Algorithms IRL: Gradient Descent and Start-up Ideas (2/3)
See Part 1 here: Algorithms in Real Life: Gradient Descent Pt. 1
Alright, so last time I went through some really non-controversial examples of how thinking in terms of gradient descent leads us easily to lessons we’ve already learned the hard way. Now I want to apply the thinking to something that’s a bit less obvious.
I want to start with an oft-repeated mantra of start-up people: the idea for your start-up isn’t that important. Sometimes they go one step farther and say it doesn’t even matter.
Your idea doesn’t matter.
I don’t know Brian and I don’t mean to pick him out in particular. You can find this identical sentiment everywhere. The point has been made more times than I care to count. Serial entrepreneurs, both successful and failed, start-up gurus and junkies, and venture capitalists of all variety have bought into a particular piece of wisdom that has become quite popular in the start-up world: start-up ideas don’t really matter, the team does. Here’s some other examples:
Your idea can suck. Just get started.
The idea doesn’t matter much; it will change anyway.
If you are anything like me, listening to all of these people encouraging other people to start companies without a good idea hurts my soul. It seems like such terrible advice. Even though I agreed with many of their points, I still felt the message was wrong. It took me awhile to put my finger on it. I’ve decided that they aren’t really wrong, they just aren’t fully explaining what they mean.
The Advice isn’t Wrong (as long as you are listening closely)
Their hearts are in the right place. If you listen closely, what are they really saying? They are saying “Stop being paralyzed waiting for the perfect idea to come around. Find smart people and get started”. This is the message I believe these people are trying to convey.
This tends to be shortened into “start up ideas don’t matter”. I finally figured out the reason that this version bothered me so much. It’s because its wrong. Start-up ideas matter! They matter alot! Maybe I’ll give Paul Graham a chance to explain:
The fact is, most startups end up nothing like the initial idea. It would be closer to the truth to say the main value of your initial idea is that, in the process of discovering it’s broken, you’ll come up with your real idea.
If you read closely, it becomes obvious that Paul isn’t saying that start-up ideas don’t matter.
Wasn’t This Supposed to be about Gradient Descent?
Yes! I’m getting there. Paul’s detailed explanation of why he believes ideas don’t matter “much” actually tells us alot. I think the nuances of this advice are important. His approach is actually screaming gradient descent.
You need to start to ever succeed. Once you start, you get a prototype as fast as humanly possible, gather feedback, and head downhill. They are all advocating the gradient descent method of finding a successful business idea.
There is only one simple problem with translating this into the shorter “start-up ideas don’t matter”. Where you start determines where you finish! There is absolutely no way of knowing how your idea will evolve over time, but it will. The competitors you have, the lessons your learn, the customers you gain, and the network you create will all color your view and ultimate determine where you end up. This is why its important to choose a market niche that has potential because you need to be in the ballpark of a good idea to actually get there. The primary purpose of your idea isn’t to be an idea worth millions of dollars but to be an idea that puts you in the ballpark of million dollar ideas. With that understanding, it matters a great deal what it is, but its value isn’t necessarily its worth as a business.
It’s also worth pointing out that the pains of software planning don’t hold a candle to business planning. You simply cannot look at the business landscape at all and get a feel for it. In fact, in business, it’s extremely difficult to even figure out which way is downhill. It can lead you all over the place through seemingly unrelated places and back around. The business landscape is extremely convoluted and so there is skill in both choose a starting location, and evaluating the environment to figure out what to do next.
A Quintessential Example
(The following story is paraphrased from Founders at Work by Jessica Livingston)
There are a lot of examples like this from the business world. I’ve chosen one of the more recent ones to try to keep it as relevant as possible: Paypal.
The technical co-founder of PayPal, Max Levchin, set out originally to make cryptographic libraries for Palm Pilots. He wanted to use palm pilots to do access control. This way you can have all your FOBs replaced by a single Palm Pilot that you could use to gain access to various areas. Since no one was using Palm Pilots, no one was making software for them, and that meant no one was going to buy their libraries. The second thought, then, was that they’d make the software for the Palm Pilots themselves. It solved the immediete problem but not the larger one (no one was using Palm Pilots in enterprise).
When that didn’t work, he got a better idea. How about letting people use Palm Pilots to exchange money? He had all the cryptographically secure infrastructure to do it. He raised a big pile of money to do secure money swapping on Palm Pilots. He built a website demo to support his Palm Pilot money transfer program. Look at this silly web demo, pretty neat? Download the Palm Pilot version, its nifty! Suddenly, people from a strange site EBay started wanting to use their silly web demo. Eventually, the world’s message got through and they went to work full-time on the website.
That’s gradient descent my friends. This little tale illustrates both points brilliantly:
- Start-up ideas don’t really matter. Getting smart people working together and heading “downhill” is far more important than having a killer idea out of the box.
- Start-up ideas really do matter. Not because of how good a business they make but because of the environment they create which pushes towards better ideas.
December 3rd, 2008 at 10:02 am
Of course, you need to set some context around the quote of mine you pulled. I’m writing about people’s fetish for NDA’s and secrecy around some ‘million-dollar’ idea they have. The point of the post was that ideas are cheap, and we all have them willy nilly. Ideas are never the problem. Prioritizing, prototyping and responding to *real* conditions are where things start to get valuable. What is the actual experience of the thing in context? An idea in a pitch deck, meh. Really, who cares? Can you make the call to abandon a game your team has been working on for a year plus to start to develop out a new idea for photo sharing? Can you make the call to throw more resources to an SMS based status update service that pivots off of one small feature that a defunct mobile service once offered? Etc. Of course, great ideas are a necessary, but insufficient condition. Also, the best way to have a good idea, is to have lots of them.
December 3rd, 2008 at 11:28 am
I’ve been on PayPal since the Palm days, and may have actually used that version to split a restaurant tab with a friend (not sure anymore). I was very amused last year when they rolled out PayPal mobile.
December 3rd, 2008 at 12:31 pm
To some extent I’m repeating Brian Oberkirch’s comment, but the way I’ve always interpreted this little chestnut is this: I have probably an idea per month that, if perfectly executed, would be ‘million-dollar’ ideas. I suspect most smart people do. It’s the execution that is the hard part. I’ve always taken the chestnut as a retort to those people who come up with a good idea and then guard it as if it’s the first original idea in the history of the world, who believe that having the idea is the hard part and then all too often proceed to totally botch the execution because their eye is on the wrong ball.
December 3rd, 2008 at 5:14 pm
@Brian (and Joe),
I hope you don’t feel I unfairly quote-mined you or put words in your mouth (or anyone else’s). I don’t disagree with anything you’ve written (including the post I quoted). I hoped to make that clear in the following sections that by actually “listening” to what you guys said, it was clear you were making an altogether different point.
I think the problem comes when “we” read posts like yours, paul’s, and dharmesh’s and distill it into a general belief that ideas are largely irrelevant.
In a way, we are agreeing, through analogy. I’ve said that the business landscape is complicated in such a way that figuring out which way “downhill” is is extremely difficult. That is precisely what you are saying. In fact, you’ve gone a step farther and said that the starting condition isn’t nearly as relevant as your ability to find and head downhill (because it is so much easier to find a good starting spot than it is to figure out which way to go). I’d agree with that statement, as well.