Hacker News new | comments | show | ask | jobs | submit login
[dead]
on Feb 20, 2007 | hide | past | web | favorite



Disclaimer: I don't have the book. Grr! Linebreaks do not appear in the comment text! The premise of the book seems to be that you're going to learn about how to run a startup by reading about successful ones-- that is, you're going to look at a bunch of successful startups, and say Hmm, what makes for a successful startup? Another way to put it is that you're trying to train a classifier-- how do you tell whether a startup is going to succeed? Well, if it has a, b, and c qualities in x, y, and z amounts. However, a key part in training classifiers is having negative as well as positive examples. It seems that from reading *just* about successful startups, you could incorrectly infer a bunch of necessary qualities for successful startups, when in fact those are qualities of all startups, or those are just qualities of the sample taken, and focusing on them is bogus. Having no direct experience in the matter, I can't give good examples, but it seems like you could fall victim to a breakdancing chicken problem: (made up example) all the interviewed startups wrote software, so writing software is an important step towards having a successful startups. Well, no, but it just happened that the interviewer focused on software startups. If you had examples of failed startups, you'd see that there was just sampling bias to


There is some sampling bias in terms of what's discussed, but you can still get good information about the good moves and missteps each company has taken. I suspect that mistakes like "Didn't listen to users" are valid mistakes whether you're talking about Infogami disappearing or about Intel's Pentium II bug. That said, there's definitely sampling bias -- but these are more fun to read because it's hard for failed founders to be honest about what was their fault.


I've read it. I'm pretty clued up on the history of IT and the various backgrounds of those involved, but I still found that it had a lot of original, inspiring content. I would definitely recommend it.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: