Standards as social contracts

Look­ing at the efforts Dave Winer is under­tak­ing in terms of get­ting OPML to become yet another stan­dard, I’ve been think­ing about how for­mats get adopted. The key insight I came up with is that stan­dards are actu­ally a form of social con­tract and increas­ingly, data for­mats is fol­low­ing the same path.

Look­ing at the history

In order to look for­ward, it always pays to look back. The dom­i­nant stan­dards for the web today are unde­ni­ably HTML (or its vari­ances like XHTML) and HTTP. More recently, XML has emerged and, increas­ingly, RSS is becom­ing the dom­i­nant type of XML for shar­ing a vari­ety of data.

How did each of those stan­dards become a stan­dard. It is obvi­ous now (hind­sight is always 20/20) that stan­dards bod­ies have rel­a­tively lit­tle bear­ing when it comes to influ­enc­ing the suc­ces of a for­mat. Take, for exam­ple, SGML, which was the dom­i­nant stan­dard­ized for­mat for doc­u­ment for­mat­ting. It was quickly superceded by HTML which, at the time, was not con­sid­ered a standard.

The same is true of RSS and other stan­dards for syn­di­ca­tion. For­mats like ICE, CDF, and NewsML were touted as the future when they were first intro­duced. How­ever, they’ve recently been superceded by RSS.

And even within the RSS world, for­mats like RSS 1.0, which was sup­posed to be more seman­ti­cally sound, and ATOM, which was sup­posed to be more for­ward think­ing that RSS 2.0, have been los­ing the war to RSS 2.0.

Boot­strap­ping is a social phenomenon

What Dave Winer under­stood, when he shep­erded RSS 2.0 into becom­ing the dom­i­nant mean of deliv­er­ing syn­di­cated con­tent is that the life and death of a new for­mat is pred­i­cated on its wide­spread adop­tion. And, in order to increase adop­tion, one has to make some­thing generic, easy to under­stand, and simple.

Many of the peo­ple in the early days of the syn­di­ca­tion space failed to see it as Dave did. We believed that a seman­ti­cally sound for­mat was bet­ter and we were wrong. Purity, it turns out is not always a good thing, espe­cially if it gets in the way of peo­ple imple­ment­ing something.

The same is true of HTML. I’d ven­ture that, from a devel­op­ment stand­point, the biggest boost to HTML was a sin­gle menu fea­ture that appeared in early browsers and remains there to this day: view source. In the early days of the web, count­less devel­op­ers learned how to do cool things with HTML by read­ing the source of pages designed by other people.

In a recent issue of ambidex­trous mag­a­zine, Jef­frey Schox talks about the three stages of tech­no­log­i­cal devel­op­ment: appro­pri­a­tion, early inno­va­tion, and sus­tain­able inno­va­tion. Here’s how he describes the appro­pri­a­tion stage:

an issued pat­tent allows inno­va­tors to con­struct road­blocks behind them as they travel down a par­tic­u­lar tech­no­log­i­cal path… Dur­ing the appro­pri­a­tion stage, patent road­blocks waste time and money… The coun­tries, need­ing to catch up with the designs and tech­nolo­gies of other coun­tries, should focus on col­lect­ing rev­enue and knowl­edge streams to fuel later stages of tech­nol­ogy development.

While he focuses on hard­ware and elec­tron­ics in a glob­al­ized mar­ket­place, the same truth can be applied to stan­dards. With few bar­ri­ers in adopt­ing a new stan­dard and by fos­ter­ing a cul­ture of appro­pri­a­tion, one can eas­ily estab­lish a base of peo­ple who under­stand a new for­mat. As more peo­ple under­stand it, they start imple­ment­ing it and, after even­tu­ally get­ting smarter about it, start build­ing on the efforts of pre­vi­ous cre­ators. Even­tu­ally, those masses of tin­ker­ers get to a crit­i­cal point, push­ing the new for­mat into areas that were unex­pected. Some com­pa­nies even­tu­ally get smart to it and see growth in that area, which trig­gers them into exper­i­ment­ing with that new format.

Even­tu­ally, due to a gen­eral agree­ment among all devel­op­ers, the for­mat becomes a de facto stan­dard. It does not have to have the impri­matur of a stan­dard body (except for some very late adopters or pock­ets where such imprint is con­sid­ered impor­tant) and moves forward.

What is inter­est­ing is the next stage, the one where stan­dard bod­ies see the area as hot and decide that they need to play in that field. A good exam­ple of that is the ATOM for­mat, which has been enshrined into an IETF approved for­mat, and to date has failed to stop the RSS 2.0 juggernaut.

So what happened?

The amaz­ing thing is how sim­ple the issue is. The rea­son RSS 2.0 has been win­ning is that it has devel­oped a fol­low­ing. With every new devel­oper learn­ing RSS 2.0, the for­mat goes stronger and the same is true of every com­pany imple­ment­ing it. Because it is sim­ple, it’s easy to pick up, which means that new devel­op­ers can do inter­est­ing things with it rel­a­tively quickly, giv­ing them a chance to become active mem­bers of the com­mu­nity and there­fore become hooked on it.

The other issue is in keep­ing things rel­a­tively open, while still main­tain­ing some level of con­trol over the gen­eral direc­tion. A suc­cess­ful future stan­dard has to allow peo­ple a chance to con­tribute but, in the end, it also needs some gate­keep­ers who decide what goes in and what doesn’t. The same truth can be applied to any sofware devel­op­ment cycle: for exam­ple, Linux may be a wide­spread open source phe­nom­e­non but the num­ber of peo­ple who decide what goes into the core ker­nel or doesn’t is still rel­a­tively lim­ited. The same is true of any suc­cess­ful open source project: some level of cen­tral­ized deci­sion mak­ing and dis­tri­b­u­tion of the work: any­one can con­tribute but not every con­tri­bu­tion makes it into the final product.

I’m now see­ing some of the same his­tory repeat itself in the OPML space. It’s a for­mat that is very sim­ple and Dave is work­ing very hard on get­ting peo­ple left and right to sup­port it. It’s the same sce­nario he’s used to boot­strap the RSS for­mat and to boot­strap con­cepts like blog­ging and pod­cast­ing into the main­stream. It’s a for­mula that works: keep it sim­ple to imple­ment, main­tain some level of cen­tral­ized con­trol over the roadmap and then evan­ge­lize it left and right until it can no longer be stopped.

Related Terms

More in Analysis,Business,Technology (5 of 5 articles)