[olefoo]

Recognizing the harsh practical reality is not "giving up"; if you're acting under the constraint that the end user of a web service must also be the customer, then you cannot under any circumstances get people to pay the actual costs per/customer of building something like Facebook or Google. There are other businesses that you could build at that scale that do fit under that constraint, but not web services that depend on millions of simultaneous users to produce any value at all.

The best estimates I have for what it costs google to provide service to one customer for 1/year is $20 < n < $24 and Facebook is probably more, can you name even one person of your acquaintance who would pay even $12/year for either of those services...? Even if hardly anyone else was using them? Before they were big?

So yeah, impossible, at least in the given constraints; like asking for a 50 seat airplane that's self-piloted, safer than current small planes AND costs less than $500 to build.

You might argue about my definition of impossible, but this isn't reddit, and I hope you recognize and understand that difference.

[jseims]

According to Facebook's S1 filings, their expenses correspond to roughly $1 per user per year, and their revenue is $4 per user per year.

I would pay > $4 per year if it meant I controlled my Facebook data (i.e., I was the customer, not the product).

[xiaoma]

I would absolutely be willing to pay $24 a year for what Google offers for free if the alternatives were what existed before those services got big (i.e. altavista and hotmail). Search and spam were both huge time-sinks.