We’re frequently told that if cable companies would allow us to buy channels à la carte, we could save lots of money by not having to pay for all those hundreds of channels we never watch. As the following numerical example illustrates, however, channel bundling may actually enable cable companies to provide offerings that many of us like better than what we’d get under an à-la-carte system.

The key feature of the example is one that also figures prominently in other digital content markets—namely, that the costs of production are mostly fixed: Once content is in hand, the marginal cost of serving additional consumers is essentially zero. Normal intuitions about pricing often break down under these conditions, and that’s what happens here.

For simplicity, suppose there is one premium channel that everyone would like to have (call it HBO) and 100 fringe channels, none of which is valued by more than a small proportion of potential viewers. Suppose there are 120 million potential subscribers in total, that their maximum monthly willingness to pay for access to HBO and the 100 fringe channels is as summarized in this table:

If the cable company sold HBO separately, their best bet would be to gouge the rich TV fans, selling 10 million subscriptions at a price of $70 each, for $700 million in monthly revenue. They’d also sell 120 million subscriptions to the bundle of 100 fringe channels at a price of $5 each, for another $600 million in revenue. Total monthly revenue: $1,300 million.

Now suppose they can offer a premium bundle consisting of the 100 fringe channels plus HBO. Their best bet would then be to charge $20 a month for that package, at which price they’d sell 50 million units, for monthly revenue of $1,000 million. They’d also sell another 70 million monthly subscriptions to the 100 fringe channels at $5 each, for another $350 million in revenue. So total revenue would be $1,350 million, or $50 million more than before. Yet no subscriber would end up with a package that’s worse than before, and 10 million former HBO à-la-carte subscribers would be paying $55 a month less than before for the combination of HBO and the fringe channels.

This is obviously just a hypothetical example. But it captures the basic idea that when the marginal cost of serving additional subscribers is zero, the average cost of serving each one necessarily goes down whenever sellers can price their offerings in ways that result in increased numbers of units sold.

Tech industry analyst Ben Thompson has described the current cable bundling system as “socialism that works,” offering data suggesting that the implications of my numerical example are hardly farfetched. No one should be surprised, then, if prices for channels like HBO and ESPN go up sharply if cable providers are forced to offer them as stand-alones. That wouldn’t be a good thing for the people who now get those channels as part of a cheaper cable bundle, nor would it help those subscribers who don’t care about the channels.