FOSS development. Until that time, there were no websites that provided integrated version control, bug tracking, developer discussion, and continuing integration. It was a patchwork of systems before that, and Sourceforge was extremely exiting to lifelong FOSS developers precisely because the need for better solutions was so great.
VA Linux initially was a good community actor: they released the entire codebase of Sourceforge under GPL, and many contributors began to work upstream on Sourceforge itself.
During the dot.com boom, VA Linux IPOed under the ticket symbol LNUX. Like airline scams of the 1920s, where companies named themselves with ticker symbols that sounded like airlines, many people thought that they were buying stock in this new operating system they were just hearing about, not one of many service companies in the space.
By late 2001, the dot.com boom was over, LNUX stock had tanked, and, as most FOSS companies do when times are tough, VA Linux ran to the oldest scam in the software industry: licensing all their software that they could as proprietary.
There's a link in the slides to an excellent article at the FSF Europe from October 2001 (written by Loïc Dachary), that describes VA Linux's behavior. As Loïc points out in his article, VA Linux did underhanded tactics to pressure developers to assign copyrights so that VA Linux could relicense Sourceforge wholly proprietary.
As a side note, this was one of the catalysts for the creation of the Affero GPL. In this case, since all the HTML and Javascript files were also GPL'd, VA Linux needed universal copyright assignment to proceed with a wholly proprietary product. Ultimately only a few developers like Loïc refused to assign copyright, but VA Linux as the overwhelming majority copyright holder simply wrote their changes out of the software, and relicensed.
We definitely encourage you to read Loïc's essay on FSF Europe's site, because he makes a truly excellent point: that the Free Software community could “Fork and ignore”: IOW, take the last GPL'd version that was released as a gift to the community, and proceed development from there — ignoring sourceforge entirely.
There was a somewhat golden period after that from 2001 until about mid-2004. Sourcforge, quite unsurprisingly, rather quickly switch to an ad-based system whereby they would show you an add if you wanted to download a tar.gz file of a project. Developers were truly stuck: SFC's own Inkscape project spent years well into the 2010s trying to fully divorce from Sourceforge, and ironically, the successor in interest, Sourceforce, Inc., realized Inkscape downloads were one of their largest downloaded projects, and constantly pitched us toxic revenue-sharing schemes until Inkscape finally escaped Sourceforge. [ next slide ]
Sourceforge Diaspora: Slide 4
The reason some projects had such trouble with getting away from Sourceforge was because there wasn't a primary fork, but dozens. Here's a list of the ones that Bradley could remember, but he recalls at the time there were at least four or five more that split the attention of the community so badly that, by the time AJAX and Web 2.0 came along, there was not enough leadership to move an old PHP+HTML application to the more interactive and modern looks that users were demanding by the late 2000s. [ switch slides ]
Preston-Warner: Slide 5
Enter GitHub, founded in October 2007. GitHub from its very founding learned the VA Linux lesson: don't ever give anyone code, and in fact, take great efforts to convince the FOSS community that copyleft, particularly the Affero GPL are terrible.
Enter GitHub, founded in October 2007. GitHub from its very founding learned the VA Linux lesson: don't ever give anyone any code: keep it all proprietary — and if you must release FOSS, do it in a way that doesn't allow people to make their own system.
This is an old slide of Bradley's, which he insisted we include even though he's shown a slide like this on the main stage at FOSDEM at least once before, simply to point out that convincing users of a fully GPL'd VCS (namely, Git) to switch to a forge that locked them into proprietary services.
While Preston-Warner was pushed out of GitHub due to an unrelated HR scandal, his vision, warped morality, and aggressive hatred of copyleft was baked into GitHub culture.
In fact, GitHub took great political and advocacy efforts to convince the FOSS community that copyleft (particularly the Affero GPL) are terrible license beyond this OSCON keynote of Preston-Warner's. We have seen cases, BTW, of GitHub employees, proudly sporting their “GitHub employee” achievement badge right next to their user icon, go into AGPL'd projects that they aren't even contributing to with rants about how the AGPL is a bad license.
We've also heard reliable intelligence (confirmed by multiple sources) that GitHub has an internal fork of Git itself. Now, the GPL of course allows internal forks, since its copyleft requirements (in most cases — there are a few exceptions) trigger only on distribution. Obviously convincing our own Git member project to switch to the AGPL is way beyond politically viable and we haven't even asked. But, this situation shows that if GitHub has a legally viable choice between liberating code on their own accord and proprietarizing it, they chose proprietarization every single time.
[ next slide ]
Microsoft Acquired GitHub: Slide 6
Ultimately, Microsoft was a perfect match for GitHub. [ Karen should say: ] I once spoke at Microsoft and asked them publicly apologize for calling the GPL a cancer back in 2002. The main feedback I got from high-ranking Microsoft employees was a “How Dare You Even Ask!?!?” kind of response.
Microsoft, and indeed most proprietary software companies, are not our friends. They don't want to help us make more FOSS (not copylefted, anyway). GitHub is much smarter than SourceForge. Instead of pushing advertising into FOSS (as SourceForge did and failed), Microsoft offers GitHub as a loss-leader product for FOSS developers, so that they are trained.
Bradley mentioned that he presented his capstone undergrad project at an ACM conference in 1995. Every single attendee was given a gratis copy of Windows 95. (Bradley confirms that this is the only Microsoft license of any kind that he's ever agreed to.) The point of Microsoft's methods are clear — going back decades: get people addicted to our proprietary stuff by offering it gratis at first, and then finding ways to sell add-ons.
While we don't like the term, Bradley calls this “free as in cocaine”.
[comment]: <> Denver: you can edit above all you want [comment]: <> bkuhn promises to not work above here [comment]: <> NO ONE BUT BKUHN EDIT BELOW YET:
[comment]: <> Next Slide.
Final Slides
The remain slides are very wordy. I did start to feel my symptoms works around 13:45 local, so while I'd sped up, I then slowed down. The slides should probably be edited, the text of what's there placed here, and the words on the slide should be less.