Thu

Mar 8
2007

Marc Hedlund

Marc Hedlund

Jedi build their own lightsabers

I was down at Stanford recently with Adam, and sat in on one of the classes he's taking there. Later on, I looked around at some of the resources Stanford makes available on the web. They provide a lot of fantastic material for free. One thing I found was this video series of Larry Page and Eric Schmidt from Google, speaking at Stanford in 2002. I'd highly recommend it for entrepreneurs. (Each video is a few minutes long, and the whole set is about an hour.)

Larry's segment on tips for entrepreneurs, for instance, is fantastic. As he says, the advice is counter to a lot of what you'd hear as "standard wisdom" about technology startups. His point about the value of solving hard problems is one of the most difficult things for me to get through to people who take my Entrepreneuring for Geeks tutorial. A lot of investors and advisors will tell entrepreneurs to do the least amount of work possible, so that they can get going quickly. You can get going more quickly this way, but then you're stuck trying to build a real business, instead of just a thin UI on top of someone else's business. If you're making all your money off of AdSense (and you're not Google!), who really owns the relationship with your customers? Most people don't think this way, and they should.

Joel Spolsky has a great article on this called, "In Defense of Not-Invented-Here Syndrome." He talks about the Excel team making their own compiler and the benefits they got from that. He proposes a great test for making decisions like this:

If it's a core business function -- do it yourself, no matter what.

I've always referred to this idea, geekily, as, "Jedi build their own lightsabers." If you're going to depend on your lightsaber as your principal tool and weapon, you'd better know that it works.

We faced this decision at Wesabe when it came to syncing data from banks and credit cards. There are a couple of existing businesses that will do that for you, but when we looked at them, we realized that there were huge problems with going that route. First, we'd have to completely give up any control over our users' privacy, since those companies would need to hold all of our users' bank and credit card passwords, and they insisted on keeping a full copy of our users' transaction data. That wasn't compatible with our Data Bill of Rights, which we hear positive responses to every day on our support line. Having control ourselves leads directly to being able to provide control to our users, and that's what our users want. Second, syncing transaction data is central to our business. Let's say that we started to become successful -- if we had a single-source provider for that data, that would mean they would have full price control over us, and would get the full benefit of any value we created. It would be like installing a black hole at the bottom of our own bank account. Finally, we would have to pass that cost onto our users in one form or another -- either by charging them a high price, as some of our competitors do, or by selling their data to advertisers and marketers, which would completely corrupt our vision, to help people get control of their money.

So, we built our own transaction syncing infrastructure, and we did it our own way. We can provide those services to our users for free (in both senses -- for no money, and with freedom to control their own data), and with much higher privacy protection -- by giving people a client program that keeps their bank username and password on their own computer. Since we started with that as the first option, we now can extend to other syncing formats, including ones that are more convenient than installing a client download. People who took the easy route will be left with only one option for how to sync data, while we will have greater range of motion.

As Larry says, this isn't the common perspective, and others will surely disagree in the comments. But hearing Larry and Eric speak about this is great, and if you're interested in building a business, I think their talk is a fantastic resource.


tags: backstory  | comments: 11   | Sphere It
submit:

 
Previous  |  Next

0 TrackBacks

TrackBack URL for this entry: http://blogs.oreilly.com/cgi-bin/mt/mt-t.cgi/5317

Comments: 11

  Ian Muir [03.08.07 01:41 PM]

I'd have to agree with 100%. I've worked on a lot of projects that relied on 3rd party technology for core features and almost all of them turned south. Almost without fail, there was an update or change our tools a week before a release.

In other cases, we spent as much time learning how to use the tools as we would've just building them on their own. This is especially bad if we had to make any customizations. Along with your lightsaber example, this is like Darth Maul. His double bladed lightsaber wouldn't be very effective if it was just 2 lightsabers duct taped together.

  Blocky [03.08.07 04:26 PM]

Your point is excellent, but the metaphor is lousy. Building lightsabers is hardly a core competency of the Jedi order. The fact that they don't oursource it (as most armies do outsource weapons manufacture) simply shows that they are far too dependent on tradition for their own good.

Yeah, it's true. I'm a total geek. I have no life.

  Marc Hedlund [03.08.07 04:44 PM]

Blocky, building C compilers isn't a core competency for the Excel team, either. But they thought the dependency was critical enough to want to do it themselves. The lightsaber analogy was that -- not that they were necessarily the best builders of weapons, but that it mattered more to them than anyone else that they get it right.

And I am second to no geek in having no life. :)

  Jack [03.08.07 10:34 PM]

I'm with everyone else. Every venture I have tried and relied on 3rd party code for has not worked out how I expected at all. So, for my latest project (my blog) I have done all the work myself. I do the PR (that WAS a hard job to figure out), I sort out advertising and advertisers, and the whole thing is built on my code. And it works. If something goes wrong, I can fix it, because I made it.

Unlike a friend of mine who uses an online "push button publishing" blog, tried using someone else's template for it, and ended up coming to me asking how to fix it when it all went wrong and her blog's content was thrown all over the screen. I told her I had no idea how to fix it, because I hadn't written the template, or the technology behind the blog.

  Nivi [03.08.07 11:27 PM]

I think "core competency" and "core business function" are a little too vague to be useful.

Check out Clayton Christensen's Innovator's Solution for a more sophisticated approach.

Christensen says something along the lines that businesses should "in-source" the parts of the value chain that drive performance. Performance is defined by the customer. And the customer's definition typically changes over time in a predictable way as the product becomes "good enough". See Chapter 5 of the book for details.

  Steffan [03.09.07 05:55 AM]

This is very similar in sentiment to Alan Kay's "People who are really serious about software should make their own hardware."

Apple, of course, are very keen on that.

  Bob [03.09.07 10:46 AM]

Perhaps a better analogy is that skydivers pack their own parachutes.

  Simon [03.09.07 10:57 AM]

A long time ago in a galaxy far, far away, before becoming a venture capitalist, I had the adventure of working at a hedge fund with two of the world's must successful traders.

In addition to developing trading algorithms and risk management strategies, we spent a lot of time analyzing exactly what these top traders did in order to figure out what made them so successful.

Among the things that we discovered was that it didn't matter what instrument you bought, so much as when you bought it and when you sold it.

As it turns out, the most important factor in determining your discipline to buy and sell was the confidence that you had in your system (and, interestingly enough, it wasn't even the fact that your system was that great, just the ability to stick with it).

Low and behold, the best way to gain confidence in one's system (read, 'discipline to stick with the system') was to build it yourself.

As a result, when training new traders the first thing they always did was to get every trader to build his or her own system.

So, the act of building ones own lightsaber is a worthy exercise in and of itself in the sense that it imbues young Jedi with a sense of confidence that might otherwise be lacking.

In terms of outsourcing the production of lightsabers, no real Jedi would ever consider doing this!

A lightsaber is not a mass produced DL-44 off the Blastec assembly line, a Jedi has a special bond with his or her lightsaber.

Returning to the somewhat more mundane yet equally dangerous world of Web 2.0 et al, I think that "building your own lightsaber" is valuable in two respects.

First, as has been well documented in other comments, you are not placing yourself at the mercy of someone else's product and second, the act of building it yourself seems to have an intrinsic value (I sense the gleanings of a universal rule lurking here).

Who knows, the fate of the galaxy may some day depend on it!

May the Force be with you.

  Kevin Curry [03.10.07 12:04 PM]

At risk of this becoming a Mutual Admiration Society, I get the point, too. Yet, I wonder if there aren't other lessons here, bordering but outside the realm of technology. Simon alluded to it:

"In addition to developing trading algorithms and risk management strategies, we spent a lot of time analyzing exactly what these top traders did in order to figure out what made them so successful."

In other words, alignment to effective business practice is what mattered most. I'm not a fan of big-bang, Nth-party, COTS product integration either, for all the reasons stated. But perhaps more than just needing to know how it's built, Jedi are building their own sabers because Jedi understand best how they will use the saber(above, re: Darth Maul).

  Mike Pearson [03.12.07 04:48 AM]

I don't think it's as straighforward a decision as people often make out. Building, adapting or buying components needs to be reviewed on a case-by-case basis.

For example, the videogame industry has moved towards a third-party supply model of physics, 3D rendering and AI components. There's even a third-party product dedicated to producing trees for inclusion in games. In these cases developers have decided that specialists have solved the issues more effectively than they could, be it in terms of time, cost or sophistication.

The focus needs to be on putting effort into areas that make your product or service unique, not on making incremental improvements to non-differentiating areas.

  lawrence eason [10.31.07 08:46 AM]

I would raelly like to build my own lightsaber, but i have no parts to it.

Post A Comment:

 (please be patient, comments may take awhile to post)






Type the characters you see in the picture above.

RECOMMENDED FOR YOU

RECENT COMMENTS