Slashdot Mirror
Ask Slashdot: Open Vs. Closed-Source For a Start-Up
atamagabakkaomae writes "Together with a friend, I am starting up a company in Japan that develops sensors used in motion capture. For these sensors we develop hardware and software. Part of the software development is an open-source toolkit called openMAT. We have some special purpose algorithms that we developed ourselves and that are better than our competitor's technology. I first wanted to publish everything open-source to spark interest in our company and to do development in collaboration with the community. My company partner disagreed and said that we will lose our technological advantage if we open-source it. So I eventually published only a part of the toolkit open-source and closed the most interesting code. How do you guys think that open-sourcing your code-base affects a company's business? Is it wrong for a small company to give away precious intellectual property like that or will it on the contrary help the development of the company?"
Tom Preston-Werner from GitHub recently posted his take on this question:
http://tom.preston-werner.com/2011/11/22/open-source-everything.html
The tl;dr version of it -- open source everything except what is intrinsic to your core business value. My personal take is that if you can't beat your competitors with a mostly open book, you won't beat them with a closed book either. Hire the best people you can find, be thoughtful about your product, and hope for a bit of luck.
For many of your customers, closed source (i.e., binary or restrictive source license) may simply not work, for example because they are at a university (and can't guarantee that the source code won't leak out), or because they need to run the software on specialized hardware that you can't provide binaries for. Your advantage may also not be as big as you think, so open sourcing the software may not matter much, and other people may provide you with useful input and improvements. So, I think you should seriously consider open sourcing the software. You could make it a dual license (GPL + proprietary).
The best choice would be if you could incorporate those algorithms into your hardware. Can you add a small DSP do the hardware? That doesn't just protect your code, it actually may also make your hardware easier to use (fewer software dependencies). On the other hand, that way, you won't get any improvement from the community.
I think the key here is the question of what is it you plan to sell. If you plan to sell the software.. opening up the source would probably be counter-productive. If you plan to sell a solution, of which the software is a part.. then, you might have some advantage.
Red hat, for instance, does not sell operating systems. They sell support. Indeed, most of the software they ship isn't even theirs, but by going open source, they have the license to ship it all together and support the whole package.
Can you be Even More Awesome?!
If it's a cost center, open source it. It may offset the cost slightly. If it's a profit center, hold it to your chest.
There is a little company called Red Hat, perhaps you have heard of them. Their competitors have had a distinct habit of taking their Free Software and adding a few pieces of proprietary code. These additions generally made the competition nicer to use than Red Hat, but for whatever reason the competitors never were able to gain any significant market share.
Caldera, SuSE, Novell, and most recently Oracle have all taken a crack at Red Hat using software that was largely based on Red Hat's own distribution. So far this strategy has produced nothing but failure.
Willow Garage open sourced the software they use to run their PR2 robot
I think Willow Garage almost had to because they were using lots of open source tools in the first place. ROS is based on playerstage, which is GPL, and a lot of the heavy computer vision stuff is OpenCv, which itself was originally open-sourced by Intel. And the deal with everyone using ROS had a lot to do with development shifting from playerstage to ROS 'cause they were similar but ROS was saner, so they became the standard in large part 'cause they improved on the existing open source standard rather then trying to create some kind of large scale shift in the community. Plus, Willow Garage is as much experimental lab as company, so I don't know if it works as a good case study 'cause it sort of has a weird mix of end goals.
Willow Garage also gained a lot of cred by taking over OpenCV from intel and actively maintaining it, which isn't something a fledgling company can do but is worth considering. They adopted the library 'cause it was critical to their business and considered something of a standard in the vision community, which meant a lot of people were already using it, so it was popular enough that maintaining it was seen as a good thing.
open source modern art: laser taggi
The question is: What are you selling? Hardware or software?
If the software is the product, then close it obviously. There's money to be had from support contracts, but that's more of a pathway for monetising an existing free software project than for setting up a new business.
If the hardware is the product, then open the software. In doing so you effectively recruit every university doing research in the field, since they will all have tweaks and improvements. They publish their research, along with the software used (copyleft is good for that) and you either modify your own default software, or add the code to a repository for special purpose software. Your code is continuously improved and supports an increasingly wide range of applications.
Your competitor can adapt the results to their product as well, of course, but first of all they've got to port it. Meanwhile the number of applications for your sensor with custom software from third parties is going to grow and grow...
Don't let THEM immanentize the Eschaton!
The important thing that OP pointed out is that the SW is his core product. There is no harm in open sourcing your collaterals, if you can't make money of them. These colaterals are normally just a cost center and putting the software out there may even reduce the maintenance effort.
The thing looks differently with your core product. The question here is in what business are you? If your are in the business of selling software, open source is clearly not a real option. Why will people buy software they can get free from others. If you are in the business of providing paid support (Canonical) or paid training (MySql AB) then open sourcing your software makes sense, since more people will use it. If you are in the business for selling hardware it may make sense to open source the software that works with the hardware. It depends where your "company secrets" really are.
Intel has an interesting approach with their thread building blocks; the provide the library under GNU GPL or a commercial license. This is ingenious, be cause the GNU GPL states that any derivative work is also GPL so it bars any closed source software. This may work for you.