Russ Roberts

Eric Raymond on Hacking, Open Source, and the Cathedral and the Bazaar

EconTalk Episode with Eric Raymond
Hosted by Russ Roberts
PRINT
Fazzari on Keynesian Economics... Roberts (and Hanson) on Truth ...

Eric Raymond, author of The Cathedral and the Bazaar, talks with EconTalk host Russ Roberts about the ideas in the book--why open source software development has been so successful, the culture of open source, under what conditions open source is likely to thrive and not to thrive, and the Hayekian nature of the open source process. The conversation closes with a discussion of net neutrality.

Size: 30.8 MB
Right-click or Option-click, and select "Save Link/Target As MP3.

Readings and Links related to this podcast

Podcast Readings
HIDE READINGS
About this week's guest: About ideas and people mentioned in this podcast:

Highlights

Time
Podcast Highlights
HIDE HIGHLIGHTS
0:36Intro. Open source movement. Hacker: different meaning from media use of the term, which is corrupt and wrong. Not someone who commits security crimes or breaks into machines. Since 1960s, hackers built programs and give away software, including building the internet and the world wide web. Term got hijacked because crackers, people who break security generally aren't very skilled, and appropriated the term. Going to use the word correctly in this podcast, isn't a better word for the members of this culture. Liberal has since come to mean the opposite of its 19th century meaning. What is open source software? Folk practice that evolved over period of decades, building software in a fundamentally different way. Outside the culture of hackers, secrecy in small teams. Within the culture, working collaboratively and subjecting software to peer review. Linux started to attract notice: unconscious folk practice made conscious. Effect of articulating it, having a theory of open source, meant could improve the process. What are some of the programs that use open source? Linux, Firefox, open-source browser. Linux, flagship project. Apache web server. What kind of market share? Apache running somewhere between 60-70% of the websites in the world, particularly running websites that have to be reliable, where performance is an issue. Logistics of open source--Wikipedia, anybody can play with it who wants to. Linux--Linus Torvalds, proprietor, manager, only big editor, originator, gatekeeper. People make suggestions to the code and he either gives them a checkmark or doesn't. Somewhat more complicated than that. Lieutenants. Typically one founder or small group of core developers; job is also to filter patches coming in from outside. Stability provided by core groups. If you make a change that is turned down, you are still free to use it; just can't distribute it as part of Linux, though could distribute it in competition with Linux. No legal bar to doing that; keeps leaders on their toes and from becoming autocrats. Huge controversy if forked. Why does that cause controversy? Inefficient, divides community's resources. Customary law, like common law property rights: people are in effect farming software for reputation gains. Glory. Worst thing, ultimate taboo, you can do is to file somebody's name off a project.
10:16How does a hacker get glory? Patch or add feature. Good hackers use their own names. How do I know who were the people involved in different parts. If new, often a credit file in source code. Notice who the steady people are, who does good work. If I make a small change, is that recognized? Might get credited in the commit log of the project but not in the credits. Usually they have a problem with the software they want to solve and it's more efficient for them to send it than to maintain their own branch.
12:19Incentives. Obvious advantage of Linux over other system. "With enough eyeballs, all bugs are shallow"--Torvalds operating as if that were true. Twenty years of observation to go on. Culture already knew it. Implication is that all software is buggy; as systems become more complex, bugs spread. Nothing short of total process transparency actually works. Linux kernel is about 6-9 million lines of code. More than PL1 code, 1973, that could make change. Distributed peer review will also run out of steam one day. Old idea in social science: wisdom of crowds, market, Hayekian idea that prices contain knowledge that no individual participant in the market can possibly have. Scientific peer review process: if you have a fundamental new insight in science, it's not accepted till it's been subjected to review by your peers. Secrecy hid the code in the name of keeping the profit stream. Metaphor: cathedral and bazaar. Fred Brooks book, The Mythical Man-Month, mid-1970s, dominated thinking, fundamental problem of software is that it is complex, problem with complexity control, need to keep teams really small and objectives tightly defined and managing the process with tight control. Long release times, most of time obsessing about squeezing out the last bug. All opposite of Linux. Cathedral builders, central, hierarchy, aspiring toward perfection. Linux community violates all these rules in effort to get maximum feedback. Messy, horizontal process, opposition to the cathedral: bazaar. Cathedral and the agora--Greek term for public meeting place. Bazaar is flat, very wide, not narrow and tall, not a spire. Self-organization from the bottom up, Hayekian sort of idea.
20:28Puzzle: On the surface, you would think closed source software would be elegant; open source would be weird, Rube Goldberg-ian. Stephenson. Opposite is the case. Why? Architects. All good software has to have an architect. Hear same argument about a market system--that "in a centrally planned system there's planning but in a market system there is no planning." But there's planning in a market system, just not central planning but by individuals. In closed source, can do review internally, second group of engineers. Problem is process is subverted because both groups are reporting to the same people, pressure to minimize problems, problems that the customer trips over. Discordian snafu principle: law of social dynamics. Elaborate joke posing as a religion or a religion posing as an elaborate joke. In any hierarchical authoritarian organization, inferiors will be more consistently rewarded for telling their superiors what they want to hear than telling the truth. Result is disconnect between the people at the bottom who can do but not decide and the people at the top. Deciders get perpetually worse information. Arnold Kling, geeks vs. the suits--people who understand the risks vs. those at the top at Fanny and Freddie. Conversely, in open source community, not all reporting to the same bosses. Tradeoff: role of profit motive. Most good corporations never have as their motto or mission statement that they want to make more money. Inspiring things about their product and consumers' lives. Open source advantage is that it is only focused on the consumer: software works better, hard to keep the connection in their brains. Open source is not a way to make money. Enabler for other activities that make money. Suppose you are a vendor and you want to sell a networking card. Traditional closed source group, ship binary driver with that card. People who buy that card take risk that you will go out of business and driver won't be maintained or that there are bugs they can't fix. Skimp on people working on bugs. Or: If you sell ethernet card with disk with source software included and pointer to website with community maintaining those drivers. Death of your company won't kill the software you are relying on. Value to you is increased; and can rationally expect fewer bugs because of transparency and simply having more people working on it.
27:59Lots of different things motivate people: money, glory, thrill, artistic satisfaction, craftsmanship. Walter Williams: if you want to get steak and potatoes to NYC from Nebraskan farmer who has to get up early and work hard, could choose to rely on good nature or on profit. Profit works extraordinarily well; good nature, maybe. Brother taking you to the airport vs. a taxi. Deadliest thing an engineer can say about a project is that it doesn't scale: that is, it works at small sizes but not large sizes. Nonlinearities make it stop working at large sizes. Love doesn't scale. Why do people see Linux or Wikipedia as a labor of love? Many people just do it for the fun or excitement. Why does it work so well? Very particular set of circumstances where love scales. Open source, not Wikipedia. Scales when you have: 1. capital goods are cheap; 2. limiting factor is human creativity and attention; 3. work is intrinsically rewarding; 4. an objective metric for success. Wikipedia is an interesting case, fulfills all of the conditions except 4, no objective metric for success. When writing software, program either runs or it crashes. For an encyclopedia entry, no objective measure of truth or of whether it's a good entry. Without that condition you get thrashing. Without condition number 1, capital goods too expensive, then no one can afford to do it for free. That's the situation the hacker culture used to be in. Computers used to be really expensive. Need all four conditions to have swarm attack. Other things like volunteer work work with varying degrees of success. Fifth factor: zero costs to communications, cheap ways to communicate. Before the internet. For general public, 1993-1994; hacker culture using it 15 years before but only some had access. ARPAnet.
36:10Great that people have this hobby and tinker, but it doesn't scale in another dimension: everybody can't have it as their hobby. Have to have a real job. How little software is actually sold. Little software done as piecework. Most software expertise is developed for businesses. Report generators for databases, COBOL problems for banks. False that typical programmer is working for Microsoft or Oracle. Position advertisements in local paper on online: how many are software for capital goods inside corporation vs. packaged product. Most is for inside corporation. One exception: computer games, which do have to be sold on the merchant market; short-lived and very few people employed producing them. What's a small number? Design side. Probably not more than 3000-4000 nationally. Not enough money in that industry, tight product margins. Once a project gets going, useful and going to attract the eyeballs. Could you imagine a world where all software is open source? In that world would the genesis of projects be the same? Gimp comes along because Photoshop is expensive. Gimp is popular because Photoshop is expensive; it didn't happen because Photoshop is expensive. Happened because a few people thought it would be cool to write a graphics program. Open source projects typically start because somebody has an itch. If they're good and if they're lucky, others look at it and say they could make it cooler with a patch. Range of motivations: some in game because it's fun and useful; others for ideological reasons, against corporate policy--that group not dominant. Sometimes explicitly to break a monopoly. Monopolies not liked because engineers are suspicious of systems with a single point of failure. Also monopolies focus on secrecy and control, stopping me from getting my work done. Secrecy is a barrier, it's the enemy of quality. Three types of organizations: unmanaged, crowd-sourced; Microsoft; Apple. Different monetary, glory returns; and different secrecy. Outside impression: Apple seems to be good at attracting good people. Burn them out very fast. MS stereotype stodgier; it burns them out, too. Successful projects results. What internal mechanisms? Mystique associated with Apple. One thing Apple does extremely well: industrial design, making things pretty. People who are attracted to that like the idea of working where it's a significant focus. Screaming successes of open source are products consumed by hackers. Firefox is the exception.
46:54Another factor: objective metric of what is good. Harder to apply in a user-interface environment than anyplace else. Open-source culture realizing this and appointing UI (user interface) dictators within each project, and they get their way. Committee just doesn't work. Flood-individual vision. Not everybody has an aesthetic sense. Could argue negative correlation between hacking and aesthetic sense. Wrong. Stereotype. Programming only looks like a mechanical, logic-saturated activity at its lowest levels. Have to be able to do logic, know a lot of mechanics. As you get better at programming, that stuff becomes quasi-automatic, learn entire programs in days. Play the saxophone, can learn the clarinet. Like music. Mastery of technique and discipline. Once you reach a certain level, good taste and aesthetic sense. Good engineering, stability, robustness over time. At highest levels, aesthetics is all-important. Left-brain, right-brain: hackers tend to be left-brained. Suggesting right-brained aspect. Early days when first Apple computers came out, relatively attractive screens compared to IBM. Different fonts, nice screen, certain part of the user community disdained that and wanted familiar green/black. Ultimate command line guys. Didn't think it was an appropriate use of computers at the time because there were better things to do with the limited computing power of the day. Nowadays, lots of resources, much cheaper. Programming time is a more limited resource, but not that much more difficult any more to make things pretty. Disdain for the pretty interface has pretty much subsided.
53:09Public policy. Pandora's boxes: property rights and net neutrality. Net neutrality: link on subject. Issue is that telephone companies want to double-dip. Want to charge residential and business subscribers for access to contact sites, places like Google; also want to charge content suppliers for supplying fast access, say Google, you are using a large percentage of network traffic so we want to charge you for using up resources. Friction-free communication. Carriers want to be able to put up arbitrary friction barriers. Per-bit metering, real cost in networks isn't in cables or bandwidths but in intersection points: routers and people to maintain them; doesn't scale with usage. If you try to meter bandwidth, you quickly find that your metering mechanisms are more expensive than increase in margins from the metering. Flat rate peering, flat rate internet connections. Claim: allowing differential pricing will get better investment in this in the capital structure. Sung that song before, give us this market rigging and it will happen. Never happened. Why should government have any role? We don't tell car companies what to charge. Between 1880 and 1920, Peter Abow (sp.?) first honcho at AT&T which became Bell, made double-bargain with U.S. Government: You give us control over the last mile and we'll give you universal service. Bell System appropriated the last mile, creature of government regulatory fiat, want to have it both ways. Ideal solution: take as much radio spectrum and fiber as possible out of regulation and let market operate. Mesh networking might emerge. Buy mesh nodes like pencils. Internet architecture already took a long step toward this with packets. Last mile is being end-run by wireless technology. Google G1 with browser, undercuts last mile any time near a cell tower, no need for telephone company's wiring. Coalition for more regulation: government should enact mandates on what telephone companies can and cannot do. Telephone companies love that, reaffirms the system that locks them into their power. Net neutrality activists, all about the politics.
1:01:47Nostalgic about early 1990s when the net came into non-hackers' consciousness. Good place to do research. Changed our lives in unimaginable ways. In 1994 couldn't have imagined the way it changed. Speculate on next 15 years. What will creative, imaginative hackers create? More and more mechanisms for enabling markets to function friction-free. E-bay; need lots of different e-bays. Some things you can't sell on e-bay. Prediction markets. Robin Hanson. Some ability to harness crowd-sourcing and swarm effects for addressing problems more general than how you make a piece of software work. Appropriately compensating people. [Taping Jan. 8, 2009]. Month away from passing a large stimulus bill. Members of Congress's compensation. Challenge is implementation. Hackers as a group at forefront to undercut monopolies.

Comments and Sharing



TWITTER: Follow Russ Roberts @EconTalker

COMMENTS (31 to date)
Adam writes:

I can't believe you interviewed Eric Raymond! If you interview any more of my heroes I'll start thinking you're tailoring this podcast for my personal tastes!

Ted writes:

Great Podcast this week Russ.
Very good example of how free markets work if only people would give them a try.

gappy writes:

An interesting podcast on Open Source, although I didn't feel that Raymond had any deep knowledge of Economics or Hayek. The description of Open Source is probably instructive for people who have never seen any source code or used gcc. The discussion on Net Neutrality was deeply unsatisfying. First, there was an hidden assumption that, since the cost structure doesn't mimic the revenue structure, than the latter is flawed. Second, the claims on differential pricing are unsubstantiated.

Still, a stimulating conversation nonetheless.

BoscoH writes:

Russ, you could have challenged him on the whole monopoly thing. So what does he think of private property and why is that different from intellectual creations? I'm not arguing here for DMCA or Sonny Bono, but I am saying that we don't have implicit distrust or contempt for people who own their homes and won't allow us to sleep over or take bananas from the pantry. Why the explicit contempt by Eric and open sourcers for those who don't want to adopt their cultural norms?

You are right about the negative correlation between aesthetics and hacking. It is an entirely different skill set from entirely different parts of the brain. What I've found is that UI aesthetic has to be a compromise between what's beautiful and what's possible. To arrive at it, you've got to have an artist draw it up and a developer try to build the mock up. You can see that developers still win in prominent GUI systems and especially Linux by the rectangularization of everything on the screen. It's convenient, but it's not pretty. In fact, it's confusing to those not familiar with it. I've sat down with my grandfather and discovered that he doesn't perceive the "depth" of stacked rectangular windows on the screen. You could probably replicate the same observation with any older person who hasn't become a computer geek.

Russ Roberts writes:

BoscoH,

I don't think Raymond has contempt for private property. In his book he chronicles a whole range of attitudes within the open source movement. He is not on the fringe.

You might be confusing hackers as in geeks vs. hackers as in thieves or vandals.

BoscoH writes:

In the last half of the podcast, there was a constant theme from Raymond: monopolies are bad. And he did say explicitly that open sourcers distrust the closed source model because it gives the closed source developer a monopoly over his code. I'm not calling them thieves or confusing hackers and crackers, but I am saying that a very negative tone in the zeal of their position and ethos reveals itself in that description.

I'm a "closed source" developer. I make good money making products that tens of thousands of people use. I've used and use some open source software in my work. I've dabbled in open source models. I'm a strong believer in freemium models. At the end of the day though, closed source makes monetizing my work easier. I'm all for people participating in open source models if that floats their boats. The part of the ethos popularized by Raymond that strikes a raw nerve is that there is something worthy of distrust in any closed source system. It has the same kind and level of tone of hostility that a critic of open source would strike by asking for examples of originality from the open source community. For example, one "joke" about GIMP back in the day was that it stood for "GIMP Is Not Photoshop" -- poking fun at the GNU trend of left recursive acronym based naming of things, the tendency of open source projects to duplicate commercial products and add nothing original, and mixing up M and N to throw off the literalists.

I read his Cathedral essay a decade ago and was as, if not more, critical of it then as I am now, for the same "live and live" reason. Hearing in this podcast how the monopoly story spills over into his view of network neutrality confirms my perception that the monopoly story is a bit of a boogeyman in the critique of closed source software. Open source is indeed a phenomenon, and it's useful to understand it, but it is not revolutionary, nor will it crowd out closed source. And the sense of moral superiority coming from open source proponents is worn. I guess I thought there was a truce called 5 or 6 years ago. Raymond doesn't seem to have gotten the memo.

Schepp writes:

Russ,

Great job, it is always interesting how all your non-economist guest the balance the competition with cooperation.

Thanks

B. Schrock writes:

Russ,

Excellent podcast. I have been listening to your podcasts for a while and have found every one of them bursting with interesting ideas.

While I am not an economist (I haven't even taken any econ classes yet) I have been living and working in an "open-source" world for years, it is how I feed my family. I was very surprised to hear you introduce Eric in the podcast. Many of his writings have changed my world view over the years, and much of my personal experience has reinforced what I have learned from him.

I did think this podcast was great, but I don't think it captured the subtleties of his views as well as it could have. I think the topics discussed have a lot of moving parts, and from comments on this podcast so far, some people did not fully understand his views.

I encourage everyone to read his essays and truly give them time to sink in.

Thank you for your work on this podcast, I find it a very valuable information source and look forward to listening to it every week.

bluhawkk writes:

Excellent, fascinating podcast. Raymond ended by saying 'Hackers as a group are going to be at the forefront of all efforts to undercut monopolies including government monopolies.'

In light of the policies of incoming administration which seem intent on continuing the expansion of government at a more rapid pace, I would have like to have heard more on the direction of undercutting government monopolies.

Ajay writes:

Let me start off by explaining what source code is for the non-techies, as it's never defined in the podcast. Source code is like the recipe for software. Normally, you follow the instructions in a recipe and make a cooked meal. With software, you take the source code recipe and produce a binary, a collection of 1s and 0s that tells the computer what you want to do in its language. The open source movement is about having open recipes, while developers of the more popular closed source software keep the recipes to themselves so their software can't be easily copied. One might wonder why you can't just take the binary and deduce what the recipe was. You can do this but as the recipes get large enough, into thousands or millions of lines of text/code, it becomes increasingly hard to do.

Getting back to the topics discussed, Eric's mention of a hardware driver, the software that tells your computer how to control a new piece of hardware like a scanner or printer, as a good use of open source is rigged, because most software doesn't come with hardware. Since open source has weak economic justification, open source proponents (at least those that are economically literate, most aren't while Eric is) have to come up with complements that can be charged for, either hardware or consulting/internal services. Eric maintains that most software jobs are in the consulting/internal group but the fact is that even that software niche needs to be standardized and spun out at some point. For example, Intel started off by building their own semiconductor equipment but quickly found it much more cost-effective and higher-quality to outsource that work to outside suppliers like Applied Materials. I am skeptical that the same isn't necessary with internal software but I will admit that open source has a chance to compete in this niche and has been moderately successful in this category so far. However, for product software, where there is no complementary hardware and the people using it are not going to hire programmers or consultants, open source has been a resounding failure, eg operating systems like Windows or Mac OS X or applications like Photoshop or Office. Product software probably accounts for a large majority of software-related revenue, if not most software jobs as Eric claims though it's probably close even in the jobs numbers. Echoing Bosco, open source has largely failed in its revolutionary goals but it has brought to the fore new elements that will be fused into the future of software development, elements like highly decentralized, telecommuting workers and an open development process. One way to accomplish a fusion is through mixed-source licenses, like the Mozilla Public License that Firefox uses, that allow a mix of open and closed source, allowing a synthesis of the best features of both approaches. I have an idea for a mixed-source license of this sort myself that I'm trying to get written up in legalese right now, hence my deep interest in these issues.

Other pushback that I have is that writing software is not intrinsically rewarding to most people and just because open source software only requires time doesn't mean that that time doesn't cost money. The fact is that most open source programmers make a living writing costly closed-source programs and write open source programs on the side as a hobby. Eric and others would like to extrapolate from the open source work they've produced to something revolutionary but it hasn't been and cannot be, because ultimately people have got to get paid and you cannot feed yourself with glory. Another issue he raises is that he bemoans the secrecy of closed source. Well, secrecy is scarcity for information goods. Since you can make exact copies of information products at almost no cost, unlike physical goods like an orange or chair, the only way to get paid for the fixed cost of creating them is to maintain some scarcity, meaning secrecy. What's important is what kind of secrecy you employ, whether it's globally harmful methods like patents for trivial ideas, which are only possible because of our horribly broken patent system, or whether you simply keep all or part of your source code to yourself. My prediction is that the most efficient method is to keep parts of the source closed and parts open, which nobody does today even though there are existing licenses that allow it.

Russell Nelson writes:

BoscoH, I doubt very much that Raymond meant you to perceive or receive any moral drubbing. It's fine with us folks in the Open Source camp if your software exists in a marketplace which supports proprietary software. Go for it! But if it doesn't, then prepare to have your free lunch eaten. We've already eaten the market for C and C++ compilers. Other markets will fall. Perhaps not all. But don't be smug.
-russ

Speedmaster writes:

As an IT guy (developer/tester) by trade I found this episode of EconTalk particularly interesting.

One note on the net neutrality discussion ... it's funny how a non IT pro (Dr. Roberts in this case) would never think of telling an IT expert which OS or language was better, but people not versed in economics rarely seem to hesitate to expound on economics. Something unique about economics in that regard I can't quite figure out.

listener writes:
Getting back to the topics discussed, Eric's mention of a hardware driver, the software that tells your computer how to control a new piece of hardware like a scanner or printer, as a good use of open source is rigged, because most software doesn't come with hardware.
Actually, most of the software that consumers use on a daily basis does come with hardware. Operating systems, web browsers are pre-installed after all. Modern mobile phones have web browsers too. Eric's driver example was ill-chosen.

Open source has failed its "revolutionary goals" on the desktop because of momentum. Consumers expect Windows on desktops and notebooks.

I believe that it will eventually win because of commoditization. Cheap netbooks are the beginning of a trend where Linux gained a foothold.

Eric maintains that most software jobs are in the consulting/internal group but the fact is that even that software niche needs to be standardized and spun out at some point. For example, Intel started off by building their own semiconductor equipment but quickly found it much more cost-effective and higher-quality to outsource that work to outside suppliers like Applied Materials. I am skeptical that the same isn't necessary with internal software but I will admit that open source has a chance to compete in this niche and has been moderately successful in this category so far.
I believe you are missing the point, since spinning out software is precisely where the open source model is attractive to companies. Just look at Apple's decision to adopt the open source KHTML engine for it's own Safari webrowser.

Or look at IBM and Asus. Both, IBM and Asus are contributing to Linux even though they are selling Linux with completely different types of machines (Servers vs. EeePc). Back in the day, companies like IBM either had to develop their own incompatible, proprietary operating system (AIX) or had to rely on a potential competitor (Microsoft) to supply an OS for them. Linux gives IBM the benefits of both approaches. Independence and the ability to customize plus the cost savings of outsourcing by sharing the burden of developing shared infrastructure with other companies (or altruistic hackers).

@Russ:


In his book he chronicles a whole range of attitudes within the open source movement. He is not on the fringe.
Well, Eric Raymond has frequently been accused on slashdot.org of remaking the hacker community in his own image and is considered to be somewhat of a looney because of his fringe views on a range of topics. He frequently claimed for example that hackers are mostly libertarian. I would argue that only a minority are and those are the kind that prevented me from seriously looking at libertarianism for a long time. (meaning economically illiterate libertarianism of the Joe the Plumber variety)

The free software movement of Richard Stallman, which is the bitter rival of Eric Raymond's open source movement, has more support and mindshare. Both basically defend the exact same software model for different philosophical reasons which are considered to be the wrong reasons by the other group.

Here is an actualy quite conciliatory explanation by Richard Stallman: Why “Open Source” misses the point of Free Software
Think "People's Front of Judea!" vs. "Judean People's Front". ;-)

Wow! What fun to listen to this meeting minds. I think of myself as being a member of an even smaller sub-sub-culture within hackerdom -- perhaps I could call myself "a Hayekian Hacker". Mark S. Miller is my role model. (Robin Hanson is pretty good, too.)

Ajay writes:

listener, I think you mean that some of the software that is used most comes with hardware. I thought about the fact that operating systems like Windows usually come with hardware but listed the open source OS as an example of failure anyway, since it is a failure. I didn't go into the reasons why it fails despite coming with hardware, though I have commented on why this is at Eric's blog before. The main reason open source fails with product software is because it doesn't scale, which is funny as Eric regards scaling so highly. Product software is very successful and makes so much money because you make a product once and copy it many times over, it scales. However, you can't do this with the GPL, as you can see with the dozens of linux distros, because any distinguishing feature of one distro can easily be copied by all the rest. That plus the fact that you cannot charge for the actual product, since you cannot restrict copying, makes the most profitable segment of software, product software, off-limits to open source. Yes, Windows has a stranglehold on the desktop but one would expect open source OS's to do better than they have, given their advantages, if the model were working. The word commoditization is probably misapplied to software since software is fundamentally different, in how malleable it is and how much it changes. It has been refreshing to see the open source OS's make headway on the new mobile hardware platform but I predict they will be defeated by either mixed or closed source software soon enough. It is a battle open source cannot win. As for the web browser, a free product that has stagnated because of standardization by committee- I dunno when the software progressives will figure out that this committee standardization of software is useless- is neither here nor there, who cares?

I believe you are missing the point about companies spinning out software. Of course, Asus and IBM can spin off their modifications into linux, the problem is that there are unlikely to be third parties that add important features to this software. IBM or Asus might add features, in the course of tailoring linux to particular uses, but no third party has an incentive to do so under the GPL as their modifications can immediately be taken by somebody else. The question ultimately is whether this consulting model can compete with the property rights of closed source. We have seen the battle of property rights vs the commons many times before and property rights have usually won resoundingly. What we're grappling with today with software licensing is: what are the optimal property rights for this new kind of property? I believe it will be mixed-source licensing. Ultimately, you speak as someone who is in favor of open source and seem to view me as against. Rather, I believe that both closed and open source have their strengths and would like to see those strengths comingled to create something far better. I have an idea of what that best mixed-source license might look like, and somebody associated with FSF Europe might soon help me translate it into legalese. :)

Ajay: do tell! I too have been working on a social/economic/legal hack in the attempt to combine the merits of open source and commercial cooperation. The idea that I hit upon was a straightforward combination of two of the greatest hacks ever: the Time-Limited Monopoly (as implemented e.g. by the authors of the U.S. Constitution "to promote the progress of science and useful arts"), and the Transitive Public Licence (due to Richard Stallman).

Here is my idea nicely written up by my friend Ping Yee: An Open Source Licence Idea

BoscoH writes:

Russ Nelson writes:

But if it doesn't, then prepare to have your free lunch eaten. We've already eaten the market for C and C++ compilers. Other markets will fall. Perhaps not all. But don't be smug.

This statement hits every nerve possible. FWhy the "we"? When did Microsoft start using gcc in its development products? Why is it the aim of the open source community to eat commercial developers' lunch? Out in the commercial world, most people I know want to create great products and serve their customers. Eating the competition's lunch might permeate the mantra of a sales department, but it's not how long term players think.

A random point about the continued viability of closed source models... Tim Cook, Apple's acting CEO, laid out an Apple manifesto in his conference call yesterday. Read about it here. A salient point was "We believe in saying no to thousands of projects so that we can focus on the few that are meaningful to us. We believe in deep collaboration and cross pollination in order to innovate in a way others cannot." Where is the cross-pollination in mainstream open source? Why, for example, can't an Android phone preview and control Impress presentations? Apple, a closed shop with only 70,000 or so eyes, managed to make iPhones and iPod Touches into interactive presentation helpers for Keynote. Oh, for $.99, downloaded and installed in jut few taps.

Bryan MacKinnon writes:

I propose that open source is best suited for purposes that are well established and commoditized. Mozilla Firefox is a good example of this. Applications that have been in the market place for a while and whose authors have made a reasonable profit, are also good candidates since it would be only a matter of time before someone else develops a competitor.

For sophisticated consumer purposes such as photo editing, I have yet to see an application that is as good as what you pay for.

I've contributed to both the open source movement and written a lot of software professionally - there is room for both.

Steve Spiller writes:

Russ - I was taken with Eric's comment that one reason why engineers dislike monopolies is (1) it creates a single point of failure and (2) that monopolies tend to become obsessive about secrecy and control.

Question - Might it be wise for the EU to be a bit distrustful of Russia's recent conduct supplying natural gas? Maybe they should diversify their natural gas supplies.

As you say "it's just a thought"

listener writes:
Ultimately, you speak as someone who is in favor of open source and seem to view me as against.
Oh no, I am not ideological on this at all. Or rather, I used to be ideological five to ten years ago, back when there was a real threat by companies like Microsoft trying to corner the Internet with proprietary standards, or when the EU was pushing software patents. But these threats are pretty much gone. I don't care what other people use, as long as nobody takes Linux away from me.

I merely believe that Linux and OSS will gain significant ground for purely economical reasons.

You refered to consumer software products. Nobody is stopping anybody from developing proprietary applications for linux. I am in fact using the closed source Opera web browser right now.

I was thinking about commoditization of hardware. The race for more Ghz and more GB is pretty much over. This is where behavioural economics kicks in. The same OS that was considered to be acceptably priced when it came with a 1000€ machine, seems outrageously expensive if PCs cost only 200€. We have already seen that Microsoft was forced to reduce licensing cost for the Netbook market, or else it would have been completely left out.
And Netbooks are just the beginning. Asus is now shipping its motherboards and notebooks with a fully functional, instant-on embedded version of Linux that comes with a web browser, Skype, etc.

Soon, we will expect any 100€ LCD screen to have a fully functional PC inside, with an instant-on OS that provides most of the software you need, including an office suite, an itunes like media application with HDTV playback, 3D graphics and enough embedded storage space for regular use.

Many once great hardware and software companies have gone the way of the dodo bird due to commoditization, or have been reduced to niche vendors. (SGI, DEC, Novell, SCO and now Sun)

Microsoft and Apple are next in line and they have wisely diversified their business (iPod, iTunes Store, XBox, Zune, web applications, etc.)
Also notice that Apple is pretty much selling shiny PCs nowadays.
I wouldn't be too surprised if Microsoft stops developping Internet Explorer either.

emerich writes:

The timing of this podcast was, for me, ironic. Firefox was cited as a leading example of open source software's superiority over closed source alternatives; literally the day I listened I had finally decided, after four years of using Firefox, to switch from Firefox to Internet Explorer because Firefox has for months now been crashing about every 15 minutes. Of course, before closing it asks me if I want a "report" of the crash to be sent to Firefox, to which I alwyays say yes, and which never results in any response or improvement in stability. So far IE has been steady as she goes...

MikeR writes:

Thank you for another great podcast.

Eric Raymond put a lot of emphasis on HIS importance to the open source movement and this, I believe, hurts his argument and his credibility.

Raymond stated, "... before I articulated the concept, it only existed subconsciously in the minds of the participants." I expect that many "participants" would take offense to such an arrogant proposition. For a bottoms-up, Hayekian guy, Raymond is sure happy to take the credit.

I run for the hills when I hear politicians, and especially journalists, make such arguments which have ad hoc narrative fallacy written all over them.

Roger Avalos writes:

Eric Raymond also has a small part in this xkcd:
http://xkcd.com/225/

AndrewNYC writes:

I really enjoyed this podcast. I just discovered EconTalk after you won the best podcast award. You definitely deserve it!

SteveO writes:

As I was listening to the show, something clicked in my head. Government "running" the economy is like a monopolistic, closed-source software provider. Free markets are like open-source projects, with a much larger scale of people contributing to the information and iterative testing of the project.

Nice.

Kent Beck writes:

Thank you for hosting Eric Raymond. He made an interesting contrast with your earlier show with Chris Anderson. Something bothers me about the perspectives of both guests, however.

As I understand economic theory, the price of a good in a free market falls to the marginal cost of producing that good. Raymond and Anderson both assume that since the marginal cost of software and similar electronic goods is zero, the price is reasonably set at zero.

Since the resources required for development are scarce (programmer or author time), all the beautiful information generating power of the market is abandoned once the price drops to zero. In Raymond's model, programmer whim is used to allocate programmer time. As free software has evolved, large companies also allocate programmer time, not to create the most valuable software possible, but to deprive competitors of profits from markets in which they are weak.

All of this seems to me to get away from the basics of economy--allocating scarce resources efficiently by getting paid for doing something valuable. Is there a way to look at "free" electronic goods that results in a reasonably efficient allocation of scarce resources?

Ajay writes:

OK, I was inspired by Zooko's interest to put my mixed source licensing proposal online in blog format, that I've been pitching to various tech luminaries recently. The summary is that a mix of closed and open source is better than either extreme alone, and that closing source for a short and limited period of time is better than the current extremes of immediately and completely open source, like the GPL, vs completely closed source for a 100 years or more, like the licenses that Microsoft and other proprietary software companies use. I'd appreciate any feedback from Econtalk commenters, please post comments there.

listener, while I agree that constant hardware scaling is a hugely important force, that many take for granted nowadays and that I often remark on in wonder and hope, I don't think it will lead to much open source adoption, because of the aforementioned economic problems. When open source is competing with property rights, whether from a mixed or completely closed source app, it cannot compete because the incentives for innovation and improvement are not aligned in its favor. I disagree that your listed vendors lost because of hardware scaling, I posit that they lost because their platforms lost and they were unwilling to switch. Windows was the winning OS and x86 the winning computer architecture, they were unwilling to adopt those platforms so they shrunk, while HP, Dell, AMD, and others did fine. Microsoft and Apple have diversified but it's interesting that only the iPod makes money, of all the new items you list. I see no indication that MS will drop IE, their plan seems to be to own the browser completely instead, by positioning silverlight as a flash/javascript replacement.

Kent, I don't think any real economics analysis would ignore fixed costs entirely in price-setting, the argument is probably usually that marginal costs tend to dominate. However, Chris and maybe Eric run with this and try to assert that digital goods will be priced at just below their marginal cost, at zero, even though you could never pay for the non-zero fixed cost that way. The fundamental problem for digital goods is that once you divide the fixed costs by the amount of goods, you get a price of a couple cents. Since there is no current payment system that easily enables such a price, ie a micropayments system, people either try to make do with advertising or give away the goods for free. Giving them away for free is only possible because idiot VCs fund it in the hopes that they can capture the market and start extracting monopoly rents someday down the line, with no conception of how they will actually do so. They look at how the newspaper business was able to extract monopoly rents for decades, and possibly other businesses like TV too, and assume that they will come up with some similar formulation some day, or they sell the startup to some idiot corporation with such a justification (skype, zimbra, mysql). This is a fool's game that has unfortunately flourished while we haven't built a working micropayments system all these years. Once we build out a working micropayments system, all this vacuous nonsense ends and real progress begins.

Padraic writes:

Great interview. I have really enjoyed your focus on technology thinkers - Shirky, Anderson, etc. Please keep them coming!

Ethan writes:

I cringed quite a bit. So Raymond claims that real hackers don't use handles. Who is he to say this? I've known MANY great people that used handles. Just because he couldn't think up a good one doesn't give him the right to put down everyone else's clever handles.

He also dumps on software crackers, that they are not hackers. He claims hackers are those that write code. Similar to the classes taught at every college and most high schools. Software crackers that reverse engineer the byte code of an application, removing challenge - response hardware dongle checks. That takes quite a bit of skill that generally isn't taught in high schools. I'm not saying illegal use of software is okay, but those same skills are what people use to reverse engineer malware and viruses and write fixes and patches. To put them down because you (and I) can't do it is wrong. Sorry, I have more respect for those skills versus writing the ncurses library.

In the end, the open source that is wildly successful is that which interferes with large companies revenue stream. RedHat versus Microsoft server products. MySQL versus Oracle. Apache versus the commercial servers that used to exist (Netscape and MS and whoever else). I use open source, I use closed source. But in the end, I don't really respect Eric Raymond.

Kind of how I have little respect for economics people who didn't identify the oncoming bubble 4+ years ago. Since I don't have a background in economics and it was obvious.

You should do an episode on the bubble blogs, and how they had it 3 to 4 years before mainstream media did. They had the facts, they had the figures. Pointers to gov't data that was clear evidence of what was going on.

Ray G writes:

Two observations:

Prof Roberts said that "some" in the hacker community were simply against the idea of the commercial product, and all that goes with it, and the author immediately identified himself as being one of the "some."

Secondly, that programmers are aesthetically retarded is a foregone conclusion to everyone but programmers - hackers or not.

I listened to Hanson and Roberts on biases before I listened to this one so it was all the more entertaining.

It was a very interesting interview, and I will read the book. Raymond is an interesting guy with much insight despite his inability to see some very obvious things about his own sub-culture.

One last observation, even though Raymond failed to see Apple's superiority in bringing together an end-consumer product, (or at least why it is superior) it is the down in the trenches programmer writing code for guys like me at work that are probably making the larger impact on society as a whole.

Keith Hopper writes:

Best econ talk yet. I was looking forward to the discussion on intellectual property that got tabled. Any chance of revisiting this with Raymond or others?

I think that Raymond's historical description of the success of the OSS approach is now a little dated. I'd love a reassessment (beyond software) of online collaborative efforts that found inspiration in the OSS approach but have learned much since then around how to apply the approach to different modes of peer production. Shirky touched on this. Maybe bring in Yochai Benkler?

Comments for this podcast episode have been closed
Return to top