The Business of Software

Hello-

    In Joel's article entitled "Converting Capital into Software that Works" ( http://www.joelonsoftware.com/articles/fog0000000074.html ), Joel makes the statement that "anything really interesting or hard to copy won't even get funded".  Unfortunately, he didn't answer the immediate question that came to mind: "like what?"  Does anyone have any examples of software that falls into this category?

Regards-
    Eric

Google's engine is interesting and hard to copy. I don't know that they had problems getting funding though. There were some problems and they got less than they should of since the wall street cats were playing media games to drive the price of the IPO down.

The next generation Intel chips are also interesting and hard to copy. I guess they are self funded though.

I'm not going to answer your question, but if you put yourself in the place of the investor, you can answer your question yourself.

If it's hard to copy, you're unlikely to get funding because you won't be able to benefit from shared marketing.  Your company will be the only one marketing WidgetX to the world.  That's a tough row to hoe. 

If it's interesting, it means it's too difficult for the general public to adopt.  People like TVs, lawnmowers, and email.  They really don't care how many bits of encryption their bank transactions have as long as the money's there when they need to withdraw for their next grocery purchase.  Mass adoption means mass profits which requires simplicity which means programming the mundane boring stuff every programmer loathes.

Think of things that require significant research and know-how. You're project is based on having some of the most talented experts developing the technology. That doesn't mean compitent and capable, but truly outstanding people in the field. It has to be so hard to do, even great engineers will get stuck.

Guess what? That basically guarrenties that it has to be a niche product. You need to pay your team what they're worth (which is a lot), cover the high R&D costs, and make reasonable profits. Only customers like the U.S. DOD will pay for such products, because it just won't be economical for other companies.

Imagine you invested a teleporter that can safely transport people and cargo. The R&D was high, and the specialty hardware doesn't come cheap. Do you really think FedEX will buy it? How many of their customers will pay 100x more for immediate package delivery? Would you pay 100x more that a 15 hour flight?

At the very best, in 10 years the hardware and knowledge will build so you can make cheaper versions. By then, competitors will appear copying your technology. They might even have a snazzier business plan and their lack of debt makes smaller ventures (that you ignore) attractive. You get beat at your own game, despite doing all the hard work.

So why should an investor risk the money on a project with low ROI?

require significant research and know-how

google, pentium

You're project is based on having some of the most talented experts developing the technology.

google, pentium

That doesn't mean compitent and capable, but truly outstanding people in the field. It has to be so hard to do, even great engineers will get stuck.

google, pentium

Guess what? That basically guarrenties that it has to be a niche product.

bzzt, wrong answer

Umm, so you believe that an VC group will invest in a startup designing a radically new microprocessor, as complex or more so than what Intel offers? Something that Intel can't copy, because its just too hard? I meant starting from scratch, not from over 20 years of being in the industry. Designing the a complex chip like the Pentium didn't occur in a dark room while the industry was elsewhere. When the playing field was level, Intel had many competitors with equivilant technology, and even today has quite a few.

Google came from academia - as a PhD research project. Many people would also argue that it is not hard-to-copy technology.

All of my statements came from a startup getting funded, not from an existing company funding a next-generation in their product line. That's not to say there aren't exceptions, but I was trying to follow the general gist of Joel's article. He was talking about markets where you get to keep your inventions for yourself.

Actually, years ago, before I did what I do now, I did work on a startup that made microprocessors and we did get VC funding. Eventually it all paid off - the company no longer exists since it was bought out by an established chip company that needed the patents. The VCs did well in the transaction.

Don't forget that interesting == patentable and patentable == hard to copy.

Okay, since Joel had thrown out patents as protection, I was trying to think of radically hard to copy ideas. Most ideas, once implemented, are pretty straightforward to emulate. I was trying to come up with something that could answer his question, which is pretty tough.

"Umm, so you believe that an VC group will invest in a startup designing a radically new microprocessor, as complex or more so than what Intel offers? "

umm... this is actually the type of project VCs invest in on a regular basis. many (most?) VCs are old school silicon valley, that is the guys investing got rich at a semiconductor startup in the early 80s. thus they understand low-level and hardware stuff. there is an obvious exit strategy, too: sell company to intel.

regarding what joel had in mind, i'm not sure. interesting and hard to copy sounds like mathematica. maybe if you designed a bunch of algorithms that could take a bunch of genetic data, and show you visually what sort of organism would come out of it. mathematica for developmental biology. that would be interesting and hard to copy. and impossible to fund. (and impossible to do)

I concede that microprocessors aren't the best choice, but it wasn't an example I'd have used. It doesn't fit the definition of being so unique as to not have competitors. The startups aren't radical enough for what I had meant. Maybe one that moves away from traditional architectures towards one based on probabilities, would fall into that unique catagory. I think the mathmatica example is a pretty good one, though.

Google is still a counter example. There are tons others but I'm sticking to microprocessors and google as examples since they are the first two I thought of and I think its funny you are coming up with ways to say that google doesn't apply.

And microprocessors and google are both hard to copy, even if you toss out the idea of patents!

And just so you know, our microprocessor was more complex than anything intel had at the time, and was a radically different architecture. 1 million transistors designed and connected at the gate level all put together by two guys in two years with $100,000 (not including fabrication costs, which were born by a licensee).

"It doesn't fit the definition of being so unique as to not have competitors."

Your definition didn't say anything about not having competitors. Anything the least bit profitable is going to have competitors. Changing your definitions is a poor way to argue. Look, its obvious that we could come up with counterexamples all day long and you'd have explanations, clarifications and new definitions for each one. Your mind is made up. You know for a fact that interesting and hard to copy technology projects NEVER get VC funding. That is a cold hard fact and any one like me with counterexamples or reason is just plain wrong. So no need to continue here.

I think I took it too strictly, which is why I already said I concede. :)

OK, I apologize for my premature exasperation.