Werd I/O

I write a lot about the intersection of technology and society here, and lately a lot about AI, but over the last year I’ve written a little less about what I’ve been up to. So, this post is an update about some of that. This isn’t everything, by any means — 2023 was, frankly, a hard year for lots of reasons, which included not a small amount of personal loss and trauma — but I wanted to share some broad strokes.

We’re now based in the Greater Philadelphia area, rather than San Francisco. There have been all kinds of life changes: it’s the ‘burbs, which is weird, but I’m writing this on a train to New York City, which is now easily within reach. I grew up in Oxford and could easily go to London for a day trip; now I have the same relationship with NYC. We haven’t yet brought the baby to the city, but that’s coming. (He’s not a baby anymore: we have a delightful toddler whose favorite things, somehow, are reading books and brushing his teeth.)

I joined ProPublica as Senior Director of Technology after working with the team as an advisor on contract for a while. ProPublica publishes vital American journalism: you might remember the story about Supreme Court Justices with billionaire friends that broke last year, or the story about Peter Thiel’s $5 Billion tax-free IRA. You might also have come across Nonprofit Explorer and other “news apps”. Our technology philosophy is very compatible, and it’s a lovely team. I’m hoping we can revive The Nerd Blog.

I work mostly remotely and spend a lot of my time at my desk looking like this:

(Guess the books! Yes, that’s also an issue of .net — specifically, one from decades ago that showcased Elgg.)

My website is still powered by Known, and I still intend to invest time and resources into that platform. I’ve also finally accepted — between having a toddler, a demanding job, an ongoing project (more on that in a second), and other commitments — that I’m not going to be making a ton of contributions to the codebase myself anytime soon. But there’s a pot of money in the Open Collective, and I’m eager to support open source developers in adding functionality to the platform. The first stop has been adding ActivityPub support to make Known compatible with the fediverse. The next stop will be improving the import / export functionality so that it (1) functions as expected (2) is in line with other platforms.

I’ve been struggling with writing a book. I’ve had the benefit of really great 1:1 coaching through The Novelry, and was making great progress until I realized I needed to revise a major element. It’s been a slog since then: I have printouts of my first draft covered in Sharpie all over my office. My fear of being terrible at this increases with every sideways glance at the unfinished manuscript (which seems, somehow, to be staring back at me). I’m certain that as soon as I send it out into the world I’ll be ridiculed. But I’m determined to get it to the finish line, revise it, send it out, and do it again.

As painful as writing the draft has been, I also love the act of it. Writing has always been my first love, far before computers. Don’t get me wrong: I don’t claim any sort of literary excellence, in the same way that I enjoy making dinner for everyone but would never call myself a chef. I’ve got huge respect for anyone who’s gone down this road and actually succeeded (hi, Sarah, you are radically inspiring to me). It’s a craft that deserves care, attention, and practice, and stretching these muscles is as desperately uncomfortable as it is liberating. I find the whole process of it meditative and freeing, and also simultaneously like pulling every fingernail from my body.

So, uh, we’ll see if the end result is any good.

I’ve been helping a few different organizations with their work (pro bono): two non-profits that are getting off the ground, a startup, and a venture fund. Each of them is doing something really good, and I’m excited to see them emerge into the world.

Also, my universe has been rocked by this recipe for scrambled eggs. So there’s that, too.

What’s up with you?

404 Media reports that Automattic is planning to sell its data to Midjourney and OpenAI for training generative models:

The exact types of data from each platform going to each company are not spelled out in documentation we’ve reviewed, but internal communications reviewed by 404 Media make clear that deals between Automattic, the platforms’ parent company, and OpenAI and Midjourney are imminent.

Various arms of Automattic made subsequent clarifications. Specifically, it seems like premium versions of WordPress’s online platform, like the WordPress VIP service that powers sites for major newsrooms, will not sell user data to AI platforms.

This feels like a direct example of my point about how the relationship between platforms and users has been redefined. It appears that free versions of hosted Automattic platforms will sell user data by default, while premium versions will not.

Reddit announced a similar deal last week, and in total has made deals worth $203M for its content. WordPress powers over 40% of the web, which, given these numbers, could lead to a significant payday for the company. Much of that is on the self-hosted open source project rather than sites powered by Automattic, but that number gets fuzzier once you consider the Jetpack and Akismet plugins.

From a platform’s perspective it seems like AI companies might look like a godsend. They have an open license to tens or hundreds of millions of users’ content, often going back years — and suddenly, thanks to AI vendors’ need for legal, structured content to train on — the real market value of that content has shot up. It wouldn’t surprise me to see new social platforms emerge that have underlying data models designed specifically in order to sell to AI vendors. Finally, “selling data” is the business model it was always purported to be.

It’s probably no surprise that publishers are a little less keen, although there have been well-publicized deals with Axel Springer and the Associated Press. The deals OpenAI is offering to news companies for their content tend to top out at $5M each, for one thing. But social platforms don’t trade on the content themselves: they’re scalable businesses because they’re building conduits for other peoples’ posts. Their core value is the software and an enormous, engaged user-base. In contrast, publishers’ core value really is the articles, art, audio, images, and video they produce; the hard-reported journalism, the unscalable art, and the slow-burning communities that emerge around those things. Publishing doesn’t scale. The rights to that work should not be given away easily. The incentives between platforms and AI vendors are more or less aligned; the incentives between publishers and AI vendors are not.

I don’t think bloggers and social video producers should give those rights away easily either. They might not be publishing companies with large bodies of work, but the integrity of what they produce still matters.

For WordPress users, it’s kind of a bait and switch.

While writers may be using the free, hosted version of a publishing platform like WordPress, they retain the moral right of authorship:

As defined by the Berne Convention for the Protection of Literary and Artistic Works, an international agreement governing copyright law, moral rights are the rights “to claim authorship of the work and to object to any distortion, mutilation or other modification of, or other derogatory action in relation to, the said work, which would be prejudicial to his honor or reputation.”

The hosted version of WordPress contains this sentence about ownership in its TOS:

We don’t own your content, and you retain all ownership rights you have in the content you post to your website.

A reasonable person could therefore infer that their content would not be licensed for an AI vendor. And yet, that seems to be on the cards.

So now what?

If every platform is more and more likely to sell user data to AI platforms over time, the only way to object is to start to use self-hosted indieweb platforms.

But every public website can also be scraped directly by AI vendors, in some cases even if they use the Robots Exclusion Protocol that has been used for decades to prevent search engine bots from indexing unauthorized content. A large platform can sue for violation of content licenses, but individual publishers are unlikely to have the means — unless they gather together and form a collective organization that can fight on their behalf.

If every public website is more and more likely to be scraped by AI vendors over time, the only way to object is to thwart the scrapers. That can be done electronically, but that’s an arms race between open source platforms and well-funded AI vendors. Joining together and organizing collectively is perhaps more effective; organizing for regulations that can actually hold vendors to account would be more effective still.

It’s time for publishers, writers, artists, musicians, and everyone who publishes cultural work for a living (or for themselves) to start working together and pushing back. The rights of the indie website are every bit as important as the rights of organizations like the New York Times that do have the funds to sue. And really, truly, it’s time for legislators to take notice of the untrustworthy, exploitative actions of these vendors and their platform accomplices.

Hunter Walk writes:

The checks being cut to ‘owners’ of training data are creating a huge barrier to entry for challengers. If Google, OpenAI, and other large tech companies can establish a high enough cost, they implicitly prevent future competition. Not very Open.

It’s fair to say that I’ve been very critical of AI vendors and how training data has been gathered without much regard to the well-being of individual creators. But I also agree with Hunter in that establishing mandatory payments for training content creates a barrier to entry that benefits the incumbents. If you need to pay millions of dollars to create an AI model, you won’t disincentivize generative AI models overall, but you will create a situation where only people with millions of dollars can create an AI model. In this situation, the winners are likely Google and Microsoft (in the latter case, via OpenAI), with newcomers unable to break in.

To counteract this anticompetitive situation, Hunter previously suggested a safe harbor scheme:

AI Safe Harbor would also exempt all startups and researchers who have not released public base models yet and/or have fewer than, for example, 100,000 queries/prompts per day. Those folks are just plain ‘safe’ so long as they are acting in good faith.

I would add that they cannot be making revenue above a certain safe threshold, and that they cannot be operating a hosted service (or provide models that are used for a hosted service) with over 100,000 registered users. This way early-stage startups and researchers alike are protected while they experiment with their data.

After that cliff, I think AI model vendors could pay a fee to an ASCAP-like copyright organization that distributes revenue to organizations that have made their content available for training.

If you’re not familiar with ASCAP and BMI, here’s broadly how they work: when a musician joins as a member, the organization tracks when their music is used. That might be in live performances, on the radio, on television, and so on. Those users of the music — production companies, radio stations, etc — pay license fees to the organization, and the organization pays the musicians. The music users get the legal right to use the music, and the musicians get paid.

The model could apply rather directly to AI. Here, rather than one-off deals with the likes of the New York Times, vendors would pay the licensing organization, and all content creators would be compensated based on which material actually made it into a training corpus. The organization would provide tools to make it easy for AI vendors and content creators alike to provide content, report its use in AI models, and audit the composition of existing models.

I’d suggest that model owners could pay on a sliding scale that is dependent on both usage and total revenue. One component increases proportionally with the number of queries performed along a sliding scale at the model level; the other in pricing tiers associated with a vendor’s total gross revenue at the end-user level. So for example, if Microsoft used OpenAI to provide a feature in Bing, OpenAI would pay a fee based on the queries people actually made in Bing, and Microsoft would pay a fee based on its total corporate revenue. Research use would always be free for non-profits and accredited institutions, as long as it was for research or internal use only.

This model runs the risk of becoming a significant revenue stream for online community platforms, which tend to assert rights over the content that people publish to them. In this case, for example, rather than Facebook users receiving royalties for content published to Facebook that was used in an AI model, Facebook itself could take the funds. So there would need to be one more rule: even if a platform like Facebook asserts rights over the content that is published to it, it would need to demonstrate a best effort to return at least 60% of royalties to users whose work was used in AI training data.

Net result:

• Incumbents don’t enjoy a barrier to entry from copyright payments: new entrants can build with impunity.
• AI vendors and their users are indemnified from copyright claims against their models.
• AI vendors don’t have to make individual deals with publishers and content creators.
• Independent creators are financially incentivized to produce great creative and informational work — including individual creatives like artists and writers who might not otherwise have found a way to financially support their work.
• The model shifts from one where AI vendors scrape content with no regard to the rights of the creator to one where creators give explicit consent to be included.

The AI horse has left the stable. I don’t think shutting it all down is an option, however vocal critics like myself and others might be. What we’re left with, then, is questions about how to create a healthy ecosystem, how to properly compensate creators, and how to ensure that the rights of an author are respected. This, I think, is one way forward.