Thoughts for February 20, 2022
Good afternoon. Today’s topics include the alignment problem, vacancy taxes, and homeless services.
The Alignment Problem
On my Tumblr blog, I wrote a post about the alignment problem, which is how we insure that when artificial intelligence is advanced enough to act autonomously, it acts in a way that people would desire. I’m not going to link the post because it wasn’t very good.
The reason the post wasn’t very good is that, try as I might, I just can’t get myself to be very concerned about this topic. In the abstract it seems like a major problem. But it also seems like a very hypothetical problem. At the apparent rate of development, it will be decades at least before there exists an AI system advanced enough to pose a threat, and in the meantime there are much more pressing concerns. I have no problem with other people investing their efforts into the alignment problem, and it is probably a good thing they do so, but I won’t be joining them.
With software development, there is a major risk from dependencies. All projects, including software projects, must depend on some other things, including other software. However, if the dependencies are not reliable, there is a risk that the software project will fail as a result. Cybercrime—caused by malicious humans rather than bad software—is estimated to cost trillions of dollars annually already (the figure in this report seems high to me, but I don’t have another to go by. Regardless, it’s a major problem). There are other issues that could be discussed. I strongly suspect that dealing with these two problems—over-reliance on dependencies and cybercrime—will do much more to reduce alignment risk than direct work on the alignment problem.
Outside of the software world, there is a parallel with the problem of institutional drift. Governments, corporates, universities, nonprofits, and all other kinds of institutions drift in their missions in ways that are beyond the control of any individual. Thus institutions frequently come to operate in a way that is neither in the interest in society as a whole or in alignment with the intended mission of the institution.
Vacancy Tax
There is a proposal by Supervisor Dean Preston to institute a residential vacancy tax. The measure would impose a tax on vacant residential properties based on how long they have been vacant.
The city of Vancouver, BC instituted a vacancy tax a few years ago, which has had modest effects on prices. Washington DC applied a vacancy tax to deal with blight in commercial properties; the effect of that is unclear.
Even though a vacancy tax might have modest benefits for residential prices, I’m not too keen on the idea. It is a blunt, punitive instrument, seemingly akin to trying to deal with unemployment by taxing unemployed people. Vacant units are self-taxing in that the owner loses out on potential revenue by not renting them.
The political economy is bad too. There is a persistent myth that throughout American cities, there is an abundance of vacant units that are only vacant because of greedy speculators, and if only they can be appropriated for use, problems of high housing costs would be solved without any new construction. Preston promotes this myth, which no doubt is why he proposes the tax. It is a foolish and unserious idea, the equivalent of “we don’t need GMOs because we can increase ‘yields’ even more by eliminating food waste”.
I am scratching my head, then, as to why the YIMBY movement in San Francisco is investing their limited political capital into an issue with piddly benefits, crafted by a politician who is hostile to the movement. Maybe it is a case of the Shirky Principle.
Homeless Services
For the last three weeks, I have volunteered with an organization called Operation Nightwatch, which provides meals, clothing, hygiene kits, and other staples to Portland’s homeless population. For as much time as I spend reading academic papers, it would be worthwhile to get out and do some stuff.
A common idea is that most people are homeless because of mental health and/or substance abuse problem. I’m not sure what the data says, but at least from what I can observe, this does not appear to be the case. Granted, my vantage point is from getting some goods to people waiting in line and is thus very limited.
Last night also brought the first serious problem I have seen, and according to the organizers, the first security incident in two years. A clearly agitated woman, who had evidently been there before, began screaming obscenities at staff and eventually threatened them with a knife. We were forced to shut down operations early, after which she proceeded to throw a brick and a scooter at the window. One of the staff called the police, but they refused to respond, as there was also a mass shooting in Portland last night which occupied the police’s attention. This is the first time I have experienced 911 refusing to respond to a call.
Overall, I think Operation Nightwatch does good and badly needed work. We are responding to problems that the city is unable or unwilling to deal with adequately.