What this covers

Eric Raymond and Russ Roberts discuss why open source software has succeeded where traditional hierarchical development has stalled. Raymond, author of The Cathedral and the Bazaar, walks through the structural and economic reasons peer-reviewed, decentralized code development outperforms closed-source alternatives at scale. The conversation ranges from first principles in software engineering—why aesthetics matter more than mechanics at the highest levels of programming—through the sociology of how organizations corrupt their own information flows, to the economic basis of open source as a capital-goods phenomenon rather than a merchant-market product.

The terrain widens considerably. Raymond traces the Snafu Principle, which explains why peer review fails inside hierarchies but thrives in open source, and defends the hacker reputation system against the charge that it ignores most contributors. He argues that software engineering historically anomalously hid its experimental designs (source code) to extract secrecy rents, violating a Hayekian principle about distributed knowledge that science and open source exploit. The conversation turns contentious on net neutrality, where Raymond contends that regulation has created rather than solved the problem, and on Wikipedia, where he argues the absence of an objective success metric prevents crowdsourcing from functioning. He also addresses where open source frays: user interface design, where the lack of objective metrics has forced projects toward benevolent dictators rather than committees.

Sharpest takeaway

Raymond argues that open source's superiority comes from decentralized peer review plus a specific set of economic conditions under which non-monetary motivation ('love') scales, and that most software is a capital good not sold for money, so open source's inability to be sold directly doesn't matter.

  • Decentralized peer review ('with enough eyeballs, all bugs are shallow') beats all other quality-assurance methods at large code-base scale
  • Love/intrinsic motivation only scales when four+ conditions hold: cheap capital goods, creativity as the limiting factor, intrinsic reward, an objective success metric, and near-zero communication costs
  • Most software is internal capital good, not merchant-market product, so open source being unsellable is economically irrelevant

The claims · ranked38 claims · weighted by value

This asset isn't compiled yet

You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.

0.79

Programming only looks like a mechanical, logic-saturated activity at its lowest levels; as programmers improve, the mechanics become routinized (good programmers learn whole languages in days, like a saxophonist picking up the clarinet), and what increasingly distinguishes top performers is aesthetic sense, because good taste correlates in mysterious but definite ways with stability, performance and robustness—so at the highest levels aesthetics is all-important.

causalhigh valuecontestednovelty 4/4durability 3/4· Eric Raymond

“as you get better at programming a lot of that stuff becomes quasi-automatic. It becomes routinized.”

0.79

Decentralized peer review—captured by 'given a sufficiently large number of eyeballs, all bugs are shallow'—trumps every other software quality assurance method, a principle Linus Torvalds operated on before it was consciously articulated.

causalhigh valuecontestednovelty 4/4durability 3/4· Eric Raymond

“decentralized peer review basically trumped everything else as a quality assurance method. Everything else.”

0.78

The Snafu Principle: in any hierarchical authoritarian organization, inferiors are more consistently rewarded for telling superiors what they want to hear than for telling the truth, producing a progressive disconnect between those at the bottom who can do but not decide and those at the top who can decide but not do—and this is why peer review within one organization fails while open source peer review works.

causalhigh valuecontestednovelty 3/4durability 4/4· Eric Raymond

“in any hierarchical authoritarian organization inferiors will be more consistently rewarded for telling superiors what they want to hear than they will be rewarded for telling the truth”

0.78

The telecom market's distortion is a historical contingency originating from Theodore Vail's roughly 1880-1920 devil's bargain with the US government—'you give us control of the last mile and we'll give you universal service'—after which the Bell System squatted on the last-mile monopoly as if it were a property right rather than a creature of regulatory fiat.

factualhigh valuecontestednovelty 3/4durability 4/4· Eric Raymond

“there was a guy named Theodore Vail... He said, 'You give us control of the last mile, and we'll give you universal service.'”

0.74

The freedom to fork an open source project, though rarely successful, is very important because it keeps project leaders on their toes and prevents them from becoming autocrats.

causalhigh valueestablishednovelty 3/4durability 3/4· Eric Raymond

“The freedom to fork is very important. It's what keeps project leaders on their toes and keeps them from becoming autocrats.”

0.74

The Cathedral-and-Bazaar essay is a reply to Fred Brooks's The Mythical Man-Month: Brooks advocated controlling software complexity via small teams, tightly defined objectives, exacting central control, and long release cycles spent squeezing out bugs—exactly the practices the Linux community violates by using maximal feedback and very short, sometimes daily release cycles.

factualhigh valueestablishednovelty 3/4durability 3/4· Eric Raymond

“I was referencing a book by Fred Brooks called The Mythical Man-Month... there's one level on which my essays are a reply to The Mythical Man-Month”

0.73

Internal peer review in closed-source companies is dysfunctional because both reviewing and reviewed engineer groups ultimately report to the same people, creating subtle pressure to minimize problems—and the problems that get minimized are exactly the ones customers later trip over.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“the whole process is subverted by the fact that ultimately both groups of engineers are reporting to the same people. Right. So, there's all kinds of subtle pressure to minimize problems.”

0.73

Net neutrality activists who push to solve the problem with more government regulation miss that such regulation created the hole in the first place; telcos love these activists because every call to regulate reaffirms the system that locks telcos into power, letting them keep margins up by buying another handful of regulators.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“they completely miss the fact that that kind of regulation is what got us into this hole in the first place.”

0.73

Once a code base exceeds a certain scale, traditional verification methods like formal proofs of correctness and structured walk-throughs run out of steam, and only total process transparency with distributed peer review works—though that too will eventually run out of steam (requiring strong AI to debug), since there is no silver bullet.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“the size of the average code base has reached a level where nothing short of total process transparency and openness of source actually works”

0.73

Hackers dislike monopolies primarily not for ideological reasons but because, as engineers, they are deeply suspicious of systems with a single point of failure; monopolies also tend toward obsessive secrecy and control that blocks problem-solving, and secrecy is the enemy of quality.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“we don't like monopolies because we're engineers and we've learned to be deeply suspicious of systems with a single point of failure.”

0.73

Hacker reputation follows a power-law distribution, with a few people who are everywhere; small patch contributors are usually motivated not by glory but by efficiency—shipping a fix to maintainers is cheaper than maintaining their own branch.

factualhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“there's a sort of power law distribution, there are a few people who are everywhere”

0.73

Wikipedia fulfills all the conditions for crowdsourced production except an objective metric of success: there is no metric for truth in an encyclopedia entry, only the judgment of other humans, so crowdsourcing produces 'thrashing'—and factual accuracy alone is not sufficient to make a good entry.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“The only one they fail is there's no objective metric for success.”

0.73

Per-bit metering of network traffic is economically unsustainable because the real cost in networks is not the cables or bandwidth but the intersection points (routers and the people who maintain them), which do not scale with usage, and because metering mechanisms tend to cost more than the extra margin they generate—which is why flat-rate peering and connections remain economically dominant.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“the real cost in networks isn't the cables. It isn't the bandwidth. It's in the intersection points, the routers and the people needed to maintain them. And that doesn't actually scale with usage.”

0.72

Open source has a strong taboo against forks because forking is inefficient (it divides the community's resources, fixes don't propagate across the divide) and because there is a customary-law system, like common-law property rights, in which hackers 'farm' their software for reputation gains; making off with someone's patch or removing their name is the ultimate taboo.

causalhigh valuecontestednovelty 4/4durability 3/4· Eric Raymond

“there's a there's a sort of customary law almost like common law property rights in the community. I analyzed this in my second paper Homesteading the Noosphere”

0.72

The most important economic fact about open source is that it is not a way to make money but an enabler for other money-making activities: bundling open-source-licensed source code with hardware (e.g., an Ethernet card) raises the hardware's value because customers no longer price in the risk of the vendor going out of business and can rationally expect lower defect rates from peer review and larger volunteer communities.

causalhigh valuecontestednovelty 4/4durability 3/4· Eric Raymond

“Open source is not a way to make money. It's an enabler for other activities that make money.”

0.72

Profit motivates economic production far more reliably than goodwill or love—captured by Walter Williams's point and the engineering concept that 'love doesn't scale' as a solution to economic problems.

causalhigh valuecontestednovelty 2/4durability 4/4· Eric Raymond

“what Walter Williams was pointing out was that love doesn't scale. Correct. As a solution to economic problems.”

0.72

Most software and software expertise is not deployed for sale on the merchant market; the overwhelming majority of programming jobs are for internal-use software (database schemata, report generators, COBOL banking programs) because software is a capital good used to solve business problems—which is why open source's inability to be sold directly does not matter.

factualhigh valuecontestednovelty 4/4durability 3/4· Eric Raymond

“most software and most software expertise isn't deployed for sale on the merchant market. Mostly it's deployed because people want to solve business problems”

0.71

Both closed and open source require an architect or small group of cooperating architects; the difference is not the presence or absence of architects but the social machine around them and how it reinforces and checks what they do—directly analogous to the central-planning-vs-market debate, where markets also have planning, just done by individuals rather than top-down.

causalhigh valuecontestednovelty 3/4durability 4/4· Eric Raymond

“all good software needs to have a an architect or a small group of cooperating architects at the center. That's something that's constant across both closed and open source.”

0.71

The open source insight is the application of a Hayekian idea—that prices/markets contain distributed knowledge no single participant can possess—and is analogous to scientific peer review, where results are not accepted until experimental designs are reviewed; software engineering was anomalous in hiding its 'experimental designs' (source code) to protect profit streams.

causalhigh valuecontestednovelty 3/4durability 4/4· Eric Raymond

“It's a Hayekian idea and my debt to Hayek is very explicit.”

0.68

Closed-source software development historically relied on small secretive teams hiding source code, which was tied to a desire to collect secrecy rents on shipped products, whereas hacker culture developed a tradition of collaborative source-sharing and maximal peer review.

factualhigh valueestablishednovelty 2/4durability 3/4· Eric Raymond

“not showing the source code to anybody was frequently tied to a desire to collect secrecy rents on a product that you shipped”

0.68

The mainstream media use of 'hacker' to mean someone who commits computer crimes and breaks security is corrupt and wrong; properly a hacker is a member of a culture existing since about 1960 that builds and gives away software, including the internet, the World Wide Web, Linux and open source.

definitionhigh valuecontestednovelty 2/4durability 3/4· Eric Raymond

“properly, a hacker is a number of a culture or a connected group of cultures that has existed since about 1960. Uh and hackers build programs and give away software.”

0.66

Open source was an unconscious folk practice for decades; articulating an explicit theory of openness, transparency of process, and peer review allowed practitioners to reflect on and improve their process, which is why the movement caught fire quickly.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“we had an unconscious folk practice, and I was the person who made it conscious, who articulated the ideas about openness and transparency of process and peer review”

0.66

Open source projects typically start not because proprietary alternatives are expensive but because a developer has an 'itch' and finds it intrinsically cool to scratch it; if they are good and lucky, others see the code and improve it with patches—so the generative motive is intrinsic interest, not cost-driven substitution.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“open source projects typically start because somebody has an itch and they feel like scratching it.”

0.66

User interface design is the area where the objective-metric condition is hardest to apply, so committee-based open source methods work less well there; the open source culture is responding by appointing 'UI dictators' within projects, having learned that a flawed individual vision beats design by committee.

causalhigh valuecontestednovelty 3/4durability 3/4· Eric Raymond

“we are gradually evolving a practice in which we essentially appoint UI dictators within each project.”

0.63

The telco claim that differential pricing privileges will fund broadband investment is false: telcos have repeatedly promised fiber buildout in exchange for rate privileges and never delivered, instead using the rents to build their monopoly walls higher, buy legislators and bribe regulators, making the situation progressively worse.

causalhigh valuecontestednovelty 2/4durability 2/4· Eric Raymond

“They said, 'Oh, give us these market rigging and these rate military privileges and the ability to raise rates, and we'll build fiber out to everybody'... And it's never happened.”

0.63

The net neutrality issue arises because telephone companies want to 'double dip'—charging both residential/business subscribers for access and content suppliers like Google differentially for fast access—and Raymond argues allowing this is a bad idea because friction-free communication is what makes the net valuable.

normativehigh valuecontestednovelty 2/4durability 2/4· Eric Raymond

“the issue in net neutrality is that the telephone companies want to double dip.”

0.62

Over the next 15 years hackers will create more mechanisms making markets effectively friction-free—needing many competing, federated eBays so no monopoly can lock in the terms of sale for particular devices—and prediction markets will become important because they harness crowdsourcing/swarm effects for far more general problems than software, especially as communication costs approach zero and if participants can be appropriately compensated.

forecasthigh valuecontestednovelty 3/4durability 2/4· Eric Raymond

“more and more and more mechanisms for enabling markets to be effectively friction free. eBay and things like it are a long step in that direction.”

0.61

The ideal solution to the telecom monopoly is to deregulate as much radio spectrum and fiber as possible and let the market operate; what would likely emerge is decentralized mesh networking with cheap, intelligent self-organizing, self-healing routing nodes that optimize the network bottom-up with no central control choke points.

forecasthigh valuefringenovelty 3/4durability 2/4· Eric Raymond

“the ideal solution is to take as much radio spectrum and fiber out of regulation as possible and let the market operate.”

0.59

Non-monetary motivation ('love') scales into productive work only under a specific set of conditions: (1) the capital goods to do the work are cheap, (2) the limiting factor is human creativity and attention, (3) the work is intrinsically rewarding, (4) there is an objective metric for success, and (a fifth, taken for granted) near-zero-cost communications; if any condition fails the system works poorly.

causalhigh valuespeaker onlynovelty 4/4durability 4/4· Eric Raymond

“One, the capital goods required to do the work are cheap. Two, the limiting factor on the work is human creativity and attention. Three, the work is intrinsically rewarding. And four, there is an objective metric for success.”

0.59

Unregulated wireless spectrum is the solvent that can break the telco last-mile monopoly: a fully internet-capable wireless device like the Google G1 makes the wireline company's ownership of fiber to the house far less important, because internet access is available anywhere near a cell tower.

forecasthigh valuecontestednovelty 2/4durability 1/4· Eric Raymond

“it undercuts the last mile monopoly that the wireline telephone companies are squatting on.”

0.58

Apache runs somewhere between 60 and 70% of the world's websites, and is especially prevalent on sites requiring 100% reliability and high performance, so the more demanding the web application, the more likely it is hosted on Apache.

factualhigh valueestablishednovelty 2/4durability 1/4· Eric Raymond

“Apache hovers somewhere running depending on whose market share figures you believe, it's running somewhere between 60 and 70% of the websites in the world”

0.43

The 1880-1920 era and the early-1990s internet explosion are real periods; for the general public the internet became a reality around 1993-94, but the hacker culture was using it effectively about 15 years earlier, with access limited to large universities, military and military contractors.

factualestablishednovelty 1/4durability 3/4· Eric Raymond

“for the general public, the internet became a reality of in the early 1990s. I'd say around '93, '94 was the big internet explosion. For the hacker culture, we were using the internet fairly effectively 15 years before that.”

0.41

The term 'hacker' was hijacked in a way analogous to the word 'liberal', which in the 19th century meant belief in individual freedom and limited government but has come to mean nearly the opposite.

factualcontestednovelty 2/4durability 3/4· Russ Roberts

“It reminds me of the appropriation of the word liberal which in the 19th century meant a belief in individual freedom and limited government and has since come to mean something close to the opposite”

0.26

Hackers as a group will be at the forefront of all efforts to undercut monopolies, including government monopolies.

forecastspeaker onlynovelty 1/4durability 2/4· Eric Raymond

“one thing you can count on is that hackers as a group are going to be on the forefront of all efforts to undercut monopolies including government monopolies.”

0.26

Even the hacker culture that did imagine the internet's broad future failed to anticipate all of its second-order effects.

factualspeaker onlynovelty 1/4durability 2/4· Eric Raymond

“Even we didn't anticipate all of the second-order effects, though.”

0.26

Russ Roberts is skeptical that the impending stimulus bill will work and suggests members of Congress should be compensated based on objective aggregate economic measures (like GDP or unemployment) rather than on their motives or desire to be seen as helpful—though implementing this is blocked by the monopoly's unwillingness to change the rules of the game.

normativespeaker onlynovelty 2/4durability 2/4· Russ Roberts

“I wish that the members of Congress were compensated based on say some aggregate economic measures... rather than their motives and their desire to be seen as helpful.”

0.25

Firefox is the best-known open source program to non-technical people and has been gaining browser market share because it is simply better, including a beautiful and pleasant user interface—making it the exception among open source products that are otherwise consumed mainly by hackers.

factualestablishednovelty 1/4durability 1/4· Eric Raymond

“probably the best known open source program at this point is Firefox... that has been making great strides in market share because, well, it's better.”

0.18

Computer games are the exception that proves the rule—they genuinely must be sold on the merchant market—but they are very short-lifetime products employing very few people (Raymond estimates no more than three or four thousand full-time computer game programmers in the US), with tight profit margins due to short product lifetimes and ruthless competition.

factualspeaker onlynovelty 1/4durability 1/4· Eric Raymond

“The one exception is computer games. Uh computer games really do have to be sold on the merchant market.”