2 ms·

All sorts of stuff in there. Patrick stated that you can submit journal articles without being employed by a university or think tank. This is true. He also s

All sorts of stuff in there. Patrick stated that you can submit journal articles without being employed by a university or think tank. This is true. He also stated that, while it is perhaps thrilling to get a journal article published to an audience of hundreds, it is perhaps nearly as thrilling, plus one hell of a lot less work, to write essays on HN for an audience of tens of thousands. This is also true. Don't conflate the two statements. Scientific papers and blog posts are not the same thing, I don't think Patrick is claiming they are, and I'm actually quite doubtful that they'll ever be the same thing. (Anonymous peer review is important, blogs don't support it, and the online systems that do support it don't look much like blogs, they look like online scientific journals.) Now, on to other good stuff: academia... [is] really quite industry-like Indeed, it is very much like an industry. One with a tiny handful of customers, one of which is usually a major government. Often the government is the only customer. The difference between working for, say, a major military contractor, or a major pharmaceutical company, and working for a university is getting smaller every day. It's nice work so long as your customer is buying. Once your only customer stops buying, however, finding another one can be a real pain. Meanwhile, it's true that industry doesn't generally give a hoot about publishing new discoveries, often preferring to keep things secret. This is too bad. However, before we conclude that the only alternative is to become a professor, let me suggest that one could also look for a job where you can sell solutions to customers and keep control of your own IP, publishing it as you will. If you want to be told many stories about how this can work, go to an open source software conference.

_delirium 15y ago
let me suggest that one could also look for a job where you can sell solutions to customers and keep control of your own IP, publishing it as you will. If you want to be told many stories about how this can work, go to an open source software conference. I believe this is possible, but if your goal is to do novel scientific research and publish the results, it's exceptionally uncommon. I rely on all sorts of statistical and machine-learning techniques developed by other people, for example. It's relatively common to set up shop as an independent data-mining consultant, and you can fairly easily pay the bills that way. So you might think that they've produced some of the results I rely on. But the way the economics work out, independent consultants rarely spend much time on basic research. The actual techniques and advances, as a result, come almost exclusively from researchers in academia and industrial research labs. In fact, I can't actually think of any influential advances in machine learning from the past few decades developed by independent researchers. Heck, looking through the R packages I use, even the software seems to come mainly from a mixture of universities and places like Rand (biggest surprise in this mini-inventory: lots of stuff from Merck, who I wouldn't have expected to be open-source-friendly). Nonetheless, that's actually my medium-term goal, so I hope you're right. ;-)
- mechanical_fish 15y ago
  the way the economics work out, independent consultants rarely spend much time on basic research Am I wrong to read this as: "After forty hours a week spend on doing well-paid work for clients, it's hard to find another thirty hours to scour literature and write up results, while still having a life?" Because I'd be cautious in assuming that academics work much less than seventy hours a week. ;) It's true, though: Doing an actual stint in academia is the right way to get a lot of scientific publications out. But: It doesn't pay well. It's not the most effective way of keeping yourself alive, or creating clear and saleable value for others. That's why I call it a hobby. You are taking time away from more valuable pursuits to advance the field, teach people, and maybe even practice a little art. I'm sad to see that people think my use of the word hobby is so demeaning. I'm a big believer in hobbies. Here's my philosophy of the ideal life: You're born, you enjoy a series of hobbies (some of which may be deadly serious, like Steve Jobs' hobby of inventing the future), and then you die. I suppose I could look for a better word, but if the word doesn't express the concept of loads of fun it's not the right word. Apparently writing essays on HN is the hobby of the day for me...

_delirium 15y ago

let me suggest that one could also look for a job where you can sell solutions to customers and keep control of your own IP, publishing it as you will. If you want to be told many stories about how this can work, go to an open source software conference. I believe this is possible, but if your goal is to do novel scientific research and publish the results, it's exceptionally uncommon. I rely on all sorts of statistical and machine-learning techniques developed by other people, for example. It's relatively common to set up shop as an independent data-mining consultant, and you can fairly easily pay the bills that way. So you might think that they've produced some of the results I rely on. But the way the economics work out, independent consultants rarely spend much time on basic research. The actual techniques and advances, as a result, come almost exclusively from researchers in academia and industrial research labs. In fact, I can't actually think of any influential advances in machine learning from the past few decades developed by independent researchers. Heck, looking through the R packages I use, even the software seems to come mainly from a mixture of universities and places like Rand (biggest surprise in this mini-inventory: lots of stuff from Merck, who I wouldn't have expected to be open-source-friendly). Nonetheless, that's actually my medium-term goal, so I hope you're right. ;-)

mechanical_fish 15y ago
the way the economics work out, independent consultants rarely spend much time on basic research Am I wrong to read this as: "After forty hours a week spend on doing well-paid work for clients, it's hard to find another thirty hours to scour literature and write up results, while still having a life?" Because I'd be cautious in assuming that academics work much less than seventy hours a week. ;) It's true, though: Doing an actual stint in academia is the right way to get a lot of scientific publications out. But: It doesn't pay well. It's not the most effective way of keeping yourself alive, or creating clear and saleable value for others. That's why I call it a hobby. You are taking time away from more valuable pursuits to advance the field, teach people, and maybe even practice a little art. I'm sad to see that people think my use of the word hobby is so demeaning. I'm a big believer in hobbies. Here's my philosophy of the ideal life: You're born, you enjoy a series of hobbies (some of which may be deadly serious, like Steve Jobs' hobby of inventing the future), and then you die. I suppose I could look for a better word, but if the word doesn't express the concept of loads of fun it's not the right word. Apparently writing essays on HN is the hobby of the day for me...