Posted on 2023-07-06

Read Time: 5 minutes | 1037 words

Modern Day Trusts and the Limits of Online Platforms

There is a not-so-secret reality behind most large tech companies: they are not so much companies as they are modern-day trusts. They have spent years swallowing competitors and corollary suppliers. Large and bloated, these organizations built their platforms on the debris of thousands of companies and then used their accumulated patent portfolios to force the rest into subservience. The technology they build and sell is only valuable to the end user insomuch that it keeps the user connected to the company’s ecosystem. If a user leaves to go elsewhere, the companies swallow that up as well. If something like GitHub gets popular, for example, Microsoft buys it. Adobe buys Figma. Google buys Writely and Looker and on and on1.

Microsoft, mostly because it got large in a time when governments still pretended to act in the interests of their citizens, has long been known for its use of this embrace, extend, and extinguish model. It certainly wasn’t alone. This is the model for modern tech, and startup culture adapted to this reality in rapid pace. After all, why extinguish the tech a startup creates when it is designed to be bought and owned by the trust? Why not simply fold it in to the larger portfolio?

And here we are, a collection of trusts that run the vast majority of what most people consider the Internet. Trusts that run roughshod over policies barely updated and policed by a government split between ignorant, theocratic fascists, and those just trying to hold some semblance of a meaningful government together because of those fascists. Even if policy change were a possibility, the stark truth is that the vast majority of those dictating that policy are not equipped to make educated decisions as every embarrassing congressional testimony with a tech bro has shown.

The “AI” of LLMs and their ilk, in this context, are just the cherry on top. They let trusts better use and silo content. Remember, to a trust it doesn’t matter who created something, but who owns and controls that something. If you are on their platform and creating content, you are part of the embrace (and we all are to some extent). Now, they are extending that content, merging it with others, creating targeted tools and profiles. With AI, they can simply regurgitate your content while merging it with others and then resell it as something new to keep that engagement going. As they do, they can better follow you, track you, make predictions about you, and learn how best to exploit you…er sell to you. It’s not just their platform, either. Who needs cookies or trackers when they can just siphon the entirety of the Internet (or have it indexed for them by decentralized ActivityPub and AT Protocol servers) and use that to create their hyper-targeted demographics and tools. They can bypass all the rules around cookies and privacy. After all, these are publicly accessible sites that people willing posted on.

I am not going to pretend there is an easy solution to this. I think there is a part of the ActivityPub user base that incorrectly believes it can post publicly and not be exposed to these dangers just by defederating from these larger organizations. I don’t really think that is a viable option for the long term. I think the change comes not in the federation but in the server. If you defederate because you want to focus on your server and your community, then more power to you. We need more of a local server focus and less of a massive distribution focus. To that end, we all need to reconsider what we make available publicly and what we keep within our communities and on our own systems (hardware and software)2. Frankly, a return to non-algorithmic curated content is not a bad thing, nor are barriers to access especially when those barriers protect people from abusive and exploitative systems and people.

While I advocate and personally practice much of the above. I don’t pretend to think the needle is going to move much in the broader sense. To really see changes at that level will require, however unlikely, government action. The best I can do is focus on my communities while I advocate and work for long-term policy changes at the government level. Of course, solutions at that level are further complicated by other existential problems–the aforementioned fascists and the accelerating impacts of climate change to name two. I do think that change is coming, though. In part, because it must. None of this is sustainable, and I think that many of those with wealth know that and are cashing in while they can.

The modern tech trust may be the dominant model and imaginary–look at how many people stayed on Twitter because everything else was “too flawed” of a platform–but it is not the only imaginary. I think, perhaps optimistically, that there is a limit to what people will tolerate3. I only hope that limit arrives before systemic collapse4.


  1. To be clear, this is the general garbage state of a corporatized world. I am just highlighting tech in this space. ↩︎

  2. I once argued that the worry over Microsoft’s purchase of GitHub was misplaced because GitHub was designed to be bought. While I did mention being careful about what you kept on GitHub, I didn’t envision the wholesale theft of work for AI models. That was not a post that aged well, as they say ↩︎

  3. I am actually quite bullish on LLM and ML models in these rebuilding contexts. I think that AI’s contextual use outside of closed systems (where there can be some amazing possibilities) is limited, at best, and more useful as a scaffolding tool for new learners and users than it will ever be as a general development or content creation tool. That type of use is critical in any sort of refashioning and learning that comes with a drastic rethinking of our current sociotechnical model. Leveraging tools to better make sense of a massive number of interconnected inputs to derive some good could be truly incredible with proper oversight and inclusive management. ↩︎

  4. It is amazing to me how true this is about so many of our current issues. ↩︎

Author: Geoffrey Gimse

Tags: #ai  #technology 

© Geoffrey Gimse (2023) - Built using Hugo.

Opinions expressed here are my own and are not neccessarily shared by employers, friends, or colleagues. Except where noted, all photos are my own. Other images used on this site are in the Public Domain or have been purchased for use via The Noun Project."