[rdtsc]

> 1. Don't start a startup for the sake of starting a startup.

Seen that first hand happen. Someone at old work was enamored with "startup world". Read too much HN perhaps... Then came up with an idea for a startup. Right off the bat, could tell it was a forced idea. It was like they sat down and given a task of "come with an idea of a startup in 5 minutes" and they came up with one. It just seemed, well ... artificial. You can almost tell how they went about it in their head "Ok maybe Uber, but like for dogs. So they can go to a park play chase...". It was that kind of thing. However, the amazing thing is it didn't matter! They convinced management to spend money on him and his startup. I believe they are still bankrolling him and his idea 3 years later, while everyone there looked at each other with a look disbelief.

One can argue the startup is good enough, if you can convince some investor to invest in it. You don't need customers, a good idea, profitability, a market, etc etc. You need a dumber investor than you, who will bankroll you and you are done. After that you can always claim you were a CEO of a startup for the rest of your life, and do talks and presentations about it, put it on your resume and so on. It just feels good, you are part of something cool and exciting.

[clay_to_n]

I took a technical entrepreneurship course, and the professor forced us to come up with startup ideas very early on, and then iterate and convince everyone to throw our original ideas out. After a couple months every group had startup ideas that were clearly much more sound than the originals.

"First startup" ideas are very easy to identify, and are the majority of what you see in college entrepreneurship. The close, but often not scalable problems in competitive markets: food delivery. Textbook re-selling. Better course scheduling. Some way to make profit off of nightlife and parties. And then the large, technically unfeasible ideas: Solve X with drones, 3d cameras, machine learning.

The startups that keep going are usually not in either of these camps, but often start firmly in one and move towards the center as far as complexity.

But your last point is also spot on - a campus food delivery startup from my university got into a good accelerator and raised some money. Seems to be doing well.

[hga]

Counterexample: FTP Software (https://en.wikipedia.org/wiki/FTP_Software). A group of friends decided they wanted to start a company, they realized that with some effort they could turn an IP mostly stack for PCs into a full featured TCP/IP stack and did quite well for a while. Then again, the messes discussed in "Fate" section could be see as suggestive.

Additional details on its fate: the company's sales function got so screwed up people started contacting the president to get a copy (and don't laugh too much, this put Xerox in dire straits and improving this was a if not the most important goal of Mark Hurd when he became HP's CEO). And I was less reliably told that at one point they completely lost their source code repository.

But the idea certainly wasn't forced, full TCP/IP on MS-DOS was a killer app at the time.