Brad DeLong's Semi-Daily Journal (2004)

I think there's a distinction between open source software and free services. If a developer decides to stop working on a piece of software, the software doesn't go away. And because of the open-source license, someone else can then pick it up and work on it.

In that sense open-source software is likely to be more reliably maintained than a niche commercial product, which can easily be cancelled by a corporate shake-up, even though it might be profitable. Not all individuals, or companies, are rational actors.

I would also note that "users" here is too broad a notion. IBM does not sell to "users" like Microsoft or Apple does. Rather, IBM sells to IT departments (the majority of their business). IT departments are not always so concerned about user experience as they are on "does it work" and "is it priced right".

In this arena, Open Source is tops and will only get better. Micrsoft is not worried about Open Source desktops; it is worried about Open Source OS's like Linux kickin' its patootie on 70%+ of web servers.

> That set of social expectations is clearly not held by everyone. And I think that such clashing expectations post a danger to the long-run health of the open-source movement.

My take: that set of social expectations *is* held by the elements of the open-source movement that has been successful in the past, and the failure to hold them in mind while working on open-source software is the common thread uniting all the cases of failed open-source projects.

The open-source software phenomenon really is a competitive marketplace of ideas. Some of the sellers go bankrupt, and some of the sellers win big. What it needs for growth is a better business model— one that makes enough money available for paying all the rest of the support staff that a good software project needs. Without that, it will continue to kick around in the margins of the software world, producing only the occasional revolutionary new application.

Personally, I think the most serious threat to the open source movement is persecution by the State. (But they can have my development tools when they show up and demand them at gunpoint. Good luck on that.)

True, "users must anticipate that the open-source software will continue to be developed and improved." Similarly, for it to remain viable, users must anticipate that the closed-source software will continue to be developed and improved.

In the closed-source, copyrighted world, this truth erects implicit barriers to entry for small competitors and encourages oligopoly. In the open-source, copyleft world, this truth guarantees that the successful solution (e.g., Linux, Apache) will gain similar status.

The big difference between the two is that the copyrighted solution not only results in the use of solutions that are not ideal, but also in ground rents for the successful oligopolists. At least the copylefted solution, once established, can be rewritten by paid and unpaid contributors until it's as good as the better solution which lacked for constituents.

As for persecution by the state, if that's code for "software patents" I'm right there with you. Patenting math, the DMCA and extended copyrights have been a huge takings that are in the process of destroying a huge potential commons without replacing them with anything but comparatively small profits for the few, the connected and the lucky.

Open source software didn't need deep pockets to flourish before. The hooppla about Linux -- and all the money going into it -- aside, Apache was the dominant web server when it was just 30 guys hacking at it in their spare time.

The only thing it takes for an open source project to survive is 1 person interested in using it and improving it. If I write a peice of software and abandon it, but people still use it that's all well and good. I think you're comparison to Weblogs.com, however, is flawed because as a service it represented a recurring cost. I still get emails from people using a bit of open source something or other I wrote 5 or 6 years ago, and haven't thought about in just as long, but their continuing to use the software doesn't *cost* me anything and there are no pockets at all backing it up.

Migration assistance and all that, yes, may very well be more the purvue of the VAR, PSI set, and yes is a contract with a customer. However, I don't think it really has much to do with the long term viability of open source software. Really, aside from Mozilla, there's not a whole lot of open source software that doesn't have sysops/developers *as the user base*. Maybe that will change if the OOo stuff actually takes off, but I think much of what your are advocating for doesn't really apply to where open source really shines anyway.

Customer confidence is obviously an issue but hardly one that puts an open source solution at a disadvantage. Exactly the same issue applies to any software package but with an open source one the client has a variety of potential suppliers and will be less dependent on the health of the supplier.

Now if IBM stopped supporting linux and apache and apple stopped supporting FreeBSD in favour of closed solutions something would be lost but that doesn't seem to be the way things are going at the moment. I expect that increasing software complexity will only favour this development by raising the barriers to creation of a private alternative.

It still seems hard to see the failure of a free service using closed software reflects worryingly on paid for services using free software.

IT departments are not always so concerned about user experience as they are on "does it work" and "is it priced right".

Those must be the IT departments on Mars, 'cause they sure as hell aren't the ones in any of the dozen or so Fortune 500 companies I've worked with, where the deciding factor is usually some combination of nepotism, incompetence and kickbacks.

I still don't think you've made your argument Brad, in that your thoughts about open-source don't really connect to Winer's situation. Not that your ideas on open source are wrong; I can't really tell yet. But I don't see the connection, except that it gave you a peg to hang your thoughts on.

I still don't think you've made your argument Brad, in that your thoughts about open-source don't really connect to Winer's situation. Not that your ideas on open source are wrong; I can't really tell yet. But I don't see the connection, except that it gave you a peg to hang your thoughts on.

I do not know why it double posted... I only clicked once! Sorry...

wcw> [...] As for persecution by the state, if that's code for "software patents" I'm right there with you. [...]

As long the scare quotes around "software patents" mean what I think they mean, then it's in the constellation of things that has me worried about "persecution by the State". Stallman's cautionary _The_Right_To_Read_ is the essay that covers the most ground in the area I am trying to point out.

Note: I didn't offer an opinion about how *big* a threat I consider "persecution by the State" to be.

I'm really not following this.

In high tech one is always making trade offs between capturing the advantages of being an early mover and the risks that you'll bet wrong and get stuck with large switching costs or worse. There isn't any safety in IBM or any other single actor. OS/2 or the PC/jr are fine examples of that!

At each fork in the road the choice you make playing in these markets is full of risks. One way to temper that risk is to consider who your fellow travelers are. When things go wrong you can then hope your fellow travelers will join with you in common cause to temper the costs of dealing with that.

Some people prefer IBM or Microsoft or Oracle as a fellow traveler. The longer in one spends in this industry the more you value having a more diversified group along for the trip.

But again. Weblogs.com was never an open source thing. It was always part of the loss leaders of a commercial enterprise. Much like the way a grocery story doesn't charge you admission. Or maybe more like how Microsoft gives away a "web browser" to lock-in market share and block the build out of open internet standards.

"a mindo"?

Why do there need to be social expectations about any software? As long as there is a competitive marketplace, open source should be embraced by at least some of the players in the market, at least indirectly.

The "deep pockets" argument seems more a response to the structure of the US software market than any inherent logic to OSS development. Its also worth remembering that US citizens are much less active contributors to open source projects than those of most other advanced industrialized nations (and particularly Europe). And where are the deep pockets in Hungary?

How is open source software different from proprietary software in this respect? If anything, the situation is worse for closed source software. If IBM drops Linux, it will continue to be maintained and improved, albeit at a slower rate. If a proprietary product is killed by its vendor, that's the end. No more upgrades, no new features, and it could be broken irreparably when Microsoft releases their next service pack.

I'd say that a Linux user has a lot less to worry about regarding future support for their OS than a Mac user does, for example.

Also, you still seem to be confusing a free service with OSS. An open-source program cannot be shut down. The worst that can happen is that the developer will stop supporting it, and it will gradually become outdated. You're not going to lose your blog that way.

Some things right here, some things still wrong.

First off, there's this: "open-source software will only flourish as long as there are at least some deep pockets--think IBM--willing to fund it in the sense of paying people for spending serious amounts of their time on it." The plain fact is that a lot of key players in important open source projects were not, and are not, paid by anybody for that work. No one got paid much for anything connected to Linux until at least dozens and perhaps hundreds of man years of effort had gone into it, all volunteer. And the non-financial incentives that made them put in that effort haven't gone away just because now there's also potentially money in it.

In short, the viability of open source software projects *as software projects* doesn't necessarily have all that much to do with corporate interest in them.

On the other hand, it is correct to observe that some *users* of open source software will need concrete assurance that the stuff will be fixed when it breaks. But that's the most important role that companies with open source initiatives, including IBM as well as others such as Red Hat, are playing: customers pay them to make sure that the stuff gets fixed, and the companies hire programmers to make that happen.

And users of closed source software have the exact same problem: they must have confidence that the company that sells the stuff will continue to be around, and continue to do an adequate job maintaining it. The difference is that that company is the *only* company legally allowed to fix the stuff when it breaks. Whereas users of open source software can shop around for a new maintainer if the old one falls down on the job.

(Well, that's one difference. Another is that users of closed source software have to worry more about arbitrary changes to license terms. But that's another rant).

And I'm still perplexed how the precipitous shutdown of a free service formerly run as a loss leader by a traditional closed source software vendor is supposed to shed light on the "social expectations" of anyone in the open source community...

I would be interested in the economics of open source, because as you somewhat said, Brad, somebody somewhere is paying to have this code written. If you're a lone programmer hoping to make a name and enhance your resume, you still gotta pay your grocery bills along the way. I think the bottom line is that open source does ultimately transform high margin software-as-license-revenue to lower margin software-as-way-to-make-money-some-other-way, whether it's on support or on your other product (like Tivo, not that they're making money). Where is a real economic analysis of the phenomenon, not a religious tome?

WRT "somebody somewhere is paying to have this code written."

There are sources of talented labor that don't fall into this category. For example, upper-class undergraduates and graduate students, both at school and as barely-paid interns. The Berkeley TCP/IP stack was written in such a fashion, and has been used widely (behavioral analysis suggests that Windows uses the Berkeley stack). At one point when I was using Linux for a research project, significant pieces of the Linux networking code were being rewritten by students.

The situation does change when a company starts to sell services based on such software. Clearly, if you want to give a customer any sort of assurance about when a bug will be fixed, you have to have someone on staff whose efforts can be directed (or redirected) to fix the bug according to a schedule. That kind of development is harder to guarantee in a volunteer environment. Although, like many other users, I have personal experience with cases where I could send a bug report to the author in France before I left work, and the fix was in my in-box when I got to work the next morning :^)

Open Source does not need deep pockets to survive. To illustrate, let me tell you a story:

In 1991, UCB lost the last of their BSD contracts from DARPA, and shut down the BSD project. By 1994, you could buy BSD 4.4 Lite on CDROM,in two different variants (FreeBSD and NetBSD).

Over the years, sponsorship of the FreeBSD project has changed a multitude of times (there's been at least three different commercial entities that at one time or another have sponsored the project) and last time I heard was totally unsponsored. Yet the software continues to be used and continues to be developed, despite a user base that numbers in the thousands rather than the millions.

Okay, lesson one: open source software can survive without commercial sponsorship, without millions of users. Because the source code is freely available, it doesn't just disappear like CP/M or OS/2 or AmigaOS or any of the many other proprietary OS's that have died over the decades.

Okay, lesson two: Why use open source over a proprietary solution? Okay, here's how I learned about open source software. In 1995, SCO raised their license fees to extortionary levels. They'd gotten a lot of small VAR's like my employer hooked on their software, and now they were turning the screws on us.

The pathetic thing was that their software wasn't even that good. For example, there was a bug in their printer daemon. It would lock up under a certain well documented set of conditions. SCO was utterly uninterested in fixing the problem, despite the thousands of dollars we were shipping their way each year.

One of our customers said to me, "You ought to use Linux." And I'm, like, "You gotta be kidding! I have enough trouble maintaining our own software! I don't want to be maintaining some buggy hackerware downloaded off the Internet too!".

A few days later my boss calls me in, and says, "What about this Linux stuff that our customers have been telling me about?". I admit I don't know a lot about it. He shows me an ad for a commercially-sold version of Linux. I scratch my head, and allow as how it wouldn't hurt to see what it looked like.

So we got in the box of Linux (Caldera Linux, if I recall correctly), installed it, I looked at it and thought "Hmm, clunky, but so is SCO." I compiled our software on it, and it ran faster than on SCO. I moved to doing most of my development on Linux and then porting it to SCO, since Linux had a far more pleasant development environment than the commercial one (as you'd expect, given that it was originally written by developers, for developers, so getting the compiler suite and assorted utilities sorted out would have been the first thing they did).

Fast forward to April. My boss calls me in to his office. "We have a new school district coming up. We can spend $XXXX for SCO Unix to drive that school district. Or we can spend $XX for Linux to drive that school district. Our bid has to be the same either way. What will it take to get our product running on Linux?"

I allow as how our product already ran on Linux, and had been running on Linux for over four months (albeit only on my desktop!). So in June we go out there and install the new systems and software. All of the computers are in a school gym on folding tables, and I'm holding my breath waiting for the inevitable disaster that comes when putting new hardware, new OS, and new version of the software all into play at the same time. I turn blue. Nothing happens other than happily running computers with happy school secretaries burbling at them while our training consultants guide them through the screens. I let out my breath and start breathing again.

A few weeks later, the Linux printer daemon locks up at one of the schools. I contact the Linux software vendor. Like SCO, they're uninterested in fixing the problem. I grab the source code to the printer daemon, and note that it is a heavily-hacked version of the BSD 4.4 printer daemon. I find the bug, and fix it. For the next two years I'm the unofficial maintainer of the "working" version of the Red Hat printer daemon, until Red Hat finally quits shipping their buggy hacked version and starts shipping another version that actually works. Meanwhile, our legacy SCO systems still need their printer daemons reset regularly.

Fast forward another year. SCO discontinues support for most of the old Unix systems we have in the field. If we get new hardware not supported by the old version of Unix, we have to fork over huge amounts of money for a new version of Unix. Meantime, while that happens with Linux, we don't have to fork over huge amounts of money when it happens. So we migrate the old Unix systems to Linux as they reach their rated end-of-life (5 years) and are replaced with new hardware.

Summary: It works better, it lets us make more profit while underbidding the competition, and we have control over the platform and aren't hostage to the whims of a company that could put us out of business. What's not to like? And we even inadvertently became supporters of Open Source, even if not voluntarily, via providing a working version of the printer daemon when the vendor was not interested in fixing the problem.

Okay, that's two stories. I have two more stories. But this is already too long, so I'll wait for customer demand. Suffice it to say, there are siginificant economic reasons beyond "it's free" to use Linux -- and the "it can't be discontinued" and control issues are just the tip of the iceberg. (And note that all of this happened years before IBM came onboard).

- Badtux the Open Source Penguin

The marginal cost of software is zero.

What's a society to do? There can't be a profit in competitive distribution of software.

Option 1: Grant monopolies via copyright and other IP laws.

Option 2: Customers pay for software only if they're the first to need it - then pay what it costs to produce. Otherwise customers take, basically for free, from the commons of previously written software.

Option 1 (closed source) and option 2 (open source) are now starting to go at it.

In theory, from the consumer point of view, option 2 is clearly the way to go.

For example, one day RedHat will go out of business. If RedHat's Linux customers want to use a new OS they can, otherwise they can pay someone to maintain RedHat Linux long after RedHat is gone.

Already you can purchase support from a third party for versions of RedHat Linux that are no longer supported by RedHat - http://transition.progeny.com/faq.html

These customers will pay pretty much exactly what it costs to produce any updates by transacting in the competitive market for trained programmers.

In the fight between option 1 and option 2, option 1 has the advantage that there is a big agent in Redmond that has and will eventually attempt to use the ability to influence law to overcome option 2.

The advantages of option 2 are that the commons is just getting bigger and better and consumers would rather pay less than pay more.

After a hard fight, I expect option 2 to win.

Mass produced software will be written by whoever has the most incentive to write it - sometimes a firm will use it as a loss leader for something else - sometimes a group of consumers will pay a firm to write something they all need that does not yet exist - sometimes a bunch of kids will write something to make a name and to learn the craft of programming.

Custom solutions will be written to order using the commons as a base.

Milton said why:
"Fame is the spur that the clear spirit doth raise (That last infirmity of Noble mind)/
To scorn delights, and live laborious dayes..."

Software has become a tougher business. It's harder to find the well-paying consulting gigs that give you time off to do what you want, such as work for free on open-source. The companies have become more exploitative towards their engineers, now more under-staffed and over-worked than ever, and also threatened with offshore outsourcing. (One boss of mine used to brag to his microsoft buddy that he didn't need to send work to India because he could hire Amerislaves oops I mean Americans for $20/hour.)

In this environment I have no sympathy for whiney customers who cry in their beer when some overworked underpaid engineer pulls the plug on their hobby horse, oops I mean hobby blog.

I'm not quite sure I'm grokking your argument.

Obviously, when picking an open source software package, you want to pick one with lots of users and preferably some commercial support. But this has absolutely nothing to do with "graceful shutdown" (which is a strange analogy between services and software). Rather, it has to do with whether a given software package will tend to get useful new features in a timely fashion.

A service like weblogs.com has a substantial ongoing operational cost. But working software can be kept operational almost indefinitely by one part time hacker on the weekends (I've done it). All that you need to do is recompile it every now and then, update it to use new APIs every few years (perhaps a Gnome 1.0->2.0 port), and fix the occasional bug. For most software, this works out to a few hours a week, which can be supplied by a single developer anywhere in the world. You can migrate to a new package at your leisure.

The only really sticky part is the race to provide device drivers. A failure here could rapidly close an open source operating system off of new hardware.

But for most open source software packages, you'll see some slow years and some fast years. XFree86 has been moribund for years now, but newly re-instated Xorg is rapidly laying the groundwork for a new era of improvements.

If you're concerned, don't choose marginal open source projects. Stick with the market leaders.

I'm still looking for some actual analysis of the economics of open source, not anecdotes. I have plenty of those myself. Let me put the issue in concrete terms. I offer it as a theory, seeking real analysis to support it or tear it down. RedHat, nor any company whose business is software (and the associated paraphernalia), will ever be even remotely as big as Microsoft, Oracle, or just the software/services part of IBM. Ever. If you look at RedHat's model for software development, you will see that they absolutely rely on those hordes of paid-by-someone-else programmers to produce the product that they then make money from using a support model. If they actually paid those programmers themselves, they would have to charge support that would be *at least* equal to the cost of being a company that employed all those paid-by-someone-else programmers. In the end, I am left with the belief that the economics just don't add up. There is no free lunch. (Sadly, little capitalist that I am, this was one of my first reactions to The Cathedral and the Bazaar years ago). Open source adoption is largely driven by it being free modulo support. But it *really* wasn't free to produce, and more than it is free to support. I enjoy the politics of this blog, but I am hoping the economics bent would make it a place to get some insight into this issue. Please, no flames, just insight.

>

It wasn't *really* free to produce, but it was produced by whoever could produce it cheapest, for whatever reason.

When someone else, for some different reason, can produce it cheaper, that someone else has all he or she needs to take over.

Open Source isn't really free. Open source is subject to bidding in a competitive programmers market. Close source is a tiny monopoly.

If, for some reason, Microsoft stops being the company that can maintain windows at the lowest cost then oops, Microsoft has a copyright monopoly over Windows.

Say HP could afford to write off more of Linux development than Linus Torvalds could because HP has other products for which Linux provides a form of marketing.

HP is free to add to, or even take over Linux development. HP's development efforts can also be added to Linus' in a way that is not possible in the mini-monopoly model.

Open source is not "free". Open source is subject to competition, to changes in the identities of the developers at low costs as the needs of the developers change and to additive production, where anyone who is able can easily "join forces" with the current software producers.

Open source may not really be free, but there are a lot of reasons to expect that open source will be cheaper for consumers.

For the guy wanting a real economic analysis of open source software: I once did an analysis based on a market point of view. I think I had about a half dozen reasons why various Open Source packages had originated and thrived. All of these reasons had one thing in common: they had absolutely nothing to do with a "gift culture" or "communism", and everything to do with good solid free market values.

The most obvious thing that I pointed out was that not all customer demands are met by the market. Some customers' requirements are so esoteric that there's no way for any company to make money meeting them. For example, Linus Torvalds wanted to play with the internals of an operating system so that he could see how the i386 virtual memory worked. He tried playing with Minix, but Minux had a restrictive license that didn't allow him to distribute his fixes. He decided to write his own. This is a case where there was a demand for an operating system with source that could be modified by programmers, and nobody meeting the demand because it was not viewed as a profitable market.

It's sort of like the proverbial barn raising. If there are not enough people in the county to warrant a full-time barn builder, but there is a demand for barns, what are the people to do? They can either a) go without barns (but that's hard on their livestock), or b) band together and help each other build barns for their farms.

Software has an interesting trait -- unlike a barn, once you build it for yourself, it costs $0 to duplicate it and hand it out to other people. In fact, duplicating it and handing it out to other people can actually reduce your costs, if those other people decide to help you add new stalls to your barn and expand it for more livestock. Even if those people don't, it's not costing you any more money than you've already spent. It's not like you have to pay attention to people who aren't paying you, after all. In the case of Linux, what happened was that people looked at it and said, "Hmm, if we did just a few more changes to it, we could use it as a house!" Or "If we did just a few more changes to it, we could use it as an office building!". Etc. In other words, it started out meeting one particular need. Then other people decided that altering the already-existing Linux to meet their own needs was less expensive than creating their own OS from scratch (note that buying the source code to a commercial OS in order to produce your own customized OS is not possible at the moment -- no commercial vendor of operating systems will allow you to do that, thus driving people to Linux).

While part b) (barn raising) is what actually happened on the American frontier, part a) is what Microsoft would prefer, since they are selling castles and would really prefer that you save up your cash for years in order to buy their castle to house your livestock rather than getting together with some of your neighbors and building barns. And if you can't afford a castle for your livestock, tough -- they'd outlaw barns if they could, but have settled for driving all the barn-builders out of business or buying them out once they start selling too many barns.

People once said, "It is impossible to write a commercial-quality operating system without a big company behind you." Linux and FreeBSD prove that to be false (both were commercial quality before being noticed by large corporations... I was there at the beginning, and saw it happen). Products don't arise out of charity, they arise out of good free market reasons. Identifying those reasons is the economist's job, rather than denying that open source software exists and has existed since long before any corporate interest in open source software. Only delusional people deny reality. Oopen source software exists. Ergo, there must be economic reasons that it exists. And since open source software existed before any big company interest in it, attributing its existence to big corporations is, like, uh, dumb?

- Badtux the Economical Penguin

Great site fatty lose weight with reductil and reductil uk

Great site fatty lose weight with reductil and reductil uk

Great site fatty lose weight with reductil and reductil uk

Get it up mate, it's fun!

This will get yours up again, dude!

It gets yours up to the top dude! The girl will enjoy it!

Muppets love Viagra!

I don't really think your thoughts are right. Maybe you need a loan?

Thanks for your blog!

Hire a car to feel higher!