4 ms·

The author's view of PhD programs seems unrealistically positive. Maybe he means a CS PhD, but then the bit about changing the world makes even less sense. Sta

The author's view of PhD programs seems unrealistically positive. Maybe he means a CS PhD, but then the bit about changing the world makes even less sense. Startups are for pragmatic (as opposed to idealistic) self-starters. I think most PhD students identify with "I want to think deeply and learn a lot about one thing" much more than they identify with "I am a self-starter who loves to work in an unstructured environment."

burntsushi 13y ago
> I think most PhD students identify with "I want to think deeply and learn a lot about one thing" much more than they identify with "I am a self-starter who loves to work in an unstructured environment." Chiming in as a counter-example! Having an unstructured environment with few constraints is incredibly rewarding. If you pressed me, it might actually be my favorite part about being in academia. (My experience with other PhD students agrees with your generalization, though.)
- tracker1 13y ago
  I think it depends on what PhD program you are looking at... the people you have to deal with in say a Psychology program are a bit different than what you would see in a CS program. The culture can be very different, and the expectations of conformity can be even more so.
  - burntsushi 13y ago
    Oh, definitely. I should have added that "CS" qualifier in there.
- p4bl0 13y ago
  I completely agree with you. The few-constraints-and-unstructured-environment is one of the main reason I'm doing a PhD. I wrote about it two years ago [1] while I was questioning myself about PhD vs. startup. [1] http://shebang.ws/startup-or-research.html http://shebang.ws/startup-or-research.html
tensor 13y ago
Another counter-example. I found my PhD both more and less constraining than a startup, at the same time. For reference, I obtained a PhD, and I am now doing a startup. There are things that a startup can never replace. Most notably, access to domain experts whose goal is obtaining knowledge. In startup land, you get massive amounts of anecdote, but little knowledge. Almost all of that anecdote is geared towards earning money to the exclusion of all else. In academia, you get essentially the opposite. The reason I say that academia can be more constraining is that often you end up focussing on a single problem to the exclusion of all else. This can, but certainly does not always, harm your breadth of knowledge. In transition to a startup, the biggest difference to me was dispensing with "publish or perish." I've reproduced and experimented with more existing work than I ever did during my PhD. That's mostly because reproducing existing work doesn't earn you anything in academia, despite being a cornerstone of science. You spend as much time, and only as much time, reading background work as is necessary to do something new. Without having to publish, I've been free to really expand my breadth and understand which methods work and which don't in my new domain. All that said, what I'm doing in a startup is largely still applied research. This is a whole different beast than I see in most startups. Their goals appear to be chasing populism and money. The antithesis of academia.
- montecarl 13y ago
  Interestingly, I have a counter-example to one of your points in your counter-example. In my graduate work I have spent a great deal of time reproducing the work of others. This is because there have yet to be well defined benchmarks for the performance of various algorithms. In order to show that a new algorithm is superior to the current state of the art, it is necessary to implement the other algorithms. This has resulted in a open source program that implements various published algorithms, where the authors did not release any program to implement their schemes.

burntsushi 13y ago

> I think most PhD students identify with "I want to think deeply and learn a lot about one thing" much more than they identify with "I am a self-starter who loves to work in an unstructured environment." Chiming in as a counter-example! Having an unstructured environment with few constraints is incredibly rewarding. If you pressed me, it might actually be my favorite part about being in academia. (My experience with other PhD students agrees with your generalization, though.)

tracker1 13y ago
I think it depends on what PhD program you are looking at... the people you have to deal with in say a Psychology program are a bit different than what you would see in a CS program. The culture can be very different, and the expectations of conformity can be even more so.
- burntsushi 13y ago
  Oh, definitely. I should have added that "CS" qualifier in there.
p4bl0 13y ago
I completely agree with you. The few-constraints-and-unstructured-environment is one of the main reason I'm doing a PhD. I wrote about it two years ago [1] while I was questioning myself about PhD vs. startup. [1] http://shebang.ws/startup-or-research.html http://shebang.ws/startup-or-research.html

tensor 13y ago

Another counter-example. I found my PhD both more and less constraining than a startup, at the same time. For reference, I obtained a PhD, and I am now doing a startup. There are things that a startup can never replace. Most notably, access to domain experts whose goal is obtaining knowledge. In startup land, you get massive amounts of anecdote, but little knowledge. Almost all of that anecdote is geared towards earning money to the exclusion of all else. In academia, you get essentially the opposite. The reason I say that academia can be more constraining is that often you end up focussing on a single problem to the exclusion of all else. This can, but certainly does not always, harm your breadth of knowledge. In transition to a startup, the biggest difference to me was dispensing with "publish or perish." I've reproduced and experimented with more existing work than I ever did during my PhD. That's mostly because reproducing existing work doesn't earn you anything in academia, despite being a cornerstone of science. You spend as much time, and only as much time, reading background work as is necessary to do something new. Without having to publish, I've been free to really expand my breadth and understand which methods work and which don't in my new domain. All that said, what I'm doing in a startup is largely still applied research. This is a whole different beast than I see in most startups. Their goals appear to be chasing populism and money. The antithesis of academia.

montecarl 13y ago
Interestingly, I have a counter-example to one of your points in your counter-example. In my graduate work I have spent a great deal of time reproducing the work of others. This is because there have yet to be well defined benchmarks for the performance of various algorithms. In order to show that a new algorithm is superior to the current state of the art, it is necessary to implement the other algorithms. This has resulted in a open source program that implements various published algorithms, where the authors did not release any program to implement their schemes.