Maybe they could not offer refunds after a significant percentage of the game has been played, regardless of time spent playing. Would be hard to determine how much of the game has been played though. The only way I could think of is to check achievements, but not all games have achievements.
Or maybe expected length could be part of the store metadata. If the developer estimates 1.5hr playtime Steam could use that when processing refunds.
Either way I'd consider this a corner case. To be solved sure, but not problem enough to disallow refund functionality for other games.
Probably the most famous overtly fraudulent example seems to be "Summer of '58", a game developed by a Russian developer who claimed 30-50% of their players refunded. Here's an in-depth investigation of this claim that finds absolutely no evidence to support it: https://newsletter.gamediscover.co/p/steam-refund-worries-ki... -- it doesn't positively prove that there's no refund abuse, but it makes refund abuse incredibly unlikely. Basically, it's possible to tell if someone who reviews the game refunded the game or not. So for there to be systematic refund abuse, it needs to be people who play the game to completion, then refund the game, but don't leave a review: but this pattern is not seen in any other games where developers report far lower refund rates.
Gamediscover has also had a number of developers share refund data directly for this article and other articles, including developers of short games (some as short as 20 minutes long), and none have had anything in the ballpark of the studios who have made press cause celebre pushes claiming incredibly high refund rates.
One thing to consider about this entire claim: if someone intends to proxy-pirate the video game by buying it, playing it, refunding, and cackling off into the sunset... why wouldn't they just pirate the game, something that is absolutely trivial to do? It's not really coherent to say a huge swath of gamers are manipulative enough to refund, have no moral compunction about doing so, but are too technically stupid to pull it off.
Finally, think about this from Valve's side -- if a huge % of <2 hour games were being refunded, why wouldn't Valve adopt the apparently easy proposal of allowing short game developers to set a flag that alters their refund %age? Doesn't Valve's choice not to do this suggest that they don't really view it as a common thing?
There are plenty of ways to pad game length with minimal additional development. Grind is very much a thing for games that benefit from longer time to completion or where a certain length is expected.
Is that really a big problem? Do you have actual numbers to that, and not just a hunch? I would actually guess that most people who can actually afford a game would pay for it. In my experience, most people who cheat in this way would not have bought the game anyway, so it’s not a lost sale.
> so it’s not a lost sale
Refunds probably count negatively to your rep. with a store.
Which devs and are they selling enough copies for 46% to be a meaningful amount?
How often does it happen, though? I mean if you refund a game after finishing it, it's the same as saying 'I'd rather have pirated that game', not a good sign it's 'good-faith' game either
Perhaps such a system could work if it were based on a percentage of time-to-complete, but I have no idea how one would objectively determine that, especially before sale.