newstorkcity

Google will report how busy some business are on an hour by hour basis, I assume based on anonymous phone locations collected. If google covers the business you want to go to that would solve the coordination problem.

I'm not sure how to make google show this information. Usually I can just look up "[business] hours" and it will give me the plot, it might not be showing that information right now due to low activity because of the lock down.

This is impressive, does anyone know what model it is using for the AI. It says "neural transcompiler", does that mean it is using a neural net?

Without some kind of efficiency benchmark, it's hard to say how useful this would be. You can make a universal transcompiler for anything that compiles into byte code by writing having a set function for every byte code, but obviously that would be horribly inefficient. I assume this wouldn't do something that awful, but without seeing what kind of code it outputs we don't know what it's like.

From what I understand, this is arguing that we should be collectively doing this in order to limit overcrowding, this method is not very helpful as an individual. If based on past trends it is overcapacity at 2:00 and under capacity at 4:00, as and individual you shouldn't expect that to change and should always go at 4:00.

Without a prescription for how many people want to go at a given time, you can't really make use of this information. If every body was following this rule, there ought to never be overcrowding, so you could never know this information by using past data. So you would need to use a sign up or polling method to figure out. If using a sign up, you could just set a cap on how many people can sign up so you don't need randomness. If using polling then this method could be helpful.

Though another issue with this model is assuming a fixed positive utility from going when overcrowded and a fixed negative utility when overcrowded. In reality the fewer people the higher utility, plus each individual has different utility based on what time they are going. Taking all this into account, I feel like the only reasonable approach is what is already done, go to the store during a convenient time for you that is the least busy based on passed data. If everyone does this, we should expected to reach a happy equilibrium (sans unexpected upsets, such as the recent rushes before lock downs).