The Illustrated Guide to a PhD • Hacker News Board

Re: The Illustrated Guide to a PhD

#11

Post by ChrisLomont » Fri, Jul 31, 2020, 4:10 PM UTC

amznthrowaway5 wrote: post #8

SideQuark wrote: post #7

Earlier quoted context omitted.

>You don't need to get a phd to push the boundaries of knowledge, and often people who get phds aren't actually doing that. PhD is just a fake status indicator like an MBA. Pick any field. Find some measure of who pushes the boundaries the most in that field. Check after picking the metric, not before, what percent of those making the advances had a PhD, compared to the field overall. For example, in computer science…

My point was a PhD alone should not be taken as an indicator of skill, it is a very weak signal at best and is no substitute for good evidence of ability. Showing that most who are extremely successful have a PhD does not prove that having a PhD itself is a good indicator of ability, even a small amount of real evidence will override that information. This is why things like basic leetcode interviews are so important…

>My point was a PhD alone should not be taken as an indicator of skill

That's not what you wrote, though.

And, all else equal, I suspect a PhD is a good indicator of skill.

With no info, hiring for someone to solve math problems, which do you think will be better: math PhD or not math PhD?

Again, please provide some empirical evidence for your claims; not anecdotal feelings.

There's a reason starting salaries for PhDs are much higher than non for almost all positions, and it's not because hiring people are ignorant.

>no substitute for good evidence of ability

It's pretty hard to impossible to get one from a decent school without well above average ability.

>This is why things like basic leetcode interviews are so important, "phds" will often fail these.

Do they fail them more than non-phds? By how much? Again, please provide evidence. I'd suspect non-phds do much worse on average than phds.

I know for a fact PhDs score far higher on Kaggle competitions, for example. PhDs are over-represented in the winners, and if you look into the winners, a lot more without PhDs are PhD students. Go ahead an look there to check for yourself.

Heck, Kaggle even has a data mining degree vs pay dataset, and guess what? Check for yourself https://www.kaggle.com/salmanq/do-phds-earn-more

I'm pretty sure you're running on selection bias, not quantified measurements.

Re: The Illustrated Guide to a PhD

#12

Post by amznthrowaway5 » Fri, Jul 31, 2020, 5:36 PM UTC

ChrisLomont wrote: post #11

amznthrowaway5 wrote: post #8

Earlier quoted context omitted.

My point was a PhD alone should not be taken as an indicator of skill, it is a very weak signal at best and is no substitute for good evidence of ability. Showing that most who are extremely successful have a PhD does not prove that having a PhD itself is a good indicator of ability, even a small amount of real evidence will override that information. This is why things like basic leetcode interviews are so important…

>My point was a PhD alone should not be taken as an indicator of skill That's not what you wrote, though. And, all else equal, I suspect a PhD is a good indicator of skill. With no info, hiring for someone to solve math problems, which do you think will be better: math PhD or not math PhD? Again, please provide some empirical evidence for your claims; not anecdotal feelings. There's a reason starting salaries for PhD…

Well my point was meant to be that it is not a good indicator. I don't know if they fail as much as non-phds, but there are just so many better signals than the PhD itself to go off of, a PhD by itself is not impressive.

And it's well known that if you are smart and start working out of undergrad instead of getting a phd you usually end up significantly ahead in terms of earnings and on the corporate ladder.

Re: The Illustrated Guide to a PhD

#13

Post by amznthrowaway5 » Fri, Jul 31, 2020, 5:38 PM UTC

shard wrote: post #10

amznthrowaway5 wrote: post #4

Earlier quoted context omitted.

> But it is important to remember what you are doing. In a PhD you should be pushing the boundary of knowledge. As a programmer in a company, you should be solving problems to generate value for your customer and your company. You don't need to get a phd to push the boundaries of knowledge, and often people who get phds aren't actually doing that. PhD is just a fake status indicator like an MBA. A lot of work that ge…

The parent is talking about intent, not result. The work a PhD does is meant to push the boundary of knowledge, not that a PhD is required to push the boundary of knowledge. Same for the work of a programmer in a company, it's meant to generate value for the customer and company, not that it's the only way to generate value. A PhD works as a status indicator is valid in the sense that a Github account with lots of ac…

Lots of people get these degrees simply for the status aspect. I take meaningful Github OSS contributions much more seriously than PhDs, given all the garbage I have seen in academia. I don't understand how that can be called anecdotal when you can pick up any scientific journal and see that most of the papers are usually trash.

Re: The Illustrated Guide to a PhD

#14

Post by ChrisLomont » Fri, Jul 31, 2020, 6:10 PM UTC

amznthrowaway5 wrote: post #12

ChrisLomont wrote: post #11

Earlier quoted context omitted.

>My point was a PhD alone should not be taken as an indicator of skill That's not what you wrote, though. And, all else equal, I suspect a PhD is a good indicator of skill. With no info, hiring for someone to solve math problems, which do you think will be better: math PhD or not math PhD? Again, please provide some empirical evidence for your claims; not anecdotal feelings. There's a reason starting salaries for PhD…

Well my point was meant to be that it is not a good indicator. I don't know if they fail as much as non-phds, but there are just so many better signals than the PhD itself to go off of, a PhD by itself is not impressive. And it's well known that if you are smart and start working out of undergrad instead of getting a phd you usually end up significantly ahead in terms of earnings and on the corporate ladder.

> I don't know if they fail as much as non-phds,

Yet you claim a PhD is not a signal. Weird. I'd have suspected you would have some empirical, not anecdotal or selection bias, evidence that was demonstrable....

>And it's well known that if you are smart and start working out of undergrad instead of getting a phd you usually end up significantly ahead in terms of earnings and on the corporate ladder.

Citation? Also now you're moving the goalposts. It's especially funny since you didn't like that I pointed out how many Turing Awardees had PhDs, to which you complained. Now you see fit to sub-select your original claims. :)

Well known? Then you should be able to provide evidence, since it must be so commonly demonstrated.

I claim it's not well known, because it's not true.

I've worked with lots of PhDs, and lots of non-PhDs, and the PhDs by earn more than the non, including owning the companies.

Also, look over the boards of companies, especially tech, and see again if PhDs are over or under represented. Hint: they're over-represented.

So you are simply incorrect in this. Demonstrably so.

You make a lot of popular claims but have provided no empirical evidence. I at least provided two demonstrable cases where you're wrong.

I noticed I've offered many times for you to provide actual evidence. None so far presented as expected.