Semantic Compositions: Introducing Step 3

Sun	Mon	Tue	Wed	Thu	Fri	Sat
				1	2	3
4	5	6	7	8	9	10
11	12	13	14	15	16	17
18	19	20	21	22	23	24
25	26	27	28	29	30	31

As should be readily obvious, something has changed around here. In order to explain it, I'm going to slip out of the usual SC voice, and write in the first person for a change. And there's no better way to begin this explanation than with a story I've wanted to have a good reason to tell for a long time.

In 1996, after transferring from Harvey Mudd College to Claremont McKenna College, I had to pick a new major, as I had decided to give up on being an electrical engineer. I was having a very hard time deciding between two choices, which I spent the entire summer thinking about: linguistics...or music? I'm not a gifted enough saxophonist to have made a living as a performance major, so it would have been music theory, but truthfully, I wasn't sure if I had what it took to make a living as a composer, either. So after spending some time reading Akmajian et al.'s introductory linguistics textbook, as I mentioned once before, I made the decision that proved me to be the worst mercenary in human history: I decided to go into linguistics for the money.

If you didn't die laughing after reading that, it's safe to assume you weren't drinking anything at the time. More seriously, and I don't think I've ever claimed otherwise, I didn't want to be a professor -- as much as I admire many of the people I studied with and those who I've met through this blog, I always wanted to go build a product, and make a living by selling it. Back when I first joined SAIC, it looked like that opportunity would be the concept-based search engine that I helped develop. And speaking of SAIC, let me mention right now that I remain in a consulting relationship with them, and so I will continue my policy of not commenting on the company, except for what follows in this post.

What I will say regarding my time there is that the time that I started this blog (January of 2004) coincided almost exactly with the end of the artificial intelligence research group there that I had been part of. I was fortunate enough to be picked up by a modeling-and-simulation team that kept me from being unemployed right while Mrs. SC was in medical school, but one of the rough things about the consulting business is that your salary is tied very tightly to continuously bringing in new work. This meant that I only worked about half-time that first year on the blog, which gave me more time to read and write, albeit at the expense of a tight living situation. In June of 2005, though, I coauthored a proposal that brought in a lot more funding than I had ever been responsible for before, and suddenly I had all the full-time work I could ask for. Alas, that work had exactly no linguistic relevance, being a big database-migration project for the Department of Transportation (the project is a matter of public record, and you can read about it here if you'd like). The effects on my writing have been obvious, but more than anything, I found myself lamenting the fact that I wasn't in academia if I wasn't going to be building something.

While I have no wish to comment in any detail on my medical records, nor on any causal connections between them and my work, suffice it to say that the catalyst for change was a physical exam last year which led me to write several things on dieting, including my unbounded admiration for some work by Neil Whitman's brother, Glen. I gained a lot of weight working on the project, and while I've worked it off at this point, it forced me to reassess my priorities. Within days of being read the riot act by my doctor, I sat down to put together a presentation on a product I'd dreamed of building for years, talked to a lawyer about incorporating -- and then stopped.

An inane provision of the Internal Revenue Code imposes a certain minimum tax burden (about $800) on any corporation in its first year of business, whether or not they book any revenue. While this is not a meaningful barrier if you're planning on raising venture capital, it would have been a pointless expense since I expected to remain with SAIC at least through January. So I put the plans on hold, but continued to write code to test my ideas -- and I was pleased with what I found.

Due to the gratitude I felt -- and very much continue to feel -- to the coworkers who saved me from unemployment just a few years earlier, I ended up postponing my planned departure several more times in order to deal with various tasks at the office, not all of which were related to the project linked above. However, once I had decided that it was a matter of when, not if, I was leaving, I had to set a date, and the "countdown" post represented the public face of my giving formal notice.

I then attempted to drop some hints via post content about what was coming next, which I'll briefly review here. First came an item about uses of linguistics in stock trading, which has no small amount of relevance to what I'm trying to accomplish now. Then came a post decrying the poor service provided by both Verizon Wireless and Amtrak to me as a commuter, a hint that my patience had run out (I will no longer be commuting by train, nor will I be paying for Verizon's tethered modem service). After that, I wrote about one of many meetings I've been attending of the San Diego Software Industry Council, specifically in a context largely useful to executives at startups. And finally, I made a comment (too subtle, to be sure) about what I would do if I was starting a business, while writing about someone who did start the business we all wish we had. Then my laptop reached a point where I couldn't risk further damage to it by not having the defective case repaired, and that was it for dropping hints (and also for doing work on the new website).

So that brings me to what you see here. Step 3 Systems is a dream come true for me -- a chance to combine my interests in linguistics, finance, intelligence (the James Bond type, not the cognitive type, although I like that, too), marketing and statistical models. Three and a half years after starting a blog about linguistics, my work and my hobbies will be essentially the same thing, and that's a blessing almost too great to hope for. More than that, while starting a business is a big risk, I'm fortunate to have the support of Mrs. SC, who shares my vision and is the best wife and partner any man could hope to have. Finally, I hope it will provide me with an opportunity to eventually give back to the linguistics community, from which I've been happy to learn so much -- hopefully, the next time I'm at an LSA conference, it won't just be for fun, but for recruiting.

Of course, now that I'm in business for myself, that means my focus will change in other ways. I hope to share with you all some of the things I've learned, and will continue to discover, about how to get a business started -- getting incorporated, getting my intellectual property in order, writing a business plan. I definitely plan to write about the corporate culture I want to create -- it matters more than you could ever realize until you try to do it yourself -- and I plan to write about some of the tricks I learned to get the website up and running. I'll also write about the articles I've been reading from across linguistics, computer science, and finance, but I'll have to be very circumspect about what I'm actually doing myself, at least until patent applications are filed. I'll also have to be even more cautious than I already am regarding the criticisms leveled at certain companies; yesterday's competitor could be tomorrow's customer. If that makes Semantic Compositions a corporate tool, well, SC was one already. With all that said, I'm looking forward to living my response to Mark Liberman's LSA challenge, and I hope you'll all enjoy coming along for the ride.