I'm half-way tempted to apply for this despite the fact that a) I'm not moving from NYC to San Francisco, and b) I'm probably not qualified.
In spite of that, I have some real questions and comments about the constraints on a project like this.
1. What's the budget for this?
In order to get any generalizable data, you'd have to set up your sample plan to be geographically and demographically diverse. In an ideal world, you'd need to provide for 1,800 subjects to get genuinely usable data. Research is rarely ideal, and you can infer from smaller samples, but how many subjects are we talking about? 5? 10? 100? 500? 1,000?
Also, what constitutes a sample? Individual or household? (see 2, paragraph 2 below)
2. What constitutes Basic Income?
I'd argue that Basic Income represents enough for all human needs to be fulfilled to some arbitrarily minimal extent. Unlike some other comments, I include the need for entertainment and sex among basic human needs. As well as catastrophic health care. Does that mean everyone gets a free PS 4? No. And I also don't mean paying for prostitutes. But some budget for entertainment of some kind makes sense to me, as does a budget for dating.
Anything short of that fails, in my opinion, to be a basic income. It's more like welfare for everyone, which is conceptually different, in my opinion. I also think Basic Income should be adjusted for location and household size. Although it might not be the right metric, the most convenient one we have is centuries of data about total household income vs. household size. That's how the Census Bureau defines the poverty threshold, for example.
There are a lot of different ways to think about this, and my opinions above are just that: opinions. The point I'm trying to make is that if this experiment is going to happen, what's the best possible scenario for measuring something useful?
3. How do we research what is happening without affecting the outcome?
I actually have an answer to this one. I think it's inevitable that any time a government agency is processing payments, there are at least some hoops to jump through. Welfare has its process, as does unemployment. It's reasonable to think we can design a brief questionnaire that tracks certain key variables like productivity, happiness, artistic endeavors, anxiety (bonus points for using a wearable device to track stress levels?? Maybe?) . . . whatever. I'm not sure what those should be.
But it could be structured in a way that is realistic enough that you wouldn't do any obvious damage to the study by observation. Once your location and status have been established, the benefits do not change depending on your answers. But you have to check in and request a payment on the 25th and 10 of every month (to guarantee payments on the 1st and 15th).
4. It's important to note that there will be some people who adjust to whatever income they are making and do nothing beyond that. I don't know what percentage that is, but the funders have to understand that at least some people will do nothing but sit around and drink or play video games, and that can't be a reason to cut them off, otherwise, it really biases things in a bad way.
5. If this is able to be truly representative, then you would have to expect at least some percent of the participants in the experiment to get into legal trouble.
What do we do with them?
If you set a condition that people stay out of legal trouble to get the benefits (a genuinely real scenario if we are trying to model how this might be implemented in the future), then you are tampering with your sample.
If you don't attach such a condition, you're going to draw a huge amount of criticism for paying people to be in jail, potentially funding criminal operations of various kinds, and there will be a small portion of people who choose to go to jail who have been there before and are comfortable with it and live extremely cheaply while racking up the savings. That's a tough one.
6. Going back to budget: is there enough to do control vs. group testing?
By that, I mean we have a control group that answers the same survey every two weeks, but only gets a small incentive to do so. Nothing even close to a basic income. Enough to get them to hit a website and answer a 5-minute questionnaire about their life for the last two weeks, identical to the one the BI people are required to answer for their payments.
And how many strategies can we afford to test? If we have rep sample control, then we can rep sample a variety of strategies: pro-rated based on location and household size, flat BI same for everyone, or other nuanced approaches.
I guess I've really seriously gone and buried the lede here, but Sam, what is the scope of this project? My experience in research is that there's not really an analogue to the MVP in the startup world.
Research needs to be rigorous and correct from the outset. You can't hack something together that mostly works and then see if it gets traction and find bigger funders if it does. I'm not suggesting that's your line of thinking, but this is a big project.
Getting meaningful, generalizable data about this is going to cost a lot.