For example, if the participants who receive this money are poor and they know that the money will stop coming 5 years from now, they may save every penny they can. This may lead us to conclude that basic income does not lead to irresponsible spending. But if any government institutes basic income indefinitely based on the results of this study, it may produce a very different outcome.
Wouldn't it make more sense to start at a small-scale and validate that it's even possible that it works? What if they spend the money to do a 5-year study and find that all participants are massively irresponsible with the money and that BI would be a disaster of a social policy?
Why commit to a 45-year experiment when we have no data right now? The scientists working on nuclear fission aren't saying: "Let's give up because we can't build a commercial scale energy positive reactor". They're saying: "Let's make progress one small-step at a time on a path that gets us there eventually".
We need results for every experiment today. Unfortunately that's unrealistic and not how science works. You can't simply compromise your study to get results faster; you'll never know how valid your study's results really are in such a case.
It sucks that if you want a good study for basic income you pretty much have to do it, effectively, forever while aggregating and gathering results over time but you can't just shortcut science because "we kind of need results sooner".
The only downside of all UBI experiments I personally do not like is how there is no way to model the cultural shifts involved with it, and how the children of the UBI generation will grow up and mature into society, because I imagine it will be incredibly divergent once you can ask fundamental questions about what all the schooling is for and how one should best model their life for fulfillment.
That might be true, but financing basic income for life is a rather expensive research proposition.
You can't really judge a states spending exclusively in tax dollars in, or you get where we are in the US with derelict infrastructure and understaffed hospitals, fire departments, and public schools. You are taxing money to spend it for a reason. As long as the results are worth the price paid (and infrastructure at least is enough of a force multiplier to give you a lot of overhead for corruption before it is inefficient) you aren't "spending a lot of money" because your economy is growing more than you shrink it by taxing it.
It is only effectively dead weight taxes (corruption, the military, modern foreign aid, politicians salaries, bureaucracy, insane overregulation) that is tax money wasted. And sure, a lot of it is wasted. Which is half the argument for UBI - the metrics are simple and should be transparent to prevent anyone corrupting what would be a force multiplier.
Ultimately, shorter term studies are an approximation. They're the best you can do. We can't do a life long study on this drug before making it available but we can draw information from shorter studies. Similarly, we can't do a life long basic income experiment, there's not the money out there, but we can study the shorter approximation and do it with more people and draw conclusions from that. All experiments entail compromise. The fact that neither of these bodies of science can have the ideal experiment is not a flaw but just part of doing science.
The general critique seems to be: the only way to study BI at any level is to run experiments that last for a person's full lifetime. It seems insane to me that you wouldn't do small scale studies and pilot programs that gather data and build up to longer and longer studies.
When I'm designing a new system I don't say: the only way to validate that the concepts are correct are to write the production-grade code at all levels and see if it works. You find ways to validate key assumptions and build smaller scale proof-of-concept models to see if any of your theories are just totally broken.
No, the experiments don't need to run a persons lifetime, the benefit, however, needs to be believably of indefinite duration, because we already have both an understanding of the economic rationale and empirical studies showing that explicitly-limited-term benefits or costs induce notably different behavioral impacts than open-ended ones.
Its the benefit, not the study of the effects of the benefit that needs to be open-ended.
It's basically impossible to imagine a perfect (ie. time-unlimited) basic income experiment. For one thing, we kind of need results sooner rather than later. I don't think waiting a generation to get crucial results for the future of political economics is tenable.
Moreover, financing for such an experiment would have to be truly massive.