What transaction costs does the new technology diminish? What transaction cost does it impose? A gun, for example, radically diminishes the transaction costs for ending life almost effortlessly at a distance. A gun can’t force you to kill anybody. But it’s going to be predominantly used in ways that capitalize on its affordances.
Selinger’s questions around affordances and impositions remind me of Malcolm McLuhan’s tetrad.
One of the reasons that leads some technology to become problematic is because things become too easy. For example, the internet makes trolling easier.
one of the dominant reasons there’s so much trolling and toxic material circulating is because the transaction costs of being able to communicate have gone down precipitously. With the Internet, unlike in the past, I don’t have to write my strongly-worded letter that I slowly proofread and make sure my penmanship isn’t embarrassing, or write out the person’s address on the envelope, get a stamp, and bring it to the post office. There’s enough speed bumps that it slows you down. If you’re going to write that letter, you must really be angry and pissed off.
Another example are the transaction costs that make AI and facial recognition too easy.
Facial recognition technology has a very distinctive affordance — which is, if it works well, it identities who a stranger is. That’s a major game-changer in terms of power dynamics. Historically free association has largely been safeguarded because the transaction costs of identifying strangers were protected by a natural default state of obscurity — meaning, there’s only so many names and faces we can recall. There’s a biological limit to that. And we didn’t have technologies that could radically reduce transaction costs of determining who someone is. We’ve experienced nothing like face recognition before, which is why there are major gaps in the law.Bottom line, what is face recognition good for? Since it’s good for identifying strangers, that will be how it’s used. And since it also lowers the cost of doing related things like affect recognition — using software to infer what mood or emotion a person’s face conveys — it will be used for that purpose, too. The costs of building upon the foundation of automated face recognition are very, very small.
What often leads to a slippery slop is the absence of any roadblocks to challenge things.
When considering slippery slopes, it’s crucial to ask: What are the roadblocks? What would stop us from excessively developing this technology? Unfortunately, that’s the problem with face recognition technology. There are some protective measures, even bans. But overall, in the U.S., there are far too few roadblocks. It’s hard to challenge the longstanding assumption that, for the most part, people don’t have a reasonable expectation of privacy when they’re in public.