[LeBleu]

In implying the penny gap can be eliminated, this article fails to account for mental transaction costs. No matter how simple a microbilling service is, there is still a mental transaction cost for deciding whether to pay or not. If my time is worth $20/hour (US average wage), and it takes me 3 seconds to read your microbilling screen and understand what you are charging, it has already cost me a penny. 3 seconds is enough time for the average reader to read 5 words. I would estimate you have about a minute to explain your service and charge for it, if you want the customer to spend a quarter on something that is worth fifty cents (or more) to them. (See http://szabo.best.vwh.net/micropayments.html or http://en.wikipedia.org/wiki/Micropayment#Theory.2Fcriticism for more background on mental transaction costs.)

If the price of your freemium service is artificially high, then you are doing it wrong. The freemium model is for services where the price of supporting free users is less than the price of other ways of informing your paying customers that you exist. (Including cases where free users contribute value.)

The best way to structure the freemium model is such that you only charge for things that are natural scarcities, not things that can be freely replicated. For example, charging for support (human time to solve your problem) is better than charging for features. (Existing features aren't scarce, it costs nothing to roll them out to all users. Implementing new features is scarce.)

[davidmathers]

In implying the penny gap can be eliminated, this article fails to account for mental transaction costs.

Seriously. Clay Shirky explained this 9 years ago: http://www.openp2p.com/pub/a/p2p/2000/12/19/micropayments.ht...

I can't believe someone is writing this in 2009. There's so much wrong with it I don't even know where to begin, but this paragraph stuck me as particularly wrong:

"Fair enough. But here’s one thing freemium fans can’t deny: in their model, a tiny minority of paid users subsidizes the service for everybody. It is this simple fact that makes the freemium model self-defeating, because, for the numbers to work, the price of the paid service must be set artificially high."

That's just pure mental confusion and abuse of language by the author.

Try this thought experiment: github launched as a micropayment service. It would already be dead. How do you calculate the "fair" price of something that doesn't exist?

[hymanroth]

Shirky was talking about micropayments for content in that article. An endless stream of little decisions. I agree with him.

However, as I pointed out in a comment above. The one-off mental transaction cost for agreeing to be micro-billed for a continuous service is less than that required to sign-up for a premium version of a service.

As to my supposed abuse of language, I don't understand what you mean.

[davidmathers]

As to my supposed abuse of language, I don't understand what you mean.

Correct me if I'm misunderstanding you on this part: If your claim is that the paying users of github are subsidizing the free users, that's simply not true and is a misuse of the word subsidize.

Calling a price "artificially high" on the other hand is meaningless. The correct price is the price that create the most value. There's no such thing as an artificially high price in an open marketplace. Or, more specifically, it's impossible for the producer of a good or service to set a price that is artificially high.

[LeBleu]

There's no such thing as an artificially high price in an open marketplace. Or, more specifically, it's impossible for the producer of a good or service to set a price that is artificially high.

That's not true. The real world is not the same as an idealized economic model. It is quite possible for a business to set an artificially high price, at least in the short term. In a competitive market, it results in the business getting less revenue, and quite possibly going out of business, but that is not instantaneous. If the price is merely sub-optimal, but not outrageous, it is quite possible that the overcharging business will make enough money to stay around for years.

Even John Maynard Keynes commented on those trying to profit by shorting irrational bubbles, "Markets can remain irrational longer than you can remain solvent." -- if you're a company benefiting from the current irrationality, that means you can stay around until the market returns to rationality. (For example, see Citibank, Lehman brothers, etc....)

This is without even accounting for the fact that these services are usually not pure commodity markets. They are in monopolistic competition (http://en.wikipedia.org/wiki/Monopolistic_competition), sometimes with only very distant substitutes, giving closer to monopoly control over their market.

[hymanroth]

If you use a service without charge which entails a non-zero cost to produce and supply it, then I'm afraid somebody, somewhere is subsiding it. So that's my first abuse of language sorted.

As to the price of premium services being 'artificially high' - as I make clear in the post - this refers to paying for a benefit that is necessarily incremental (if it wasn't, nobody would use the free service). The price of that benefit is generally excessive when compared to what can be had for nothing.

You many not agree with those words, but I don't think they, too, represent an abuse of language.

[hymanroth]

The mental transaction cost is certainly an issue with content. I wouldn't want to have to decide every time I start reading a news article whether I want to pay to get to the end.

However, the one-off mental cost of deciding whether to allow a web service to bill me for future usage is not a big deal, and, anyway, it applies even more to the choice of upgrading from a free to premium version.

In fact, it applies to any instance where the user is faced with having to pay for something.

What do you propose? Everything should always be free?

As for the scarcity argument, charging for support is a horrible business model. Look at Red Hat's revenues per user vs MSFT's. The key is to leverage the negligible marginal distribution costs of digital content.