Hacker Newsnew | comments | show | ask | jobs | submit | narkee's comments login

I don't know about other places, but on the highway in Michigan near construction sites they sometimes have signs saying something like "If you injure a worker, it's a $5000 fine and up to 25 years in jail".

Note the precedence of the $5000 fine before the 25 years of jail time. Clearly the Michigan DOT thinks that fines and jail time are necessary incentives to not kill or harm construction workers.


I'm sure they think it, but does that make it so?


FYI, that last line says "patents", not "patients".


The call for larger sample sizes isn't always appropriate. It can often lead to spurious inferences.

As Jacob Cohen (famous statistician) has said, "all null hypotheses, at least in the two tailed forms, are false".

That is, with nearly any hypothesis about differences between groups, given a large enough sample size, you're likely to find a significant difference.


i think the difference here is between the two different experiments (one with three rolls of the dice and one with a single roll) and not a difference between the groups. You could choose to use the exact same group of participants for both experiments.


I'm finishing my PhD, and starting a post-doc in the fall. I had wanted to break out into industry, because I didn't see myself in academia in the long term, but I didn't get anywhere with job applications, so here I am.

I have a few questions:

1. My work consumes most of my time and energy - how would one find time to work on side projects and build a portfolio, when the academic workload is so all-consuming

2. I felt like I had to apply for industry jobs in my niche, otherwise I would be competing against a much larger pool of general engineers/science graduates. Considering that I've spent most of the last 5-6 years focusing on work in a specific niche, how can this be used for leverage for more general technical positions?

Thanks for the interesting article!


1. My work consumes most of my time and energy - how would one find time to work on side projects and build a portfolio, when the academic workload is so all-consuming

Turn off the TV, close Ph.D. comics stop procrastinating. That's a fairly glib and slightly offensive answer, but it's the best I've got. There are no shortcuts. Getting yourself ready to work in industry will take time and effort.

I can, however, suggest that adopting a more rigid industry-style work schedule can help. In grad school, my time management skills sucked, and I'm pretty sure I wasn't alone in that.

One thing you can sometimes do to speed things up is to take pieces of your research and turn it into side projects. For instance, something I didn't do (but should have) was properly package/test my numpy shared memory library.

2. I felt like I had to apply for industry jobs in my niche, otherwise I would be competing against a much larger pool of general engineers/science graduates.

You overestimate the number of general engineers who are capable of quantitative work. When I said there is value being the guy in the room who understands regression and confidence intervals, I meant it.

The much larger pool of general engineers/scientists is the target audience of blog posts like this one: http://www.zedshaw.com/essays/programmer_stats.html

Do things like Naive Bayes, SVMs and Black Scholes seem straightforward to you? If so, you are a quant. Now you just need to become a developer.


"You overestimate the number of general engineers who are capable of quantitative work. When I said there is value being the guy in the room who understands regression and confidence intervals, I meant it."

I'd just like to second that. If you've learned enough math to be an expert in a scientific niche, you're almost certainly miles ahead of most developers you'll work with.


Heck, I just have a bachelors in math and I'm the "go-to" guy for all problems mathematical where I work. Nothing more than high school trig usually -- but I'm the only one around who can do it.


What is your field?

(Because if you're a theoretical chemist trying to become an industrial chemist, the advice below is probably useless. But since you refer to "side projects and a portfolio", I'm going to assume for now that you must be some kind of programmer.)


First, a tiny story: At the corresponding time in my own career, I turned in my thesis, turned down a couple of jobs in my old field, moved into my parents' spare room, and spent three or four months teaching myself web development from online materials and a handful of books. (SICP, Learning Emacs, Introduction to TCP/IP, SQL for Smarties… the usual suspects.) Then I got a job. (It was the peak of the 1990s bubble, a good time to get a programming job. But, then, today is also a good time to get a programming job.)

The right way to save time on your academic workload is to stop doing academic work. Do you see yourself in academia in the long term? No. So why are you working on a postdoc? Who ordered you to get a postdoc? You did. Who is making you spend time and energy on that postdoc? You are. Write yourself a resignation letter, give yourself two weeks' notice, and quit.

The right way to escape a niche is to leave. You don't need to expunge the niche from your permanent record or anything - it's nothing to be ashamed of, it deserves a nice spot on your resume and is good for years of future anecdotes and impromptu lectures. But if the niche doesn't get you paying work, it may be time to let it rest for a while.

The right way to get any technical position is to demonstrate that you've done and enjoyed the kind of work that the position will ask you to do. If you want to build web sites, build a web site or two. If you want to build iOS applications, build an iOS application.

These things will not be sexy to an academic audience. You have to deal with that.

Don't worry about immersing yourself in the "larger pool". The reason there's a large pool of people doing X is: X is where the money is. And X is where the money is because, no matter how many people are doing X, there always seems to be more work to do than there are talented people to do it. Programming is in an expansionary phase, and there's a lot to be done.


Concerning the workload: As far as I'm concerned, the academic workload is definitely not all-consuming. From watching my colleagues, I came to the conclusion that everybody chooses their own workload. Some of my coworkers come at 9AM and leave at 4PM; others think they absolutely must finish that experiment and stay until 10PM. Some people think they must immediately rush to the lab if their supervisor sends them an email on Saturday afternoon; others just ignore the email until Monday morning.

I have never heard of anybody getting into trouble for turning their computer off on the weekend, or working normal hours; I think much of the pressure is self-inflicted.

Personally, I currently work in academia (doing my PhD), and I run a profitable business on the side (selling my own software). Sometimes this is a bit stressful (when you have to prepare a talk and keep getting emails from customers because of a nasty bug you introduced in the last release), but most of the time this works out just fine.


I think much of the pressure is self-inflicted.

This is true, but that doesn't make it less stressful. It's self-inflicted in the sense that the pressure running a startup is self-inflicted: you are competing with a bunch of other highly skilled individuals for a fixed slice of cake and if you want to be one of the successful ones you need to out-compete the others. The fact that no one forces you to work isn't really relevant.


Seconded. Some of the most successful academics I know keep regular 9-5 hours and leave their work at the office, have spouses and children, etc. I've noticed a lot of them are fantastic at time-boxing and that some of my less successful colleagues aren't so much, and the freedom the academic schedule gives them ends up being more of a hindrance than a help.


Are those tenured professors that have their grad studends and post-docs do all the work for them ? :)


Haha, nope, lots of PhD students and junior profs that have their act together.


As a researcher who just accepted a private job, I do have some experience here. My advice is to make use of the fact that academic jobs usually have unsurpassed autonomy and flexibility. If you aren't looking to get a faculty job, think of the postdoc as a relatively low-stress way to prepare for the transition out of academia. Not to say you should blow your job off, but make use of the opportunity to learn things that increase your cross section for industry and are still relevant in your research (even if it would not be the optimal thing to do if you were looking to crank out papers for your career in academia). Most of the academic workload is self-imposed, in the sense that you worry about the rat race for getting the next job, but if you are not interested in continuing your academic career, that's a non-factor.


This accords with my experience. I have a lousy academic publication record, partly because I wasn't that bothered about getting the next post-doc (although I was never short of offers, strangely enough) and partly because I liked to build things that facilitated research, which was a gap in our field of research which nobody seemed to have the time to fill. That isn't publication material, but it was a lot of fun and allowed me to investigate all sorts of interesting avenues.


I felt like I had to apply for industry jobs in my niche

It's your job to apply for positions, and their job to filter the applicants. Don't think you have to do their job for them.


What an amazingly complex game it would be if he decided to include relativistic effects of time-dilation.


I don't understand that distinction.

In selecting a typeface, you're also necessarily selecting the font, so why make the distinction?

From an artistic and technical point of view I get it, but from a user perspective the two are not separable quantities.


> In selecting a typeface, you're also necessarily selecting the font

No, the reverse of that. A typeface of the overall look, a font specifies the size and things like bold, italic.

So what is commonly called the font is really the typeface, and what is called the size is actually the font.

Back when fonts of a specific size were physical thing this mattered a lot - you only had a limited number of fonts for each typeface.


Typeface plus size plus bold/italic/etc is the font.


I think research funding in general could take some cues from the Y Combinator funding model, in the sense that if you give smart people money and time, they'll discover/create amazing things.

You would think the value and return you get from investment in science would be apparent to anyone living in the modern age.


Is that supposed to be a trick question?

I can't tell if the answer is n(n-1)/2 or just n(n-1), because I'm unsure whether you would be looking to count a "bidirectional TCP connection" as 1 or 2 TCP connections...


When I give this question I always make sure it's clear that I mean the "/2" case by saying something like "so there's only connection between each pair of computers" or "so you only count one connection between each pair computers".

Anyways, the point is to weed out people who say nonsense like 10^9 and 9^10 and 10! and whatever else pops into their head. In cases when people give these incorrect answers, it's always clear that they're just guessing, and when I tell them it's wrong they stress out and that's that. They refuse to even answer the n=3 case, because they've already given up.

The reason we found this a good filter is because people who are unwilling to think a little to give the right answer, are usually unimpressive in other parts of the interview, too. Ie. they give stock answers, and if we ask more or would like to drill down, they give up. That's our limited experience.


Yeah, I wasn't sure the way the question was posed either if it was n(n-1)/2 or n(n-1).

The bidirectional piece needs to be clarified in the way the question is posed.


Don't forget battery life.


Thanks for the explanation, closures have been a little tricky for me to wrap my head around.

I'm confused a little with regards to the closure closing over the "value" of free variables, or the reference.

For example, in python:

    i = 1
    def f():
        return i

    i = 2
    def g():
        return i
Both f() and g() return 2, even though when f() is defined, shouldn't i=1 be closed over in this case? That's what I find confusing here, in the sense that the state of the environment changes outside the closure affect things inside the closure.


This is a property of how javascript and python do closures. In the above code i refers to the same piece of memory in every case. So f and g are closed over i. The reference is preserved but the value is not.

This behavior is the motivation behind the do keyword in CoffeeScript.

Here's more on this:




Applications are open for YC Winter 2016

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