If you think each business is a "random attempt", I hope you're never a founder. Modern businesses are highly non-random. It's true that investing in startups is in some sense gambling, but it's gambling on a highly filtered subset of all existing businesses, which is in turn a tiny fraction of all possible businesses.
The "where is the data" line is especially misleading here because no data just means no data. There are no high-n, double-blind studies validating the concept of technical debt. But there are also no studies demonstrating it's not a useful concept. And that's most of life: the decisions we have to make rarely can be solved by looking up journal studies. When that's the case, we have to apply other tools.