The reason that Colin can get away with the picodollar model is that the type of people who use tarsnap would get annoyed if they were charged for 1gb if they were using 100mb. _For those customers_, there is a clear correlation between the price being charged and the value provided.
If Colin is attracting people for whom the difference between $0.03 and $0.30 is meaningful, then Colin is attracting people that do not deserve to do business with Colin. Colin was involved in security with BSD, and made a no-BS almost bulletproof encrypted backup solution on top of untrusted hardware, and has created something of actual value. Anyone whose data is worth less than, oh, call it $100 a month (at the low end) doesn't really need Tarsnap.
I respect that everyone has their own motivations for doing things and that economic optimality is not the be-all-end-all. Colin can run Tarsnap as an underpriced public utility if he wants to. But, Scout's honor, it is underpriced. The pricing model actively discourages large, savvy customers whose lives and businesses would benefit from switching to Tarsnap in favor of pathological bottom-feeders who gain no significant benefit from Tarsnap. My personal opinion is, in addition to costing Colin money, this moots the impact that his (extraordinarily impressive) engineering achievement could otherwise be providing to the world.
This is a very awkward conversion for me, so I'm going to refrain from the unsolicited advice for at least a few months until picodollars just make me snap again.
You've disastrously misunderstood what Colin is doing.
Colin "gets away" with "picodollar pricing" because he is deliberately segmenting away most of his customer base. I am not entirely sure why he is doing that, but he has made a strategic decision in the short term to make less money in order to build his service up more slowly and, I presume, carefully. He exclusively wants the kind of customers who will not be annoyed with this annoying price model; in other words, the kind of customer that has time to reason through the idiosyncratic pricing scheme of a new backup service.
Unless slowing the growth of your own service is one of your goals, do not mistake what Colin is doing here with a pricing strategy that works in the general case.
I can't help agreeing with Patrick here, but I'm not Colin so I can't really presume too much.
In any case I always thought that Colin could get out of the low price quicksand by launching an entirely rebranded service with updated pricing based on value to customer, etc. sure, keep the original guys around too. Throw in some perks on the "new" service that are attractive to enterprise customers. It's not every day you have a product of this quality and (potential) customers willing to pay so much.
Assuming the goal is to have only technically sophisticated users who want to use the most durable data store known to man (S3) and have a rock solid & secure backup solution, the pricing structure being 3-6 times that of s3 pricing is a great value on all sides.
actually: colin, could you tell us perhaps what the total monthly storage and or bandwidth volume is? I don't think that if has ever been stated on the internet :)
caveat: i've been playing with tarsnap on/off since it was first launched, and now that i'm doing my own wee startup, i'm using it quite seriously and religiously to backup my working directories. Love it. (however much i email the tarsnap user list with some confusion thats easy to sort out every time i've resumed using tarsnap after some hiatus :) )
