|
|
|
|
|
|
by tnecniv
3276 days ago
|
|
|
I'd be interested in hearing someone with knowledge talk about how these models get used in practice. From an engineering perspective it seems like actually applying them would be very hard. For example, how good are parameter estimates and how do we know if they are good? If I was designing a controller for a plane, I can spend a lot of time getting good parameters for the plane model and design a controller that looks good in simulation. However, when I take the plane out, it might not be stable (or doesn't meet some other metric I designed for) because the parameters changed due to a variety of physical factors and/or the model is a model and thus doesn't account for anything. So I go back and forth tweaking the controller design until I get something that works. That process isn't really feasible when it comes to economies, though. |
|
|
Would building high speed rail from NYC to Chicago be a net positive for consumer welfare? Well we can use estimates of how much people pay for various modes of transportation throughout the country, and how often they're used. We can also come up with an estimate of the cost of building the rail. The analogous effects to your feedback loop are equilibrium effects - for example if people start taking the new high speed rail then plane ticket prices will fall a bit. But it adds more uncertainty to the model, and we know that plenty of flights from NYC to Chicago are connecting flights, so probably the effects are pretty small, and we can just assume those prices stay fixed for the sake of tractability.
If you do this kind of calculation and find the rail project is a massive multi billion dollar money pit, you can confidently recommend against it despite the fact you've ignored some of the feedback effects.