If he now interviews a new set of couples, does his predictions according to the developed model, and it _still_ gives high accuracy, then we can call it science.
The power of a good model lies not in how good it predicts what you know, but how it can be used to model unknown/new phenomena.
The author makes a couple of assumptions that I don't agree with.
The first is that success is measured by the amount of money you earn. For me it is much more rewarding to work on interesting and inspiring problems than having a huge income.
The second is that people that are good at hard science problems are good at solving the "stupid" problems. You need a different set of skills to solve the latter.
I disagree somewhat with your final statement. As long as we're talking about a creative job, it's not easy no matter what. As a scientist/researcher, you have to be every bit as creative as an artist or a musician, IMO.
This is somewhat true of engineers as well, but you can certainly get an engineering job where you don't need to exercise your creativity.
This used to be the case, but universities are getting better at this. Usually it's pretty easy to spin off a company, where the university retains some ownership. It's a win-win situation, really, a university is a great place to foster ideas, but not very good for bringing a product to market.
If you think of your supervisor like a boss, that's wrong, IMO. You should pursue your own ideas in your own way, and your supervisor should do just that, supervise you and guide you as best as he can, then let you do what you want with his advice.
Of course it's hard to build a team as a PhD, at best you can get some reasonably competent master students to do some work for you.
Well, I accept the term boss doesn't fit well. But still, in very competitive institutions you need to deal with lots of politics, do post-docs over post-docs to get a lab of your own or at least be more in control of the research. I think a startup that is financially successful is a good shortcut to that. You can become a CTO and go whatever direction you want.
It's a good roundup, but I think you missed the most compelling reason to do a PhD, at least it was for me. And that is that you get the opportunity to dig really deep into a problem for three year or more. And that is your only concern, no business crap, no customers, just you and your problem.
Also, the freedom of doing a PhD is great. Want to take two weeks to learn a new programming language? Sure! Want to take two weeks off and go surfing? No problem!
In fact, some of my most important work on my PhD was done during one of these "diversions".
Finally, some jobs require a PhD. You may want to consider if that's the kind of job you want. Hard to answer that before you finish, though. For me the answer was a resounding YES!
There is a saying in the scientific community..."if you want to make sure that your competitor doesn't get any work done for the next two years, give him your code!".
In all seriousness, though, a lot of labs view their in-house codes as a competitive advantage. You may not like it, but science has become commercialized, the labs/unis are all competing about the same grants, which creates a competitive enviroment.
The problem is that there is a whole industry created around Photoshop, all artists use it, it's taught in every college/course, and there are a million books written about it.
For a startup to compete against that momentum would be extremely difficult. There is also considerable competition from open source/free programs like GIMP and Paint.Net. Pixelmator seems like it's doing well, though.
I think the opportunities lie more in creating niche tools. Just look at all the tools created the last few years for photographers, the top one perhaps being Adobe Lightroom. It contains most of the image editing tools from PS that a photographer needs, but is built more around the photographer's workflow.
> For a startup to compete against that momentum would be extremely difficult.
I think it would be very difficult for a startup to make a replacement photoshop, but that shouldn't really be anyone's goal (photoshop already exists), and with actual novel ideas about how editing should be done, I think there are quite reasonable opportunities: it's always difficult to make quality software, but image editors and their market aren't inherently worse in this respect than email clients, say, or word processors, or many types of games, etc.
> considerable competition from open source/free programs like GIMP and Paint.Net
Really? I don't know anyone who uses these for serious work. I understand the GIMP gets used in particular niches, such as among free software ideologues, but I don't think competition from these would be a real problem for anyone trying to make a serious image editor.
> It contains most of the image editing tools from PS that a photographer needs
Not really. Most Lightroom users also use Photoshop, and Lightroom's goals as a product are completely different from Photoshop's (image organization, selection, comparison, rather than careful editing): they're complementary, not competitors.
I use The GIMP for serious work - production of advertising materials for print and for any non-vector elements of website designs. The quality demands (see eg http://photoshopdisasters.blogspot.com/ ) seem so low in the print media industry that most graphic designers probably aren't using CS4 to any great advantage.
The only times I've needed anything else have been because the printers say "we only accept a PSD" or "files must be from Corel 9" or whatever. They always take something else in the end. Actually it's more often I need to use AI files and Inkscape's not doing to bad on that.
Exact colour match (pantones, etc.) doesn't matter for me; my target market couldn't generally care less that the shade is ever so slightly out and in most print situations the colour is either off at print (newspapers) or off by the point of viewing (eg magazine in a rack that's faded for a month).
When you say “production of advertising materials” what do you mean? You work for an ad agency? Or you sometimes wear a “design hat” in addition to your other roles at your company? Because if you were spending 40 hours a week on design work, buying and learning and using Photoshop instead of the GIMP would pay for itself quite quickly.
I should have been clearer. I'm sure there exist people who use the GIMP, even though I don’t know them. I do not, however, think the GIMP provides “considerable competition” in the image editing space by any reasonable definition of “considerable”. I don’t have any solid numbers of my own, and really have no idea how I’d look for any, but if we just judge by, for instance, relative numbers of books offered for sale about each product, Photoshop has a simply crushing market-share advantage.
> quality demands seem so low in the print media industry
I don’t work in print media (I’m a political science student), but this seems like a pretty cheap shot. One could take similar shots at programmers, musicians, scientists, etc.
I don’t work in print media (I’m a political science student), but this seems like a pretty cheap shot.
It's my opinion. There seems to be a lot more typos and grammatical errors that should have been caught in proof stages, a lot more poor "it'll do" type photo-shopping than in the past. This may be because I've become more focussed and more observant with respect to print media given that I'm using it for inspiration for online work and to suggest what the zeitgeist might be.
Incidentally, as you note, this probably is true in other fields too.
The power of a good model lies not in how good it predicts what you know, but how it can be used to model unknown/new phenomena.