Steam games are also intangible. For each game that Mr. Rohrer sells on steam, his marginal cost is exactly 0(since Steam pays for the bandwidth). Therefore he can theoretically produce and sell an infinite amount of the product since it costs nothing to produce(I'm not counting any contracts he's made to profit share since those are certainly percentage shares and therefore not applicable to this argument). So economically speaking, the world will best be served should he sell his product for as little as possible, generating the maximum amount of total surplus.
But the market for video games is not a ideal "perfect market" like you learn in Econ 101. Specifically, two of the perfect market assumptions are broken because the publisher has a monopoly for the specific game - for which there are only imperfect substitutes - and there are barriers to entry due to the significant sunk costs necessary for game development.
Rohrer (or any content-owner with a zero marginal cost good like a game) would indeed maximize the utility extracted from a given game by giving it away for free but it would not be rational economic behavior for them to do so. Unless, of course, Rohrer were artistically or philanthropically motivated and derived more personal utility from having the maximum number of people enjoy his game than he would from the money to be made from selling it to a fraction of them.
He's not giving it away but he is (ostensibly, at least) trading money for artistic satisfaction by refusing to engage in the profit-maximizing behavior that is typical of the way games are sold. A good in a perfectly competitive market is priced lower than what almost all consumers would be willing to pay. The difference is called "consumer surplus" and is the area below the demand curve and above the price on the left side of the supply and demand chart. The typical behavior in the game market is for the publisher to try to capture as much of that surplus as possible by selling the game for exactly what each consumer is willing to pay for it by gradually lowering the price in ways that the consumers cannot really anticipate - i.e. sales.
In this example, if the price is permanently fixed at $10 with the green supply curve (flat due to zero marginal costs), the blue rectangle is the money made by the merchant ($10 * 250,000 = $2.5 million), the maximum for the given blue OnePriceDemand curve. The red is the consumer surplus - happy consumers who (assuming perfect information, due to informative reviews and demos and such) knew they would enjoy the game more than the $10 they spent.
But when consumers believe that the game will eventually go on sale, the demand curve bows in. The extreme consumers - those who don't really care how much it costs and those who would only really play it for free don't change, but the ones in the middle are not willing to pay as much because they are willing to wait for a sale. The orange SalesDemand curve assumes (unrealistically) that the merchant can perfectly set a price for every single consumer. This gives the entire area under the orange curve to the merchant (about $3 million instead of $2.5 million) and leaves the consumers with zero surplus. No consumer is truly happy because they are each only marginally more happy than they would have been if they had kept their money - not ripped off but just barely worth it.
As always in economics, the real answer depends on details that are difficult to measure in reality. No one can perfectly capture the entire consumer surplus and in reality there would be a bunch of green lines at various sale price points with little red triangles above them. Depending on the shape of the lines it is possible for both consumers and the merchant to get more value by taking advantage of the white space under the orange line, giving the consumers a smaller share of a bigger pie. But even if that is true, the people who like the game the most are getting less utility in the sales model - it's just made up for by allowing a bunch of people who sort of like the game to sort of enjoy it because they got it on sale.
Assuming Rohrer is honest about his motivations (and this isn't just a PR scheme or a plan to maximize profits by trying to make the launch of a multi-player game bigger), I think he should be commended for wanting the people who like his game the best to get the most value out of it and being willing to leave money on the table to make it happen.
Pricing models with anticipated sales are good for some consumers and bad for others. The economics of a no-sales pricing model are likely bad for the publisher but good for the consumers who are most enthusiastic about a product, so good for Rohrer.