And although I applaud your efforts and accomplishments at Tarsnap, I must say that you base your argument on significantly less than 100 startups.
I wish I could agree with you here, but I'm afraid I've seen far too many comments along the lines of "you're violating pg's rule XYZ, so you must be doing something wrong". Yes, most people here probably understand that pg's guidance doesn't apply to every situation, but I still think it's worth reminding people occasionally.
I don't believe this sort of comment is helpful. For one thing, "X works" and "Y can work, too" are not two bits of advice which are in opposition to each other, and shouldn't be treated as such. (Even when X = "Startups should have multiple founders" and Y = "Actually, you can get by with one person if...")
For another thing, evaluating expertise by weight of incomparable circumstances is not logical. If it were, we'd all be taking advice from McDonalds on how to sell software, as the sum total of all customers of every startup company which has ever been mentioned on this site would not equal the number of cups of coffee McDonalds sold worldwide yesterday. Now there are many things to learn from McDonalds, but willy-nilly application of their ideas because they have mass on their side is a bad idea. (The sweet spot for pricing is below $5. Advertising is less about motivating people to buy in the near future than building your brand image for life. Literacy is a nice attribute to have in an employee but is not necessary if you exercise proper management.)
Scribd: 5M daily uniques
Justin.tv: 3M daily uniques
last.fm: 2M daily uniques
It might be satisfying to include some reference lines on your daily user graph, so you can celebrate when you serve more pages than, say, McDonald's sells fish sandwiches, or Wal-Mart sells Spiderman action figures, or whatever.
Of course, this is a meaningless comparison, because the number of people that Scribd or Justin.TV reach on a daily basis is more closely correlated to the number of users they have than the number of customers.
On the other hand, PG has also spent the past 3 years funding different startups and must have some decent idea about what is more likely to help a startup.
Same goes for you. I would assume that you are more experience in Bingo software then other programmers even if they are better programmers than you are. But in the field of Bingo, you would probably crush several highly talented developers.
"Weight of incomparable circumstances" probably.