They could threaten EA's other apps, though, or perhaps refund all purchases for the game.

I think if you have an abusive app developer, its within reason to ban them from the platform.

Of course, if that app developer is a major player like EA, I would imagine the definition of "abusive" suddenly becomes much more liberal.

Rock Band isn't a multiplayer game and shouldn't need any servers to play.

"Shouldn't" and "doesn't" are different concepts though. Just as you can't reasonably "force" EA to continue operating a service forever, you can't go back in time to make them architect it properly either.

This is inevitable in the modern world. We want to allow for people to pay for software. We want software to engage with the broader world. Having those two things at once inevitably leads to a situation where software breaks because its "managers" don't want to maintain that engagement anymore.

So pick one of those two and throw it out: make all services pay-for-play instead of one-time-fee. Or disallow non-standalone apps. (Or I guess a third option would be to force the customer and app vendor to sign a contract with a service level agreement). None of those seem workable to me.

I worry that you might be right, but it seems to me like we're moving towards a Soviet economic model where we fleece the users today and disable the services tomorrow. It would not have occurred to me that applications installed on my phone might one day vanish due to the publisher not wanting to support them. I don't have this game, but I don't think it's likely that I would have expected this behavior. Acting like this is a normal, acceptable business practice when it is basically unprecedented in our field is not going to discourage the behavior. Between this and various cloud offerings vanishing after acquisition or because Google or whoever doesn't feel like supporting them, I think we're entering an era where software as a service has become software as a means for fly-by-night companies to defraud users and destroy their trust.

Apple can't "force" EA to architect it properly, but they can refund everyone who purchased the game and/or punish EA in any way they see fit.

Which raises the question of what is a reasonable expected lifetime of an application? If, say, my app stops working in a future iOS update years from now and I take no action to update it, should those customers still expect a full refund even though they've used my app for years already?

Assuming you weren't doing anything sneaky, I think not, because Apple (at least theoretically) takes responsibility for not breaking compliant applications with updates. There's a world of difference between Apple breaking your app and you breaking your app.

There seems to be a parallel with other warranty/consumer protection laws here.

In the UK, for example, we have laws like the Sale of Goods Act, under which anything you buy firsthand must meet certain criteria such as being fit for the purpose it was sold for. If it isn't, you are typically entitled to a repair/replacement/refund. However, that need not necessarily be a full refund or replacement with a brand new equivalent, depending on the circumstances.

Typically, if a physical product is found to be faulty within six months of purchase, the onus is on the merchant to prove that it was not faulty when it was bought. After that, the burden shifts somewhat, but the merchant can still be liable for fixing a problem even a few years after the original purchase if the item would normally be expected to last that long. As time goes by, the amount of a refund you might be entitled to receive if the product breaks and isn't be repaired or replaced instead would tend to diminish, to reflect the value you have already received from the use of the product before it broke.

Usually, in physical product world, a court could make a fair determination of how long something might reasonably be expected to last based on the general market, the original purchase price, etc. Of course with software it's harder, because bits don't magically break, but on the other hand there are platform compatibility issues to consider.

As you say, that raises the question of what is a reasonable expected lifetime. If we're talking about pure content -- an e-book, music, a movie, etc. -- then I would argue the answer is "forever". If someone has paid for the content, in the expectation that it was a full purchase and not for example a rental or all-you-can-watch fixed-time subscription, then there is simply no excuse for ever deactivating the product after the purchase.

If we're talking about executable code in an application, then I would argue that the answer must be as a minimum "as long as the supporting platform is available", where by supporting platform I mean the device/OS/etc. necessary to run the code, and not including any sort of artificial dependency introduced exclusively or primarily for the purpose of allowing the software to be disabled in the future. If there is a legitimate dependency on some system outside the consumer's personal control, for example a centralised server used to co-ordinate a multiplayer game and operated at the game developer's expense, then I wouldn't have a problem with a system whereby the dependency was OK but anyone who wanted to maintain the usual legal protections for software would have to leave in escrow a version of the software that allowed third party systems to replace such dependencies if and when the original systems ceased to be available.