The AI Misinformation Epidemic

Besides, I'm a solo founder with a meager burn rate so that by the time I have the traction the VCs want, I will be nicely profitable with plenty of cash for growth just from retained earnings.

Or, my back of the envelope arithmetic is that with common ad rates from ad networks, a $1000 server kept half busy 24 by 7 would generate $250,000 a month in revenue. For just one server, a cheap Internet connection, just one employee, that's a heck of a profit margin and plenty of cash for 10 more servers. Half fill those, say, in two spare bedrooms I have, with some window A/C units, some emergency power supplies and an emergency generator, and I will have annual revenue more than a VC seed or Series A equity check. And, "Look, Ma, 100% owner of just an LLC and no BoD!".

Sure, once I have the traction, VCs will call me, and then I will check and tell them about all the times I sent them e-mail they ignored and explain that my plane has already left the ground, has altitude, and is climbing quickly and it's too late to buy a ticket.

Sure, I typed in all the 25,000 programming language statements myself and implemented my math derivations in 100,000 lines of typing -- lots of in-line comments! The code is all in just Microsoft's Visual Basic .NET with ADO.NET for getting to SQL Server and ASP.NET for the Web pages with IIS for the Web server plus a little open source C called with Microsoft's platform invoke.

I wrote my own Web page session state store using two collection classes and some TCP/IP sockets with class de/serialization -- sure, could have used REDIS, but my code is so short and simple writing my own was likely easier. Besides, now I'm about to copy that code, rip out most of it, and get a log file server that I will like a lot better than what I'm using now from Microsoft.

Otherwise the code looks ready for at least first production, to, say, well past $250,000 a month in revenue. At IBM's Watson lab, I wrote AI code that shipped as a commercial IBM Program Product, and what I've written for my startup is more solid (got an IBM award for some of the code I wrote in an all night session -- let one of our programmers be done in the next afternoon instead of two weeks and got a MUCH nicer result for the customers -- trick was to do some things with entry variables to keep some run-time code on the stack of dynamic descendancy).

My project needs data, and I have a lot but need to get more.

Curiously, there's some good news: My development computer was crashing about five times a day, apparently a hardware problem, maybe on the motherboard. I did mud wrestling with it and, then, shopping for parts for a new computer. But my old computer now, for no good reason, is no longer crashing! The computer has plenty of free disk space for some more data. So, for now I get to set aside all the system management mud wrestling of getting a new computer and getting all the software moved to it and running, can review the most critical parts of my code a third time, write the log server, write some code to make working with SQL Server easier, get some more data, and do an alpha test, a beta test, and get some publicity. Maybe I will even go live with my development computer, get some revenue, and get a really nice first server.

The problem is important; the math is solid; the code is solid; I suspect a lot, at least enough, people will like the results (it's intended to please essentially everyone on the Internet), etc., but no VC in the country wants anything to do with my work now. Nothing. Zip, zilch, zero.

Lesson: To VCs, nothing but nothing matters but traction.

So, the flip side of that lesson is an

Opportunity: Be a solo founder where the traction the VCs want is enough for profitability and plenty of cash for organic growth.

The founder of Plenty of Fish was a solo founder who eventually sold out for $500+ million. Some old remarks by A16Z confirm the possibility of a one engineer unicorn -- at

http://a16z.com/2014/07/30/the-happy-demise-of-the-10x-engin...

with in part

"This is the new normal: fewer engineers and dollars to ship code to more users than ever before. The potential impact of the lone software engineer is soaring. How long before we have a billion-dollar acquisition offer for a one-engineer startup? "

I agree!

I saw the need -- obvious enough. I cooked up a new solution with a new UI, UX, and some new data to be just what some new math I derived needed. Easy enough -- I've worked harder single exercises in Rudin, Principles of Mathematical Analysis. The work was easier than my Master's paper, my Ph.D., and any of the papers I've published. I wrote the code for the math and checked it various ways including with programming those calculations again in another language to check. I designed the data base tables, the Web pages, and wrote the code for the Web pages. Then the code for the session state store. Then I got sick, then got well, then my computer got sick and got well, now I'm back to progress. So far, being a solo founder, nothing particularly difficult. No reason to need a co-founder.

But there are groups who are combining cutting edge neural network research with AGI and seriously trying to build general intelligence. Researchers who are working on narrow AI tasks and are familiar with this will continue to be incredulous right up to (and possibly beyond) the point where they see a system that seems intelligent to them.

Consider things like SOAR - that was explicitly an architecture for general intelligence. Whether it was any good or not is another question - but there was an active good-old-fashioned AGI field back then.

Mind you - one of the reasons I left AI research to do web stuff from '92/'93 onwards was I was pretty sure the symbolic approaches in favour back then wouldn't scale up and that there really was only intelligence on one side of a computer screen.