I found some notes I jotted last year after a now-deleted exchange on ex-birdsite, with Rachel Coldicutt, where we were discussing the UK OSA (as usual).
She had said:
“The Online Safety Act will take several years to become effectual because its focus is at the wrong end of the firehose. The focus is on post hoc proof of compliance rather than effectually shifting the business models and incentive structures.”
To which I responded:
“The focus is at the end which was the most effectively captured by lobbyists and vendors.”
And (somewhat later in the discussion) she replied:
So, as Heather says, what we have in the OSA is a process that suits big tech compliance departments and is by and large dependent on imaginary technology to enforce.
I want to come back to that: imaginary technology.
I have already expressed everything I could possibly say about how this particular piece of legislation wasn’t just drafted around the needs of compliance software vendors, as opposed to, you know, children. It was drafted to encourage the invention of the technology needed to do the things which the law legislated for. Put another way, it was a speculative call to invent imaginary technology based on the belief, also known as magical thinking, that this technology could exist, and immediately solve All The Problems.
And that, as I said, we could invent that technology and solve All The Problems if we all just nerded harder whilst humming “Jerusalem”.
I bring this up today because I think we need a term for that.
I’m going to call it speculative legislation.
Speculative Legislation is law drafted on the presumption that some form of technology will be invented to do the thing the law is enacted to do. So it sets forth compliance obligations which service providers in scope must achieve, which are so bureaucratically onerous and technically complex that they can only meet their obligations by entering into service contracts with the technology vendors who are expected to invent those future solutions.
In other words, speculative legislation is a strategy for financing imaginary technology to solve real problems.
There may be a better term for it.
There may also be a worse term for it, but it will be taken seriously because an academic puts it in an academic paper, which makes it officialy cite-able, because that’s how these things work.
After all, bad ideas which were taken seriously because they appeared in academic papers, unlike the criticism of those ideas which didn’t, was how we ended up with the OSA in the first place.
But that’s another post for another day.
This is quite closely related to a lot of the machine learning legislation: ‘yes, you are allowed to simply wholesale steal everyone’s IP because when the magic machine you are building based on this theft comes into existence it will, by some entirely unexplained mechanism, be able to, for instance, solve climate change’.
Of course, not only will the AI not actually come into existence but if it did it probably will say ‘yeah, you probably should have listened to the human scientists, because they were right’.
As with the OSA it is I am sure nothing to do with a bunch of people extracting large amounts of money from the system.
Yes! Thank you for understanding my poorly articulated lunchtime musing.