If only one out of ten failed, that would clearly be a great strategy. I think you meant 10% succeed.

Even if that stat is true, it isn't useful. It doesn't mean your chances must be one in ten, and it doesn't mean that the overall rate can't or won't rise in the future.

Assuming you meant "9 out of 10 fail", and that the average failure takes a year, including recovery time at a real job to replenish your finances and find a new idea...

That means you've had 6 chances just by the time your colleagues have finished grad school. 0.9^6 = 0.53, so you have a 47% chance of being a multi-millionaire by the time your academic friends are even entering the rat race. Compare that to the article's "If you're 95th-percentile lucky you might get a post-doc position, then you have a 50-50 chance of getting a tenure track position." Your odds are about 10 times better with the tech startup.

Then consider that at a minimum, you should expect 2 years of postdoc and another 6 years before getting tenure. That's now 14 years of training before you can actually work on what you want. 14 startup attempts with a 90% failure rate gives you a roughly 80% chance of success. The startup path is starting to look mighty good...

and plus, having started and failed at 5 startups, you learn a lot! your chances for the 6th one will definitely be higher than 10%.

Those are awesome odds compared to the "standard" way, apparently.

They would be awesome, however they're backwards. 9 out of 10 are likely to fail, averaged across industries. Perhaps more.