One of the many times I failed in business

In my business career, I’ve succeeded a time or two. And those are fun memories to relive. But I see no point in writing a self-ingratiating blog post about the singular glorious moment of victory that, in truth, followed a long series of defeats.

In fact, the moments of defeat are often the memories I most enjoy reliving — or perhaps, put better, most need to relive — because instead of reaffirming my ego, they teach me something. They are rememberances of lessons learned the hard way — and often, the only way.

The crazy start

I was in college writing a thesis on topic modeling. Topic modeling is a subject in machine learning concerned with discovering the set of “topics” discussed in a set of text snippets, without any human help.

At the time, there were excellent models for determining the “topics” present in text corpora given some amount of a priori information, like the number of expected topics in total. But the output wasn’t human readable — there was no “heading” for each topic.

My technology gave these “topics” discovered by complex machine learning techniques names. And this was powerful.

So I developed a web application that could ingest a bunch of text — for example, all the readings for one of our classes — and run it through the algorithm we used for topic models (latent Dirichlet allocation) and assigned the different topics names based on my special sauce algorithm. Then, it was displayed in a pretty, interactive graph-like interface (based on graphviz).

We thought this was really cool.

Setting up the server in my friend’s dorm room

At the time, it wasn’t so expensive to get a thin website with a pre-selected set of libs for your choosing at GoDaddy and the like. It was, however, both complicated and expensive to host a true Linux server, like the ones we were provided to use for school.

Problem was, if you were running something that seemed to be not school-related, you could get flagged by a sys admin. And that meant you got in trouble — with school. Nobody wanted that.

So I wrote a GoDaddy PHP site as our frontend that called out to a job server by posting work to a public address. The job server would pull from that address, and post back information. This server was essentially client-only, so it worked within the constraints of the firewalls of a dorm room.

So what did we do?

The obvious thing. We packed up a bunch of computer stuff in a Jeep and brought it to my buddy’s dorm room. We rigged it up and, via our incredibly slow but compliant-with-school-policy job queueing mechanism, started serving responses.

It felt like magic. Seeing our homework get digested by a computer and spit out the relevant pieces, with names we could understand, and a graphical display of the interconnctions. “Who wouldn’t want this,” we thought to ourselves.

Trying to sell it

So obviously the next step is to sell the software for a billion dollars, right? No worries! There was a conference in New York — conveniently a 5 hour drive from my college — only a week away, on the subject of text mining, and all the giants would be there!

But the tickets cost way too much. So we did the obvious thing.

We snuck in.

We found ourselves suddenly in front of reps from all the big tech companies. We pitched our idea, and actually got several meetings set up for later in the day with giants like Oracle and Google. But in the end, nobody bit in the way we had hoped — nobody wanted to bring us into their world.

And that was the only idea we had. So we basically gave up.

That was the stupid part.

That business could have succeeded. It could be huge today. But we gave up when we ran out of ideas — a very narrow set of ideas — on how to monetize the technology.

The failure

The reality is, it would likely have been a futile effort to compete in the information retrieval space — anything worth doing is usually dominated by Google and friends in some manner. But we’ll never know, because we didn’t try to do it ourselves when everybody else said no.

The lesson, on the other hand, is simple. When you have a great idea, and your first thought of how to turn it into something useful doesn’t go exactly as planned, you don’t just quit.

Yes, you may be getting indications that the market isn’t receptive — and Lean Startup wisdom would say to pivot and re-try. But there’s also the chance that your thought horizon for monetization opportunity isn’t far enough — you haven’t dreamt wide enough or big enough, perhaps, to understand the relevance of your own achievement.

Ralph Waldo Emerson tells us to “Trust thyself; every heart vibrates to that iron string.” Indeed, in all things, including entrepreneurial efforts, you should have trust in your convictions. You should search hard and thoroughly for a way to monetize an idea you think is great. You should stay true to your convictions to the extent it is sane to do so.

But then, as Emerson says in that same essay, “a foolish consistency is the hobgoblin of little minds,” and this is just as true. When proof upon proof suggests that your idea is one without a monetization opportunity, it’s time to pivot. But only once you’ve exhausted all other avenues.