At any rate, do you believe his recurring costs, however you quantify them, are likely to be anywhere near $20k a month? It seems very likely that this is a profitable venture.
2018 costs:
48% to salaries and contractors (my wife and I are the only people on payroll, and draw $60k salaries for ourselves)
12.1% to operations (health insurance, renting a desk at WeWork, minimal travel, lot of SaaS subscriptions, misc equipment and other costs)
7.5% to PayPal fees (I really need to renegotiate this)
4.9% to AWS
4.2% to payroll taxes and other business administration
23% to profit (goes to saving up to quit my job last year, finish my book, and then dedicate ourselves to this full time)
Hope that's useful! Let me know if you have any other questions about this stuff and I'll do my best to be as transparent as I can without revealing every little detail.
Update: I should also mention that the salary we draw is in now way sufficient to actually support a family of 4 in Berkeley, so since quitting my tech job last October, we've been draining our savings and preparing to find a way to turn this little business into something that could sustain us longer term. I call it our flying rickshaw (the opposite of a rocketship). Outcome still uncertain.
And yes, PayPal fees are taking money from your children.
There's definitely stuff I can do to make AWS cheaper. Starting with just remembering to pre-pay for my instances ahead of time. One of my least favorite things to do is to log into the AWS console, so I try to do it as little as possible.
The CPU intensive part is due to autosave. When a lot of people are writing concurrently (even on the order of hundreds), that was usually what would take down the site in the old days. The site is weird in the sense that there are 10x more writes than reads, and then doing all of the linguistic analysis, badge granting, stats generating cause spikes when they happen as well. I'm @buster on Twitter if you want to continue the discussion there.