Natanael

You're conflating things. We have no reason to argue those are true with any certainty, but we still can't exclude the possibility. It's the difference of "justified belief" vs coherent theory. Physics have had a ton of theories postulated without evidence where decades later one option was proven true and many others proven false. Under your assumption you shouldn't have made that theory before it could be tested.

I'm not arguing any specific purpose of controlling a simulation in these ways, just that the arguments saying it wouldn't happen are too weak. A multipurpose simulation (imagine one shared by many different teams of simulation researchers) could plausibly be used like this where they mess with just about anything and then reset. Doesn't mean it's likely, just that it's unreasonable to exclude the possibility

If you don't know what they're testing that could certainly seem excessive. But failure of imagination doesn't prove it's impossible, although you can argue it's unlikely

I'm not saying it happens, I'm just saying some of the arguments here aren't logically justified

Simulations of boats in water don't care about what's happening to the water much of the time yet it needs to be there, you seem to be way too confident in your conclusions

You don't rerun everything from scratch. Especially weather simulations can be checkpointed at places you have high certainty, and keep running forks after that point with different parameters. This is extremely common with for example trying to predict wind patterns during forest fires, you simulate multiple branches of possible developments in wind direction, humidity, temperature, etc. If the parameters you test don't cover every scenario that is plausible you might sometimes engineer it into the simulation just to see the worst case scenario, for example.

And in medicine, especially computational biochemistry you modify damn near everything

To the simulated object there's no difference between a fork of a simulation with different parameters vs directly changing parameters in a running simulation.

For one, we’d notice things changing without cause.

Maybe those reactions are part of the test? Or doesn't affect it. Or they abandon instances where it was noticed and the test derailed.

But it doesn't necessarily show if they have common sense. If you have many low complexity problems then maybe, but it can't predict the best performers

Checkpointing interesting points in simulations and rerunning with modified parameters happens literally all the time

Especially weather / climate / geology and medicine

In this instance it doesn't. But in this universe almost every industry using simulations run many different ones with different parameters. It doesn't make sense to assume simulation theory with only a single simulation without interventions, because that assumes the simulator already knew that what the simulation would produce would fit what they wanted and that's not a guarantee (just for information theory reasons alone!)

Why does testing numerous different circumstances and consequences violate the idea is simulation? A sufficiently capable simulation engine could literally be used for social experiments

Until Amazon Sidewalk make every smart TV connected against your will