It might be a weird idea, but after all, an open-source project makes something people want, it has leverage, and it can sure use some money.
So can anyone give a good reason why this will not work?
Simulation is not exactly sales automation, data capture, billing or inventory.
Your best bet is to reach out to federal safety and security agency, and sell them a heavily customized version of the tool (usually the problem du jour; few years ago was terrorism, now it's home foreclosure, job loses, etc.) you will be prepared to have a few PhDs on board to sell it to the governments and you might need to reach out to your local governments and municipalities.
(Yes, I know of SimPy, it's something different.)
What sort of game theory problems are you thinking about, I have a pretty solid background in algorithmics of that stuff, (which will transition from solid to novel research level soon :p, so i'd be curious as to what you think are settings where someone would like to do game theory modelling)
How is a general framework useful?
When you're coding a simulation, you spend about 10% of your time coding the actual algorithm, and 90% of your time coding other things. The purpose of my project is to do that 90% for you, and to do it thoroughly.
For example, saving the simulation data to a file. Or being able to rewind the simulation and change something in its world, and then running the simulation from that point and watching what that change will cause. Or having the machine calculate two different "possible futures" concurrently.
These are the basic features, and I know it doesn't sound like something very complicated, but if you think it's simple, I challenge you to find me a framework that allows to work comfortably with simulations.
What sort of game theory problems are you thinking about
Generally, whatever you want. I'm not a game theorist, but here's what I coded a year ago: A hundred computer players play 50 rounds of the prisoner's dillemma with each other, and they get paired randomly after every session. Every player had a certain strategy chosen from a pool of strategies. When a player lost a lot of points, the computer replaced it with a new player with a random strategy. It was very interesting and I saw some very cool behaviors. So you'd be able to do stuff like that on my program.
http://github.com/cool-RR/physicsthing/tree/master
Please keep in mind that it's a very young project. Run gui.py in that folder to see a demonstration. The code requires wxPython and that Windows Extensions thing by Mark Hammond.
The readme is a bit out-dated: When I wrote it I was mainly thinking about physics simulations, and today it's more general.
Come back there in a month and you'll see some serious stuff.
They are predominantly a service business, however, and provide custom software for a few very large clients (oil and gas industry, and one very large consumer products company). They've moved slowly in the direction of products rather than services, but I don't think they've ever quite made the break from being a custom software shop for (very profitable) niche markets.
I explained their business as a way of saying that while you can build a profitable business and make yourself rich on the kind of tools you're talking about (the founders of Enthought are doing very well for themselves and could probably sell the company tomorrow and retire multi-millionaires; they're also doing fantastic things for their Open Source community at the same time), you can't raise money in the valley for that kind of company. Until you figure out how to productize what you're doing, and make it possible to increase revenue faster than you increase head count, you aren't building a startup. You might be building a business, and that might be awesome, but valley investors pretty much only invest in startups.
They work with geological data to help oil companies build safe platforms (losing a platform to a natural gas deposit costs millions of dollars and occasionally human lives, for example, so the amount they're willing to spend to prevent it from happening is incredibly high), as well as profitable and efficient wells. They also work with fluid dynamics data for a consumer products company, to make sure your toothpaste comes out of the tube without splattering, etc. All seem like "simulations" to me. But I don't know. Visualization is a big part of what they do, as well, and SciPy has awesome realtime visualization capabilities (lots of folks also like MatPlotLib, which works nicely with SciPy).
Phusion offers a focused product - you install it, usually very easily, and you're done. It's hard to see a simple way to monetize it other than charge money for the product itself (thereby harming its market share). They might be able to sell support contracts or professional services, though.
On the other hand, if you're selling a service, seems to me you could benefit, or at least not lose anything, by opening your source. Would Twitter be doing worse if their software was open? They might actually have had better code by now, since so many coders use their product.
Also, you might build your product on top of an OSS codebase, using the accumulated code and domain knowledge as foundation. AFAIK that this is what Boxee is doing with XBMR. Also consider Wordpress - being open source helped it become a leading platform, later facilitating wordpress.com which has a pretty decent business model.
JBoss made it work because they were a pin-compatible replacement for something IBM was already charging a fortune for to large enterprises. But Phusion sells to the cool kids, not the suits, and the cool kids shop at thrift stores.
Hope I'm wrong about this.
We don't think so. Webmin is an Open Source project. Virtualmin, Inc. is a business. The business is new, and had pretty low recognition when we accepted YC funding. We're still not as well known as a company as we are as an Open Source project, but we are finding more and more of our customers have never used or heard of Webmin.
"neither the money nor the YC mentoring seem a good fit for them; they could have gotten funding from RedHat, IBM, Oracle or whoever else."
This makes assumptions about us and our company that aren't accurate.
We weren't looking for funding for our Open Source projects. We've funded them just fine on our own for 11 years; Jamie and I have lived off of our Webmin-related projects for most of our professional careers.
Now we are building a product startup that has a shared codebase and a large number of shared objectives to our Open Source projects. Building a business (a startup, and not a consulting business) is a very different process than building a successful Open Source project. We wanted some help with the former. YC provided it.
"i don't think the server automation tools like Webmin or cPanel see much love from hackers"
We'd rather you not group cPanel and Webmin together like that. They aren't even in the same market. Webmin is for system administrators, cPanel (and Virtualmin) is for hosting providers and their customers. Webmin happens to have been used by hosting providers a lot over the years, but it's a general purpose administration tool in use by large and small companies and individuals all over the world.
I won't make claims about how hackers feel about Webmin, but I will mention that Webmin is downloaded more than 2 million times per year. I can't think of who else would be downloading it from SourceForge. It's a UNIX system administrators tool. A very large percentage of UNIX system administrators are also hackers.
It also makes me a very sad panda to know that you think we do the things that cPanel does that makes hackers hate cPanel (and thus think those same hackers should dislike what we do). We don't generate config files from templates...we parse them and edit them. We respect comments, file order, and if we don't understand a directive (say, for an Apache module we don't know about) we won't break it. We work the way the OS does, respecting file locations and defaults (like Debian's sites-enabled/sites-available, or the per-interface iptables stuff), and go to great lengths to do so (have a look at the hundreds of per-OS config files found in Webmin for an understanding of just how great those lengths are). And, we provide copious documentation, over one thousand printed pages worth, for how things are working and how they map to the underlying system. All of these things are dramatically different from cPanel, and are designed from the perspective of "I want a tool that helps me do my job rather than forcing the world to revolve around it".
Virtualmin is also available in both an Open Source (GPL) and commercial version.
I'd be happy to answer any questions folks have about what we're doing (though I won't claim to be an expert, and I'm still figuring it out) and how well we're doing (well enough that I'm not at all dissatisfied with closing up shop on my previous service oriented business to start Virtualmin, Inc.).
You can have your closed-down code base where nobody knows your software, and you're just another hacker. You open source it, get users, and you're suddenly a software publisher; all it takes is for one university or corporation to contact you for help and you're free to ask for compensation. If your software was locked down, you wouldn't have had that lead to begin with.
However, the fact remains that going open source reduces one's ability to control one's work. Trademark protection means that you can't be out-competed by yourself.
The answer may be consulting but that's a difficult route b/c anyone can generate a profit from consulting. It's not unique. You need to figure out how to provide something that's better than everyone else. As Warren Buffet would say, you need a moat.
The problem with consulting is that it doesn't scale. You can only deliver it with trained consultants, who cost lots of money. Build a good product and you can sell it to 100 people as easily as you can to 10. That's why products are considered better money than consulting.
What I intended to say is that it's difficult to differentiate one consultancy from another. You have to do that based on clearly proven, previous performance. When you're starting a new company, it's easier to sell something that's new (like a service or product).
The reason is simple: a customer can try out an open source project for free.
That's kinda the point of Open Source as a marketing tool. If someone sees how awesome your Open Source software is, and they want more of that same brand of awesome (but to do something just a little bit different), they know exactly where to go.
I just talked about a software services business that I worked with a few years ago called Enthought, and it's deeply relevant to this threads interest. While at University the company founders built a set of Open Source Python (and Fortran and C++) components for scientific computing. They then went on to start a software services business with a little bit of money raised from family and friends. Their revenue is now in the millions (I can't be specific about how many millions, as I haven't known with any specificity for many years, and I've never been told that they would want me to talk about those numbers with specificity). Their business has very frequently first come from customer exposure to their (deeply awesome) Open Source software.
Tell us where you are. Maybe someone might have a lead.
(Drumroll...) Jerusalem, Israel.
When you do a startup you want to find a good niche that makes good money, and where competition won't start pouring in once you've proven that there is a market. So you need some kind of barrier, with YC startups it seems to be technology.
"Because making something people want is so much harder than making money from it, you should leave business models for later, just as you'd leave some trivial but messy feature for version 2. In version 1, solve the core problem. And the core problem in a startup is how to create wealth (= how much people want something x the number who want it), not how to convert that wealth into money."
Cheers
1) Dual licensing 2) Charge for commercial support 3) Sell hosted versions (might not apply to you) 4) Advertising?
For an open-source library, the right-to-fork and to use without paying licensing fees means many of the options used by other startups which defer revenue questions in favor of building an enthusiastic userbase are already off the table for you. You can't easily withhold your product unless fees are paid; you can't add advertising; you may not be in a position to sell excludable complementary offerings.
You could try a dual-licensing structure like Sleepycat BerkeleyDB -- free for other open-source projects, pay if you want to use in a closed-source application. However, that presupposes a large closed-source commercial market for the software your library-users make, requires some capability for closing big corporate sales, and it may take years of proven library quality before becoming the solid/safe choice for companies to embed in their own products.
And the unstated assumption of that interpretation is that it is impossible to make money without coercion.
Excludability ≠ coercion.
An advertising-supported site can exclude non-payers from using their advertising inventory. Money is made without coercion (and free/open-source software might be provided as an audience generator).
A consulting/services/support business can refuse to provide their service to non-clients. Money is made without coercion (and free/open-source software might be central to the delivery of services or expression of expertise).
A paid-subscription website can exclude non-subscribers. Money is made without coercion (and free/open-source software might be provided as a complement to the paise service).
The OP's proposal of angel-funding an open-source library presents a challenging case, unless some excludable complementary product can be offered. Only the dual-licensing strategy could vaguely be said to include coercion, if you believe enforcement of copyleft or copyright is coercive. And as I noted, that approach requires overcoming other market/sales/staying-power difficulties.
If you have other ideas for turning a open-source library into a money-making non-coercive business, please share -- especially if they somehow work without excludability. I'm very interested in this topic.
Your points about service businesses are well-taken, though I think this is a hard and uncertain road to turning a profit.
As for turning free software into a money-maker, I think not-for-profit is the stable state of a perfectly efficient software industry. That doesn't mean programmers couldn't be well paid, but it's likely nothing that would be terribly attractive to VC.