Because bitcoiners don't understand finance?

Anyway, some interesting thought.  I don't know if I have enough info/background to make numerical estimates, but I imagine this is what their per-customer margin looks like:

bitcoin dust
less: marginal cost of mining servers/generic work servers
less: marginal cost of advertising (acquiring customers)
less: marginal cost of promotions (avg lottery payout)

The dust revenue is just so, so low do you think that you'll be bumping against problems with just the marginal cost of having a constant connection to a server and getting new shares?  The cost of one network connection to one mining pool has got to be incredibly tiny but so is the revenue you're making from dust.  Bitcoin mining is both your most reliable source of income but also your most server expense-hungry and concurrent network connection-intensive choice.

Advertising: really tricky as you're still being trapped under the upper bound of your dust revenue.  How do you keep your average cost of acquiring a customer that low?  I guess raising tons of investor money in a formal investment series gives you a pass here for now.

Finally, I don't think non-bitcoin sources are going to be moneymakers for these guys.  People who need scientific computing aren't known for their deep pockets - they are funded by grants.  The web scraping thing makes it sound like you've already got a customer which is cool, but I don't see the sustainability for that business model (and I don't know who would pay for it in the first place).  People scrape the web so they can store, index it and query it at a later time.  Your army of smartphones is only going to give you a small bandwidth saving compared to downloading the pages yourself.  Does the cost of outsourcing it to a 3rd party with an elaborate system to manage its workers (aka higher per-scraper cost) going to work out financially?