There are some counter-examples (e.g. Franz, LispWorks), but I think it's best to just consider this sort of work a labor of love.
The game console dev kits are very expensive (at least they have been when I looked)
Is MS Visual Dev thingy free?
There is free version of probably every language out there. But, that doesn't mean people aren't making money selling that language. Cobol compilers come to mind.
The trick is to pick the right customers.
Perhaps Walter Bright can make money writing compilers, but that would be an exception. I think that you will find that compilers and other language tools are sold in conjunction with something else.
UC, MIT, Bell Labs, Google, Microsoft, Sun, ... the list goes on and on.
I say "very difficult" because search engines, for example, are non-rivalrous and non-excludable in practice. They are "public goods". The trick is, access to the attentions of and data generated by search engine users is excludable, and this is the good that Google actually sells. But any sort of non-excludable good from which your consumers cannot generate an excludable good you can sell for others would be unprofitable.
Programming languages (mentioned by another post) are another good example of a public good: they are non-rivalrous, and while they may be excludable as long as you sell the only compiler/interpreter/runtime and sue everyone else who writes a different compiler/interpreter/runtime, in practice no one will pay for a programming language anymore so you have to make it non-excludable for it to even exist in the outside world. Likewise, Google is probably excludable in the sense that they can set up a paywall before you use it, but in practice Google has chosen to make the search engine non-excludable and it's hard to see how a paywalled search engine would work (though I won't rule it out as a possibility).
I think people view for-pay information brokerage as unfair, unless it comes with service. If it's something they know costs the other party 0 marginal cost to provide, people start resenting them, which causes the paywalled service to hurt doubly hard when a free alternative comes about - even if they follow to free, people might stay away out of spite.
Alternatively, Matlab may succeed because it straddles a couple of different categories. It could be argued to be a language, an implementation of that language, a development environment... since it's very goal-oriented, it could even be seen as a more traditional piece of software with an unusually hard-core scripting language.
You can also make money by implementing a language that someone else created: java vendors e.g. BEA Systems; C++ vendors e.g. Borland; and SQL vendors.
There are a million easy ways to make $0.10 by spending $1.00
I think this is a very necessary startup, but one that's going to be very difficult to pull off well, and will require serious cash.
You probably won't profit if you make something people don't want.
But making something people want doesn't correlate nearly as well with whether or not your product is commercially viable.
But if you don't create any value, it's easy to capture it.
Few diseases are confined to only extremely poor areas. Medicines against malaria, for example, are a robust business. If you developed a good vaccine, you could make a lot of money selling it in Brazil, China, India, and other developing countries. The reason there is no vaccine isn't because there's no money to be made, but people have actually tried pretty hard and been unable to develop vaccines that aren't as risky as the disease itself.
The workaround since 2000-01 or so has been differential pricing or licensing for the Third World. But this took decades to get agreements for that to happen.
Here's a good page about the situation with HIV drugs. http://www.avert.org/generic.htm
"We've long since shed the notion of recouping costs through donations, and instead turned to ad revenue for covering operating expenditures. Our entry into the wild and not-so-wonderful world of advertising has been mixed. We've added more ad positions to the site over the years to offset rising overhead and been bounced around between so many ad networks and account managers that I'd be hard pressed to list them all. But we've succeeded—4chan is still here, after all."
If it's $10K without paying yourself, then you've got some work to do, but if you can live lean off your savings, you may still be able to improve the situation to the point where the business is sustainable or better.
If it's $10K before you pay your employees, find a new business, this one's not going to happen.
Investment Banks
Railways
Subprime lenders
So to answer your question: it is irrelevant. And I encourage you to separate "product" from "profit". Because if you want to make profits, you need to master business and marketing. Making products is for those programming guys. The great thing about HN, is that we think we can do both ... and we can! - Just remember that they are different hats for a reason.
=)
#1 is much tougher. If you're in a niche inside a niche inside a niche and you end up with 10 paying customers sending you $20/mo it's technically "working", but too small to be worth it.
#2 is easier. The situation Twitter is in. They have tons of users and now it's purely a matter of ingenuity to come up with a way to profit in a big way.
That got me thinking it may be something to exploit. Not so, with infrastructure costs swallowing up so much of the potential profits. People will just have to deal with all the channels they end up paying for. So, it's either the internet or bust if you don't want to play the cable game.
Making something people want enough to pay what it costs to create it : hard.
Hiphopgoblin had a bunch of people who proclaimed their love for the site, but I doubt any one of them would have paid cash for it. Maybe I could have cooked up some sort of new media advertising plan had I realized enough traffic, but I decided to abandon..
People's love is a revealed preference when they vote with their wallets.
