Issue 39: The Losers of the Open Source Movement
When the movement trends towards a cult
Food For Thought
Thanks to reading books like Napoleon by Andrew Roberts and Black Rednecks & White Liberals by Andrew Sowell, I've kind of been on a "nuance kick" lately. Basically, I've been interested in not just treating knowledge about popular, complicated, or hairy topics like Napoleon and his wars or the foundations of slavery and it's impact in today's race relations I get from common sense or social media thought a priori and instead use reasoning to tease out my own analysis of the subject at hand.
Last issue, I tackled the subject of surveillance, and drew up an argument that not all surveillance is bad, via negativa. This issue, I'd like to flip the script and discuss something that only gets "good" press – open source.
Being a part of the hacker/developer community at large requires one to have an understanding of what open source is and why it continues to sticks around. Because of the motivations and leadership of the open source community, the words that "open source" generally hangs out with are words like: non-profit, community, volunteering, free, sharing, equality, education. All of these words sound pleasant, but I'll make the point that they are explicitly chosen, these "word neighbors" are largely marketing. As an open source contributor and user, here are some other words that can be associated with open source: bullying, two party system, virtue signaling, free labor economics, sleight of hand, subprime markets.
As of late, we are in a societal moment where the failures of the web3/crypto movement, the power accretion of billion dollar training runs from giant GPU farms for LLMs and wealth inequality reaching all time highs in the market have people on edge about power imbalances. Over and over, new technologies come along that promise to "democratize" the (insert whatever new hype thing here) and make the world(*) more equitable. People "below" the water line of wealth are eager to tip the scales in their favor, people "above" it want to keep the systems in place that are netting them profit. Both of these desires are logically founded, and often it is not outright malice that make the rich richer and the poor poorer, but tragedy of the commons type market effects: better access to knowledge, more money to spend on riskier opportunities, connections, etc. that propel the market to reward those who have previously been rewarded.
So you might reasonably ask, why-oh-why would you come after something like open source? What about the open source movement could be bad? Isn't it doing the work to make the world more fair for developers, and for users of applications?
Bullying, the Two Party System, and Virtue Signaling
The first bone to pick with the open source movement is that leans heavily in favor of itself. In other words, the open source movement operates similarly to an ideal that can never be truly reached. The truest open source is a godhead, an unknowable and unreachable place. People will argue in forums for months about the minute differences between the Apache and MIT licenses, or nitpick the idea of attribution via forking or cloning, or drive a wedge in the leadership of communities because "they rely on the tool" or establish a "benevolent dictator for life" (Linus, Guido, etc.). This is because the term open source resists definition for it's own purposes, and is reactive to the technology of its era.
This is more true than ever in a world of LLMs, where the open source movement now is more than just a "weird thing" that the developer community does, it is "necessary" to guarantee the future of the world's safety. There is a lot to unpack there, but the short of it is that people want open source data sets from models and they want to know exactly how these models were built.
Since the people who are leading this charge of driving the open source movement forward have no financial or time/energy stake in these operations (skin in the game for any of you Taleb fans out there), they do what they can: they politically pressure (read: bully) developers and organizations to publicly share all of their research.
This bullying process largely takes place in the modern agora of Hacker News comment sections or Twitter feeds, where people will just attack organizations until they "get what they want". See for example, this tweet below, where one of the founders of Mistral replies to a user.
The way Far El writes here, and the way Arthur Mensch responds are both extremely telling. Far El's use of ellipses here is damn near criminal, and you can almost hear the venom-like sarcasm and judgement dripping off of his tongue. Here is a comparison analogy to drive the point, which any Southerners from the US will immediately recognize:
"Oh, you're just sweet as a peach cobbler. Bless your heart.............."
The point of Far El posting this on social media was to bully an organization he is not a part of into doing something that he wants by accruing social attention.
Arthur's response is no better. Here, he is responding, on the same platform, for positive valence advertising in the other direction. Truly, he may just be doing for Mistral what is on the masthead of their company website, but it could very well be...marketing.
In a way, both people here are virtue signaling their commitment to the god of "open". They are both "for the cause", but who is more "for the cause": the maintainer? Or the user?
These social media guerilla tactics and the unattainable idealism of open source slowly create a two party system, a binary. You are either open source == good, or closed source == bad. The gradations of what can be shared, or indeed asking what should be shared, wither away with each new release, each attack that further tribalizes closed source software vs open.
See why we need to apply some nuance?
Free Labor Economics, Sleight of Hand, Subprime Markets
This issue would be incomplete without talking about money. Money is indeed the underlying driver of technology as a whole, or rather, the cultural expectations of the continued development of technology. There are many, many, many routes I could address here, but I'll leave a list to study at the bottom if you are so interested(**). Instead, I'll address two that are close to my heart: working for free, and fake money in the water supply.
Working for Free
I love Obsidian. It is a great piece of technology. Scratch that, it is a phenomenal piece of technology. But what makes Obsidian truly great is not it's core functionality – it is the extensibility of the software. The community of Obsidian has done great work to turn Obsidian into a thriving ecosystem of calendars, Kanbans, queries that go across your entire vault, and more! However, while the user experience may be awesome due to the great developer community, the developer experience is not.
I've spoken to other popular plugin developers and behind closed doors, many of us feel the same way: we got the short end of the stick. The community just straight up won't pay for anything, even with a donation button available ont the site. GitHub stars don't put food on the table.
This is worse when issues start to flood in, when the community asks for features they want for you to do. In other words, free work. This causes the maintainer undue stress, as all of a sudden people rely on them to do work. This reliance causes the very natural state of human altruism to kick in, and so the maintainer will want to do this work. It's not true parasitism of course, and I would never go so far to claim that it is, but there are similarities in taking advantage of a process that is innate to get something else done. This is good locally, perhaps, but bad globally.
Creativity is quelled when developers have no sure promise that their work will be rewarded. And in fact this causes many of the open source repos we see "trend" on GitHub to be versions of "socially validated technology x but open source". Their marketing angle is, hey, you like this tech – what if it was open source? In the short term these projects collect small wins like upvotes/social validation by basically copycatting, but in the long term the developers of these projects lose because they will 1) have to maintain their new offering for free ad infinitum, and 2) they will always be playing catch up with the source code and feature offerings of the closed source code they copied to stay at parity. E.g. whenever Instagram changes its UI, all the open source copycat repos will need to update as well.
Fake Money in the Water Supply
Above, I showed a tweet from one of the founders of Mistral replying to a Twitter user. Did you know that Mistral raised over $400 million? Don't you think that this money serves as a role of a safety net for Mistral to play the long game of the social signal of the very noisy and hyper competitive AI space right now (Microsoft, OpenAI, google, your neighbor's dog, Stability, etc.) while working on their technology, or that its VCs (like A16z) won't eventually come knocking for some type of return on their money? Remember:
GitHub Stars don't put food on the table.
The economy of VC money and advertising effectively prop up the free part of open source. Developers don't need to charge for their software because they get money from deep patronage pockets or massive advertising auction networks (this is epidemic in the creator economy as well). This inflates the value of network effect software from companies like Meta who get to release models like Llama for free to get good marketing and to find developer talent to hire or stress test their models for them for free. Simultaneously this free money deflates the value of indie developers who try to fairly price their work and get told that the most anyone will pay is free thanks again to companies that can outcompete at the level of paying their devs.
Watch this talk below from Evan Czaplicki of Elm fame to learn more about the uniquely weird economics of software.
So, I ask you again, don't you think the idea of the virtues of open source require a bit of nuance?
On My Nightstand - What I'm Reading
In addition to marginal costs, software also requires ongoing maintenance to continue running successfully, regardless of how many people use it. Rather than a function of use, these costs are a function of entropy: the inevitable decay of systems over time. It’s not just code itself that requires maintenance either, but all the supporting knowledge that surrounds it. When code changes, its documentation must also change.
In the most general sense, it involves the steady flow of software developers who will leave organizations that regard them as cost centers and fungible commodities. Those organizations tend to believe you need two categories of people to implement software: business people who think, and grunts who, as James Grenning suggested, do low-level translation of natural language instructions into source code. And historically, we’ve proven them right to an extent. There are plenty of reasonably well-compensated programmers out there who content themselves with this golden coffin arrangement. But here’s the thing. That approach produces inferior software. The agile software movement suggested that we break down the barriers between business folks and IT folks so they can work more effectively together. I say we reject that premise in its entirety and go forward believing that business folks and IT folks should be the same people. This is, of course, a disconcerting proposition for the business folks. Managers and former developers will need to come face-to-face with an uncomfortable question, and that’s “What do you need me for, then?” My honest answer to that is, “I don’t know. You’ll probably figure something out.”
While most open-source developers do not intrinsically object to others profiting from their gifts, most also demand that no party (with the possible exception of the originator of a piece of code) be in a privileged position to extract profits.
There is good reason to believe that the financial foundations of the web are perhaps shakier than we think, maybe even producing the conditions for a “subprime” crisis in attention, similar to the dynamics that brought down the global economy in 2008. Examining these parallels is a key step toward thinking about how society, if it’s not too late, might want to re-architect the web for the better, and about whether the web as we understand it will endure.
-- Subprime Attention Crisis: Advertising and the Time Bomb at the Heart of the Internet (FSG Originals x Logic)
The software that matters most is the most hidden, the least revealed. Guess what? BUMMER software usually runs on a foundation of free and open software (like the Linux/Apache stack). But no one can know what is done on top of that free and open foundation. The open-software movement failed absolutely in the quest to foster openness and transparency in the code that now runs our lives.
Habitual drivers, driving habits/
a day at a time, each day to do one/
the gluttony of moderation, the immoderation of scarcity/
a habit not started is not one undone
* though it's never really stated if it's the world that the creators envision, or the one they live in in reality
** List: the drama of open source organizations like npm or React, the blow up moment of leftpad, the anti-trust cases against Microsoft, the "I just want to know I can get my hands on it, someone else will verify if it's safe" security holes of many pieces of open source software
Thanks for reading, and see you next Sunday!
ars longa, vita brevis,
P.S. If you like what you read on this newsletter, forward it to a friend! It really helps!