It's more than just sandboxing: Apps going through the App Store have to meet a stringent set of requirements, chiefly being no private API usage, no circumvention of features intended to give the user control/privacy, and no third party web engines (and all the performance and security implications that come with those). Apple regularly rejects apps that violate the first two, and while I'm sure some get through they tend to get buried by scores of more legitimate alternatives. Sideloaded apps get to bypass all of this.

If they ever do offer sideloading/gatekeeper, it should be turned off by default and turning it on should make the risks crystal clear with a scary alert and passcode prompt. Average folks should be heavily discouraged against using it. Gatekeeper works on the Mac, but iOS devices are both far more numerous and far more personal, so the stakes are much higher.

Apple's checks are not very deep. They suffice to keep honest people honest but do little against an actual adversary.

The number of safeguards they have in place make that a moot point. If it's not discovered in testing, it's sandboxed. If it gets past the sandbox, they have a killswitch that's registered to the app's specific ID. There's very little damage that an app can do if it's gone through the App Store proper.

That could be done for sideloaded apps too.

Sideloaded apps, by definition, are not tracked individually by Apple. You can't make a killswitch for an app if you don't know what the app is.

And yet the Mac has a malware killswitch. Clearly it can be done.

Source? The Mac only has the ability to prevent unsigned apps from being installed which is easily disabled. There's no killswitch for unsigned apps.

Sideloaded apps make attacks on non-sideloaded stuff a lot easier.

Not to sound like a broken record, but, how?