Assuming a 20 day work month and a 10 hour workday that will give you half an hour a month to work on each of your 400 projects. And assuming (optimistically) that you will spend 2 weeks doing each project it will take 15 years to get to 400.
And that's not counting overhead such as accounting, etc. for all of your projects.
I like the idea of the post, but maybe he took it a bit too far...
That's the point - you don't start a new project every 2 weeks. You start off with an idea easily adaptable to various scenarios. Then adapt each time, target new people, and observe if the variation makes you required $1 a day. When you have nichified a codebase that nothing more can be squeezed of it, you start on the next.
I think the idea is that each project gets an initial small period of launch work, and then after that they're mostly self-sustaining for their economic lifetime. "Optimize for time." In that case, half hour/month may be enough.
The 1$/day and 400/month figures are just for illustration. The idea is to crank on multiple simple things that share characteristics. As you optimize for time, you're also optimizing for "crank," getting better at cranking out simple things.
And that's not counting overhead such as accounting, etc. for all of your projects.
I like the idea of the post, but maybe he took it a bit too far...