Developer salaries and rent are the same no matter how many copies you sell. Each ten-cent copy sold brings you closer to covering those costs. If you don’t make enough to cover costs, you’re not selling at a loss, you’re just not selling enough.

The marginal cost on infrastructure is like the marginal cost on bandwidth: really small.

> you’re not selling at a loss, you’re just not selling enough

Ah, you could have just said it's a matter of wording. Yes, what I meant is you could sell, and make a loss. So in order to drive up sales hoping that it will actually start a positive feedback loop where the growing community attracts more buyers, you could sell your software at different prices. You probably noticed even some cloud services sell for different prices depending on country (MS gives you some cloud things 20% more expensive in Germany as opposed to US).

And saying the infrastructure is just bandwidth is like saying you just a keyboard to code. Random example, I assume many big name games face similar challenges. [0] I think we can agree maintaining a setup that allows players all over the world to play in good conditions while preventing cheating, and many, many other services costs and involve quite a bit more than just the net link.

[0] https://www.networkworld.com/article/2895129/big-data-and-ba...

Your previous comment referred to losses in a single country. It makes no sense to talk about total profit/loss including overhead in that context. How do you allocate your shared overhead between the two countries such that you can say you made a loss in one country but a profit in another? If you're "selling at a loss" in a single country then the only reasonable interpretation of that is to account for the additional costs involved in serving that country, but not any shared overhead.

Yes, there are lots of other costs for running a network service besides bandwidth. But those are heavily if not completely independent of the number of users. Your anti-cheating effort costs about the same regardless of whether you have a million users or ten million users.

If you expect your game in India will sell 50.000 copies at $60 but 250.000 copies at $10 you are losing out on some of the money ($500K in this example). On the other hand you can sell the same game in the US for $60 or $65 knowing that the predicted sales will cover up for the money that you didn't make in India. Bonus, you now have 200.000 more active players in the world than if you tried selling everywhere at a middle price point ($50-$55...).

So you can sell the game at a loss meaning once your sales have dried up you haven been able to make a profit or even cover your costs. And you can "sell at a loss in one country" meaning you're willing to let that one country not cover its share of the predicted income (meaning if you sold everywhere like that you'd go bankrupt fast) just because you can get more market share. And some other region will compensate.

I'm sorry you would rather nitpick on the wording instead of the point being made, which I'm sure I explained pretty clearly. And this already exists in practice so obviously the logic behind it is sound.

> But those are heavily if not completely independent of the number of users.

So what if the number of users/sessions hosted increased by 3 orders of magnitude, right? The databases, the game servers, the login servers, the web servers, security, management, and everything else that also has to be geo-located will just balloon to accommodate the extra users or markets at no cost as all IT infrastructure tends to do.

Again, here are some random numbers from almost a decade ago for a game that's probably far less demanding than anything today [0]. Having x US servers vs. having 3x US servers and 5x Asia servers is not quite "independent" of user numbers and location.

[0] http://twvideo01.ubm-us.net/o1/vault/gdceurope2011/slides/Ma...