Hacker News Reader

Comment by hawkice | Hacker News Reader

I am very skeptical of claims that outcomes are discontinuous in policy space. Going one step in direction A should give you information about going further in that same direction, at least on some scale of change. Clearly the people running the experiment agreed, because they designed it with the intent to get that information.

View on HN

> Going one step in direction A should give you information about going further in that same direction

So eating healthy for 1 day/week will change your blood results and tell you that a year of it will make you lose 50 pounds?

So going to 1 day of college will make you measure differently on a test that checks critical thinking?

Different things take different amount of time to see results. I could see behaving very differently between

* This is a very temporary 1 year test of UBI *

* You have a social safety net of UBI for your entire life, and can count on the income no matter what happens

It's more that shorter experiments might tell you whether there are deep downsides even in the short-term. Not great information, not full information, but information; even if it's just a one-way test.

It’s not that they are completely useless it’s that they cannot test the actual affects on the economy.
If you have a small scale UBI even if it’s on say a scale of a whole town it’s still not enough to evaluate the economic impact because the supply/demand economy isn’t that localized.
We have other experiments that have shown to have a negative effect for example rent allowance in the UK which essentially set the lower limit for rent as no one rents below what the council would pay, this has increased rent prices even in places where there was no negative change in the supply and demand.

It's good to be skeptical. There are a whole class of problems that are discontinuous, but it's way too tempting to label things that way.
That said, UBI _seems_ a lot more like herd immunity than most policy problems, especially social safety net policy.

Well outcomes are discontinuous where the policies are discontinuous. There's a huge leap in the graph from 'this is temporary' to 'this is permanent.' We can't just write off the effect that has on human behavior.

First, nobody is saying they're discontinuous. But there are plenty of continuous results we can see that are concave up, i.e. they require commitment beyond a certain point to be more effective than not implanting them at all.
For example, let's say initially we build a single road from coast to coast in the US, starting with no roads. There wouldn't be many places to drive to, there would be a small market for cars, and as a result of lacking economies of scale, cars would be expensive. There wouldn't be much infrastructure for purchasing or maybe even extracting gas, so that would be more expensive too. The road probably wouldn't get used much, and wouldn't be worth it.
But if you build a bunch of roads all over the country between where people want to go, eventually they'll become worth it.

I imagine that researchers designed the experiment they thought would be accepted rather than the experiment they wanted. Yes, they intended to learn some things and most likely did.

There are such things as network effects and thresholds for complex systems. So outcomes can and will be discontinuous in policy space. Perhaps for the simple reason that human society is pretty complex and certain things about it are still not very well understood.

If the criticisms of a policy is that subsidies for the voluntarily economically inactive can only be funded with massive tax increases and cuts in benefits for some of the poorest members of society, that the biggest chief beneficiaries will be house-rich early retirees and that it isn't economically sustainable, then a test which tops up a proportion of people's existing benefits out of general government funds for a short period and discovers they like free money isn't very informative.

Reply on news.ycombinator.com