Diversity means you don't get trapped in an echo chamber.
Seems like you are using very selective evidence there, or selectively attributing only unsuccessful products to lack of diversity.
It's more of "a person whose unique experience, drive and motivation can teach you something and who's pleasant and educational to be around in circumstances outside of employment contract".
People who come from conservative and religious societies will always be pressured by friends to not make new friends who are of different ethnicities.
Thiel is basically ensuring each company Will become a little Island of segregation and this is what I have seen in the bay area: companies that are 90% Indian, 90% Chinese, or 90% white guys.
I think the effect you're seeing (where many Asians, such as Chinese, Korean, or Indian) are more isolated has far less to do with idealogical beliefs, than it does with multiple factors, such as being immigrants in a country that has typically has had a harder time identifying with their culture as well as a large trend of pushing towards higher education, especially in demanding fields such as medicine and engineering.
But I digress :)
If not, you may be bad at leadership regardless.
It seems that him saying that you try to hire people you could become friends with is opaque. How does he determine someone he would be good friends with? Do they have to share the same political views and hobbies? Without more info, it is hard to not view this as: Hire people Peter Thiel could be friends with.
The main advantage of hiring people you would be friends with is that at least you feel comfortable with them even if they're not an optimal hire. It is almost like admitting you don't have a solution to finding good workers, so if you get a bad hire, at least you get along with them.
This helps reduce unnecessary conflict when the only thing you're focused on is growth. A prime example is why early startup employees come from the same school as the founders.
I think later on, bigger companies focus on talent and culture equally (for the most part).
I haven't seen stats on startup hiring, but I read this about investment, which in my opinion stands a good chance of having a similar effect:
http://hbswk.hbs.edu/item/7069.html
"The probability of a successful exit outcome decreases by 18 percent if two venture capitalists who previously worked at the same company partner up in the syndication.
The likelihood of success drops by 22 percent if co-investors attended the same undergraduate school.
The negative effect of shared affinity is even stronger when it relates to ethnicity: Collaborating with someone from the same ethnic minority group comes at the expense of a 25 percent reduction in performance."
Big assumptions, but those guys are in the type of position where that's likely a reliable expectation.
Keep in mind that Mark only briefly "worked for" anybody in high school and college, and most of these were contracting positions.
In a way, the only example of people he wouldn't "work for" that we know about is the Winklevosses after he decided to "stop working" on "their idea"
1. you can try and hedge yourself against employees doing bad things. Like quitting for the competition or passing information. In this model, employees are paid mercenaries and none are crucial to the company. This is favored by corporations as it is good risk management.
2. you only hire people whose values align with yours, and where one strong shared value is sticking together.
The prisoner's dilemma is a good framework to see why either can work.
1. assumes defect/defect, which results in poor but positive outcomes. Corporations are inefficient, but they are profitable. You keep good people by paying them more than the competition, not by building trust and values-based friendship.
2. is cooperate/cooperate, which results in large positive outcomes for as long as you can maintain it, but the first breach of trust (the first instance or even possibility of a defect) results in a catastrophic negative outcome. For example, if you're having a few rough months and your CTO - who built 80% of the site and is the only one who understands the data model - leaves for a better paid job.
This is part of why it is so rare for more efficient upstarts to win against large corporations in their own market - by the time they get big enough, the startup looks too much like a corporation to maintain its cooperate/cooperate environment.
Where things get interesting is Andreesen's "Software Eating the World" essay. Some companies have recently surfaced (Instagram, Whatsapp, most hedge funds) where a small number of people highly trusting each other have managed to scale the outsize outcome to corporation levels.
And I think this is what Thiel is trying to say (or at least what he makes me think about): if your co-founders are practically "family", you might be more likely to go down that path. Whereas Zuckerberg prefers playing it safe, and assuming people will defect and scaling his company accordingly - especially since there is no shortage of capital to keep the best.
It's probably not fair to put these side-by-side due to context, but Peter's approach resonates with me now. Here's how I reconcile this: Peter's "friends" concept is about the long-term effect on yourself (and your team around you). Mark's "work-for" concept is about putting the very best people around you to build and operate a company. They could certainly co-exist, depending on your personality.
Mark's approach (for me) makes great sense on paper, but Peter's approach comes with the value of been-there-done-that. Both have their place, but over the long haul, I can really appreciate what Peter is talking about.