According to the website, the fee is (number of users * 0.001). Since the current fee is $2.91, the number of people who've signed up is around 2,910. Seems like at least a few people willing to throw in a few bucks to give it a shot. It'll be interesting to see what will happen to growth as the rate goes up and up.

It seems that the fee should not be linear to the number of users.

Make it log(x) or something.

The incentive for scalping will increase.

How much they've made so far, with scala:

  scala> def income(n_users:Int) = (1 to n_users) map(i => (i-1)*.001) reduceLeft(_+_)
 income: (Int)Double

  scala> income(2901)
 res0: Double = 4206.45

You need to round the intermediate values down like they seem to be doing.

Well, I believe the main point of that is as an anti-spam measure.

Really the main point was curiosity - Joshua suggested the idea, and I wanted to see if it had legs. I thought it would be a good way to control the rate of growth, fight spam, and limit the number of drive-by users while offsetting some hosting costs. To my surprise, it turned out to be a way to raise Y Combinator levels of funding, with the big bonus of not having to deal with Paul Graham. I wish I had had the foresight to write down my expectations before starting the experiment.

At least for me--someone unsatisfied with the bloatware del.icio.us has become, willing to try something new, and really not too fussed about a signup fee lower than MeFi's--the cost was actually an attraction. It says "We're serious" and "We want to keep out spammers", both of which are fairly crucial to a bookmarking service.

It also makes the site stand out: probably only URI shorteners are more common than social bookmarking services, and the increasing signup fee differentiates it from all the others (I doubt I'd have given it a second thought otherwise).

No. The main point is to make money. An anti-spam measure would have worked with everyone paying $0.10. This just grows too fast.

But it is an excellent business model. If you can get 1,000 users to sign up each month and if you have no significant problems with scaling (because you only store small amounts of text), then you will have no trouble paying all fees.