Umair Haque / Bubblegeneration
umair haque  

 
 


Design principles for 21st century companies, markets, and economies. Foreword by Gary Hamel. Coming January 4th. Pre-order at Amazon.


 
Friday, April 28, 2006

A Tiny Model of The Long Tail of Peer Production


Nick has a great and inflammatory post about numbskulls and 2.0. Which is important, because it crystallizes a lot of the (flawed) thinking about 2.0 lately. You might also wanna check Ross's latest post for more context.

First, Nick's argument: most open-source contributors are "numbskulls", a tiny fraction are "experts" whose ideas/opinions create value.

"...Wikipedia had over 850,000 articles in English, and 2.9 million across all languages ... This content was generated by fewer than 50,000 contributors in English, and 103,000 total."

To be honest, I'm not quite sure what the point of this argument is. That the ratio of contributors:articles should be > 1? Does it matter? No; by itself, it's a meaningless observation based on silly pseudo-math.

Let's recast this question to gain some real insight.

We know preferences are heterogeneous. For the last few centuries, economics has focused on consumption preferences. We like different things; I like My Little Pony, you like Miike Takashi.

Now, the pendulum is shifting. In a world of abundance, we must understand the implications of heterogenous production preferences.

Let's assume a world of perfect information or zero transaction costs.

In this kind of world, what do heterogenous production and consumption preferences really mean?

Not what Nick is saying; that there will always be one big long tail of peer production.

Rather, they imply two things.

First, agglomeration economies mean people with similar consumption preferences will group together in communities focused on different products/markets/outputs. If I wanna discuss makeup, I have go to MUA - that's where the discussion happens. These are simple natural monopoly dynamics.

Second, for each community (or set of people with similar consumption preferences), there will be a long tail (or some similar power-law distribution) of peer production; because they will have different production preferences as well.

For each Wikipedia, for each MUA, for each Basenotes, for each Squidoo. And this is, in fact, exactly what we see in the real world at those communities.

Pretty cool.

Now, these dynamics only create problems in two cases.

First, where there is no selection of different outputs regulated by that community. In that case, the most productive members, even if they're morons, will produce eternal crap unstoppably.

Think Geocities.

Second, where there are no experts involved in regulating outputs. If there aren't, the crap will never get filtered out.

Think blogosphere.

Now, those two examples should make it clear that revolutionary peer production models - the ones that make cool stuff - work because they're the opposite of what Nick is saying.

Put another way, it is the coordination that takes place in markets, networks, and communities that filters out the power of numbskulls.

But more to the point, if you follow the heterogeneity argument above, you may be a numbskull at Wikipedia - and a genius at MUA.

The point of open source, to put it another way, is not that everyone does contribute - it's that everyone can contribute. Now, this may not yield perfect outputs - but it does yield, in many cases, hyperefficiency: outputs at radically less opportunity cost than the firm could produce them, or far more innovative ones.

So you should see by now that the argument that open source models "rely" on the intelligence of their participants is a neat sleight of hand (after all, everything does - from firms, to anthills). The genius of these models is that they tap the space where expertise and heterogeneous production preferences intersect - something that almost never happens in firms, despite big bonuses, nasty bosses, and the other nice things about the rat race.

I've italicized that because I really do think it's absolutely revolutionary from an economic point of view.

There's a longer version of all this forthcoming in the long awaited (but rarely hyped) Bubblegen book.

***

Second, a bit of selection bias. Is Wikipedia "mediocre"? Perhaps. But Wikipedia's an easy target, because it's such a mammoth undertaking.

Let's choose some other open source outputs. Is Firefox mediocre? Is Apache mediocre? Is Linux mediocre? I doubt it.

So Nick's argument begins to look a little more dubious.

Let's take it further. Let's say I'm a makeup lover. I trade info about make up at MUA. Is this info "mediocre"?

By expert standards, sure. But the standards of the community are often very different. In many cases, communities exist to overturn "expert" orthodoxies - by hacking, tweaking, and remixing things. The point of Wikipedia is not to be the "best" encyclopedia - but to revolutionize the terms of "best" from most accurate to most ubiquitous, up to date, and easily accessible one.

***

Finally, self-contradiction: in telling us that 2.0 is utopian, doesn't Nick reveal that he is, in fact, the ultimate utopian? After all, here is a guy using a blog to tell the world essentially that if everyone wasn't a numbskull, they could make the world a better place...

Ah, irony.

-- umair // 1:10 PM // 18 comments


Comments:

Umair, Very interesting. I agree with much of what you say (though I'm not sure you address the focus of my post, which was the application of Web 2.0 tools within an individual company, where there's an organizational dynamic very different from what exists in the broad marketplace of ideas).

But I think, by trying to stretch the definition of open source, you miss a crucial point. The filtering process by which the community's raw contributions to Linux and other successful open-source software programs are transformed into an actual product is not an example of what you call "the coordination that takes place in markets, networks, and communities that filters out the power of numbskulls." Rather, it's an example of, to use your words in a slightly different way than, I think, you intended, "experts ... regulating outputs."

These are two very different models, and I think you're trying to conflate them. There's quasi-democratic, market-based and technology-mediated filtering, and there's hierarchical, authoritarian filtering by a meritocratic elite. You can't use examples of the latter to praise the former.
// Anonymous Nick Carr // 2:58 PM
 

Much of this edge competency I feel is covered in The Theory of the Firm as a Nexus of Contracts. If one has a loose definition of contracts and stakeholders it leads to

"those who relations to the enterprise cannot be completely contracted for, but upon whose cooperation and creativity it depends for its survival and prosperity." (This definition stresses the importance of stakeholders for long-term cooperation and innovation.)

and

"stakeholder" embraces certain corporate "constituencies," including, at the core, shareholders and employees, but also extending to certain customers, supplier and lenders.

See
http://www.thecorporatelibrary.com/governance-research/patterson-report/LinkDetail.asp?ID=265

Does this lead to edge competency?
// Anonymous Anonymous // 5:25 PM
 

Hey Nick,

That is a very interesting point which I have to chew on. I don't think I'm conflating, but I see what you are getting at.

I agree your context was very different to my response - I don't think these models *can* work inside the firm at all.

Anon,

Any economic phemonenon can be called a "nexus of contracts".

Though the definition you cite is somewhat analogous to edge leverage, it doesn't help us strategicall - what is it that we must do to utilize this "cooperation and creativity"?

Clearly, offering everyone a "contract" in the standard sense of the word isn't nearly enough.

Thx for the comments.
// Blogger umair // 5:32 PM
 

�Any economic phenomenon can be called a "nexus of contracts".

Correct, so if we then view the Firm as you put it as an "economic phenomenon" I believe applying edge competency strategies or at least thinking in these terms is easier. I was not proposing a solution. I still am getting my mind around this. But if we(me) cannot even conceptually grasp it forget about implementing a solution that embraces it (edge competency). Definitions are important and even more important is ones perceptions. If decisions makers thus have a rigid definition/perception of �The Firm� then they cannot begin to think in the ways you are suggesting.

�Clearly, offering everyone a "contract" in the standard sense of the word isn't nearly enough.�

Agreed but if we view everyone as a �stakeholder�, contract or no contract, can we then begin to think strategically with respect to edge competency?
// Anonymous Anonymous // 6:02 PM
 

�Any economic phenomenon can be called a "nexus of contracts".

Correct, so if we then view the Firm as you put it as an "economic phenomenon" I believe applying edge competency strategies or at least thinking in these terms is easier. I was not proposing a solution. I still am getting my mind around this. But if we(me) cannot even conceptually grasp it forget about implementing a solution that embraces it (edge competency). Definitions are important and even more important is ones perceptions. If decisions makers thus have a rigid definition/perception of �The Firm� then they cannot begin to think in the ways you are suggesting.

�Clearly, offering everyone a "contract" in the standard sense of the word isn't nearly enough.�

Agreed but if we view everyone as a �stakeholder�, contract or no contract, can we then begin to think strategically with respect to edge competency?
// Anonymous Anonymous // 6:04 PM
 

Hey Anon,

Let me simplify. You're talking about a property rights theory of the firm.

That's fine.

The problem is that property rights are vaguely defined if at all at the edge (in open source communities, for example).

So thinking in terms of stakeholders and rights isn't very helpful - this is why the media industry keeps stalling at the edge; because their understanding of strategy is based around outdated property rights.

It's better that you think what markets, networks, and communities are really good at (and bad at) and reason backwards (or just wait for my book :)

Or, just search for edge competencies, plasticity, and liquidity and you will have plenty of stuff to think about.
// Blogger umair // 6:12 PM
 

Umair,
I think the projects you mention have as much selection bias as Nick's. Shouldn't you look at open source in general instead of a few select projects?

As a general rule, open source fails and is lousy. Most projects never get very far and the ones that do tend to be the ones that are most driven by a small handful of people or less. In general, the code submitted by the masses gets rejected because it sucks.

You are right that there is value in the intersection of production preferences and expertise, but I think if you look at most "Web2.0" companies, they wouldn't qualify as examples of your statement.
// Anonymous Rob // 7:06 PM
 

Hey Rob,

You're absolutely right, my sample is biased too.

While many open-source projects may never work, I think the reasons for failure are very different from the reasons for (lack of) success.

That is, an open source project might not be interesting, so it gets no contributions - which doesn't tell us much about why successful open source works.

So sample bias is a little bit necessary to begin to disentangle the hypotheses we can use in more rigorous studies in the first place.

I also think many 2.0 plays are exactly this intersection - or are at least hoping they will be. In fact, I would go so far as to say this intersection is half of what makes 2.0.

But I would certainly like to hear more about your experience...
// Blogger umair // 7:14 PM
 

Peer production is not about aggregating low level contributions - it is about adding this aggregation to the whole ecology of information production.
We still need high quality or big scale contributions - but since it is now technically possible to also aggregate the all small scale or low level contributions we should not reject them arbitrarily. There is evidence that the peer production model can create incentives for contributions on the whole spectrum of scale. Hmm - it seems that I am just summarising 'The wealth of networks' (http://www.benkler.org/wealth_of_networks/index.php/Main_Page)
// Blogger zby // 4:14 PM
 

zby,

I disagree. Would you call Digg peer production? How about delicious or flickr?

Tiny contributions can (and do) add up to something very large - it's just that the result is very different from Apache or Linux.

Thx for the comment.
// Blogger umair // 4:23 PM
 

Umair, I think Nick's key point is that contributors to peer production processes are less likely to be the best possible contributors on that subject, even if, to your point, they are better than a random group. To use your makeupalley example, contributors there know more about makeup than any other group, but those who could make the BEST contributions on any makeup topic are disuaded by the contributions of those with less insight on the topic of makeup -- even if those active contributors are insightful in comparison to the "average" person's views on makeup.

So the peer production effects are MORE efficient but not MAXIMALLY efficient because a critical mass of the highest yield contribution is never achieved.
// Anonymous Scott Karp // 9:22 PM
 

umair - yes I do include Digg and del.icio.us as peer production, but still those two use servers, software and administration that are not commons produced by the peers.
// Blogger zby // 12:24 PM
 

Hey Scott,

I think that is a pretty good way to put it, but I think that argument relies on an expert driven notion of "best" in the first place.

In a sense, a place like MUA exists to debunk the "expert" makeup reviews you will read in Cosmo.

zby,

I'm not sure I follow your argument. Clearly, the peer contributions are worth a great deal more than the servers/software/admin.

Thx for the comments.
// Blogger umair // 1:40 PM
 

Hey Umair,

Cool post. Liked the italicised sentence ("they tap the space where expertise and heterogeneous production preferences intersect"), I think it goes pretty far in explaining peer-production. Let's analyze it a bit further.

First of all, if this were entirely true, then by definition, this would lead to maximally efficient production. Basically, what your sentence says is that those who know best seek opportunities to contribute to where their expertise would be most useful. This leads to creation of production pools that attracts those with expertise.

However, as your Geocities example shows, this is not always the case. I think the case of abandoned open source projects also demonstrates the same thing. Why is this the case?

One way to think this through is that projects do not fail merely due to lack of expertise. Nor is a project's success guaranteed just because it can attract all the experts it needs. This is trivially true in the corporate world : billions of dollars and thousands of scientists do not a successful product make. Otherwise, product development woulkd become a riskless expertise in purchasing the necessary intellectual and financial capital.

I think the basic problem is with using the word "expertise" in your hypothesis. Let's first recognize that the dynamic we are witnessing is not just a deployment of expertise in peer production. I see at least three different types of contribtuion, of which expertise is only one. Here they are:

1.Microchunk of knowledge (this would correspond to your notion of expertise)
2.Micro-manufacturing (say, contribution of code)
3.Micro-influencing (say at MUA, where a contributor gets buy-in from community about her preferences- perhaps due to her superior ability to judge make-ups or discover new cool stuff)

Once you start seeing it this way, you realize that successful Web 2.0 plays are the ones that are able to
implement superior filtering mechanism. In fact, this ties in with your attention market thesis - except that this is not merely restricted to attention, it's aabout a market for contributory abilities whose actual operating context may be any of the three contributions modalities.

This then also shows you where the obvious source of competitive advantage lies for any 2.0 place. I shall refrain from spelling it out so that you continue to get mucho $$$ :-). Suffice it to say that Reversion Wars on wikipedia fascinate me for that reason, as I mentioned to you an an e-mail sometime ago in another context.

BTW, I haven't forgiven Nick Carr for foisting his IT Doesn't Matter crapola on me in b-school :-) :-). The double irony of reading this post on his blog which displays the cover of this book (man, how I wish I could get mucho $$$ for writing such crap - I suppose I need to be BS'ing about this shiite at HBS) was just priceless. Dude Nick, if IT didn't matter, why are you even talking about this stuff?
// Blogger Mahashunyam // 1:15 AM
 

Oh, and BTW, it should be obvious why this stuff just can't happen in a firm : the fundamental organizing principle of a firm is to minimize transaction costs through hierarchical structures, not connecting those with the best expertise to the problems they are best suited to solve. Trivially, nobody could fire Scott McNealy at Sun even if they had figured out how to save Sun's business. In fact, this is even more obvious in politics : those who know best how to solve our problems are nowhere near having the power to come together and implement those solutions. But that's a debate for another day.
// Blogger Mahashunyam // 1:20 AM
 

Good design!
[url=http://ogvwajqv.com/nqgb/uysa.html]My homepage[/url] | [url=http://kojqtwjs.com/csud/dhuv.html]Cool site[/url]
// Anonymous Anonymous // 6:37 AM
 

Good design!
My homepage | Please visit
// Anonymous Anonymous // 6:37 AM
 

Nice site!
http://ogvwajqv.com/nqgb/uysa.html | http://swhzrwfc.com/qlaf/zvcd.html
// Anonymous Anonymous // 6:37 AM
 
 

Recent Tweets







    input
    portfolio
    contact

    mail.
    uhaque (dot) mba2003 (at) london (dot) edu

    skype.
    umair.haque

    atom feed

    technorati profile

    blog archives