What's The Linux Kernel Worth?

schneelocke writes "What's the value of the Linux kernel? After an offer by one Jeff V. Merkey to pay 50K USD for a BSD-licensed copy of Linux, David Wheeler does some calculations and comes up with an estimate of 612M USD." Wheeler has come up with a number of interesting software-worth estimates and other quantified facts about Free software; since some aspects involve ineffables and hypotheticals, the details can be argued, but he provides a good framework with SLOCCount.

Comments

Nothing

[AKAImBatman]

By my calculations, the Linux kernel is worth: nothing.

Before you get your tights in a twist, just listen to me for a moment. The value of a product in a capitalistic system is determined by what the market is willing to bear. Yet it is not worth anything if the developers are not willing to sell it at what the market demands. Thus we have a gap. The market would probably be willing to bear a few million (perhaps as high as 50 million) dollars for the Linux IP. Yet it seems that the developers would demand a price in the range of 612 million.

The end result is that the Linux kernel has no market value what so ever. The developers won't sell it at the market's price, and the market won't buy it at the developers price.

Re:Nothing

[millwall]

The end result is that the Linux kernel has no market value what so ever. The developers won't sell it at the market's price, and the market won't buy it at the developers price.

This is not quite true. The market value is what the market would buy the product for, if it WAS for sale.

Imagine that you have a car, which you for whatever reason don't want to sell at the moment. This doesn't leave your car with "no market value". The market value is still what the market would have bought it for if it was for sale.

Re:Nothing

[wcbarksdale]

So the Mona Lisa is worth nothing, because the Louvre isn't willing to sell it?

Re:Nothing

[bickerdyke]

You are asuming that there IS a market. The OP stated that there is currently no market for linux kernels, because there is only one potential buyer and one potential seller who cant agree on a price --> no trade, no market, no market price.

In your example there is a market for collector cars which can be used to find a market price.

But it would be better to say "priceless" than "nothing"

Re:Nothing

[eln]

The discussion based on the offer is what would it be worth in a binary-only form under a BSD-style license. The developers are willing to accept a price of $0 for the code under a GPL license, but are not willing to sell at any price under a less restrictive license. Hence, there is no asking price for the actual product demanded, as it is explicitly not for sale. Since there is only 1 person in the potential market at this point, and there has been only one offer, with no counter offer (unless you count "infinity dollars" as a valid asking price), there is not enough data to set a reasonable market price for the product.

Estimating based on what it would cost in a commercial environment is also flawed, because there are too many variables to consider. If it were developed in, say, Silicon Valley, each programmer's hourly rate would probably approach $100. If it were developed in, say, Boise, Idaho, the hourly rate for these programmers would probably be in the neighborhood of $20 to $30. There is also development models, management, total personnel required, etc. These things will all vary depending on the nature of the business, what kind of timelines they were looking at, who the management team was, etc.

The bottom line is, since the developers have always been paid nothing for their work (except those that are being sponsored by commercial entities), the total value of their time put into the project is $0. Sure, you could try and put it in terms of an opportunity cost to the developer, but that sort of thing leads to inflated valuations, since in all likelihood if these guys weren't writing the code in their spare time, they would be doing some other hobby that doesn't pay them anything instead.

The bottom line here is, the only time that you can assign a value to is the time that someone actually received a wage for. This is a small minority of the overall code base, so by that method the code would not be worth much at all.

Therefore, all we are left to consider is whether or not Linux is a good value to the consumer. Generally speaking, a good value to the consumer is one in which they extract more value out of the product then what they paid for it. Since most people get Linux for free, it doesn't take much to get more than what you paid out of it, especially if you use it in a business context. There are those, of course, who will get negative value out of it (they get frustrated and throw the computer out the window, for example), but the vast majority of users will get a net positive value out of the product. Putting a dollar amount on that value is difficult, though.

Re:Nothing

[Bombcar]

You've hit the nail on the head. The Linux kernel is Priceless.

A DSL connection to the internet: $50
A decent PC: $500
Downloading enterprize-class source code: Priceless.

There is some code you can't buy, for everything else, there's Microsoft.

Re:Nothing

[ajs]

The bottom line is, since the developers have always been paid nothing for their work

Why do we keep saying this when it is so patently untrue?

Were the SGI engineers payed nothing for their work on the Linux kernel? The DEC engineers who made 2.0 possible? The IBM engineers who ported massive amounts of IBM code into Linux (the topic of a lawsuit, I'll remind you) and wrote a good amount of their own code? What about the Red Hat engineers who have contributed hugely? Alan Cox isn't paid? I think he'd be upset to hear it. Last I checked Linus was being paid specifically for his work on Linux by OSDL.

These are just the high-profile cases. Dozens of people around the world are paid to work on specific niches of the Linux kernel all the time.

Same thing goes for the rest of the OS tools, utilities, and subsystems.

The fact that some, even a majority (and I'm not conviced on that point) of the work might be gratis does not mean "the developers have always been paid nothing."

Re:Nothing

[Tassach]

Yet it seems that the developers would demand a price in the range of 612 million

RTFA. The 612M figure is an ESTIMATE of what it would cost to DUPLICATE the 2.6 kernel using traditional software development methods. And it's probably a pretty reasonable estimate. Developing good complex software is HARD and EXPENSIVE.

Re:Nothing

[poptones]

The bottom line is, since the developers have always been paid nothing for their work (except those that are being sponsored by commercial entities), the total value of their time put into the project is $0.

So... if I fix a flat tire on my car myself, that labor has a value of zero dollars? Even 'tho I just saved myself twenty bucks? If I fix my neighbor's computer for free, saving her fifty bucks in the process, the value of my labor is zero? If I spend some time I could be laboring for a paycheck instead engrossed in a hobby, that time has zero value?

Dude... you got some fucked up values

Value:

1 # An amount, as of goods, services, or money, considered to be a fair and suitable equivalent for something else; a fair price or return.

2 # Monetary or material worth: the fluctuating value of gold and silver.

3 # Worth in usefulness or importance to the possessor; utility or merit: the value of an education.

4 # A principle, standard, or quality considered worthwhile or desirable: "The speech was a summons back to the patrician values of restraint and responsibility" (Jonathan Alter).

Therefore, all we are left to consider is whether or not Linux is a good value to the consumer.

By your measure it's not - it's completely worthless unless that consumer paid for it.

What is the value to me of a diamond? Only what I could sell it for - I have no use for a diamond (unless, perhaps, I need to cut a piece of glass). Bottom line is you expended a lot of words saying nothing. Value and price are not directly related, nor even comparable to one another.

ooooh

[LiquidCoooled]

612million / [Developers.Count] = $650 (per machine)

Who woulda thunk it..... ;)

perhaps...

[Anonymous Coward]

we should get a better estimate by asking the nice folks at SCO. They seem to know much about this.

You just can't...

[numbware]

... put a price on it. Linux is priceless. Mac OSX is $50.00. Windows is a paperclip and a bubblegum wrapper.

Obligatuary quote

[rastakid]

"They want me to be a whore!" -- Linus Torvalds.

Merkey's offer doesn't make sense to me ...

[Anonymous Coward]

After an offer by one Jeff V. Merkey to pay 50K USD for a BSD-licensed copy of Linux

Why would they do that? What advantage is there to the BSD vs GPL licenses?

The only advantage is that if you redistribute or sell software that is GPLed, you have to provide source code - with BSD you don't.

So, Merkey's company wants to sell modified Linux without providing source code to the modifications. While I doubt the modifications are worth that much, he apparently does.

Why wouldn't Merkey use FreeBSD for the application he wants to sell? Almost all linux software is available for FreeBSD, and then he wouldn't have to pay $50,000 for a license.

Or can someone explain this to me?

Re:Merkey's offer doesn't make sense to me ...

[John Harrison]

I'm going to be intentionally vague here so that I don't have to post as an AC.

About two years ago I was speaking to the developer who had ported Linux to a particular hardware security device. I asked him why he had gone with Linux instead of OpenBSD as his base. He stated that it was his preference to go with OpenBSD, or any of the flavors of BSD, but he went with Linux because the company is publically committed to Linux and Linux has a marketing value that the BSDs do not. It is better to say, "Our gadget now runs Linux! Won't your developers be happy?" than it is to say, "Our gagdet now runs OpenBSD! Won't your developers be happy?"

Basically there are often non-technical reasons for wanting to use Linux even when some other OS would be a better technical fit.

Eh?

[niko9]

It's like trying to put monetary value on a Van Gogh or a Matisse. The Linux kernel is truly priceless. You could never get that kind of collaboration even with the most highly paid software engineers, beacuse they don't do it for money, neither did Van Gogh.

Mu

[Vainglorious Coward]

What is air worth? Some things have great value, but simply trying to measure that value in dollars is to misunderstand the nature of that value.

I Thought This Was Already Established

[Lethyos]

Clearly it's $699.

pretty safe offer

[Kwantus]

Jeff's offer:

This offer must be accepted by **ALL** copyright holders...

That'll be harder than getting agreement on the Charlottetown accord (a thing in which any given Canadian could find something to hate)...A) you'll never find *ALL* the copyright holders - plus the complication some have died, who of their heirs has the say? B) of the hundreds you can find it's sure a few will say No on principle.

Re:pretty safe offer

[Entrope]

No. Some groups require that (most notably the Free Software Foundation), but many do not. Linus and the other kernel folks are part of the latter group.

While there are some benefits to having a single copyright holder, you need a lot of bookkeeping to track everyone's copyright assignments and a lot of work to make sure they are proper. US copyright law requires that any copyright assignment be in writing and identify (in writing) exactly for what works copyrights are being assigned. Many European countries recognize "natural rights" or "droits d'auteur" that cannot be assigned to a third party. Some programmers have employment contracts that stipulate all software copyrights for things they write while employed belong to their employer, even if the software was written on the programmer's own time. Et cetera.

Linus decided either the effort was not worth it or that there were other benefits to not requiring copyright assignments. The Free Software Foundation does go to the effort to work all the details out. If you (or anyone else) wants to have a single copyright holder for a GPLed program, I encourage you to assign that copyright to the FSF. It saves a lot of effort on your part and ensures that the copyright holder has both the resources and resources to protect the software.

Mu

[temojen]

You are assuming that market value is the only measure of worth. Even in capitalist accounting (GAAP) you are mistaken.

To an accountant, all assets are valued at their expense, minus any prior amortization or markdowns. Most Linux users would thus have to include in their valuation any time they spent downloading, configuring, and installing the kernel.

I would have to include a few hundred dollars for the time to develop, test, and submit the (very small) patch I submitted. With ten years of their life put into it, Linus Torvalds, Alan Cox, etc would each have to value it at several million dollars.

Misconceptions

[Savant]

There seem to be a remarkably large number of people posting on this one who haven't read past the title, never mind the article.

This isn't about a consumer price for a kernel binary. Comparisons with copies of Windows are irrelevant. The $612 million dollars quoted is a suggested figure representing the kind of cost a commercial company would have to take on to develop an identical operating system kernel.

Software companies have in the past changed hands for large sums of money. The brand is of course worth some of that money, as are relationships with existing customers, but a large part of that value is the IP possessed by the company. There are few companies that have possessed software assets of a complexity and widespread use comparable to the Linux kernel that have changed hands, and such companies when sold have been bought for large sums - to pick one example, Netscape was bought by AOL at a price tag of $4.2 billion dollars.

The value of the Linux kernel code and Linux branding, if a company with sufficient resources were interested in obtaining it, and if it were for sale, would quite probably exceed this figure of $612 million by a sizeable percentage.

$50K is a derisory offer for even an non-exclusive right to develop and redistribute the IP, which is effectively what a solitary copy under the BSD licence would give. Certainly the company I work for would laugh helplessly if such an offer was made for our code, which is several orders of magnitude smaller and less complex than the kernel.

Savant

Re:GPL vs BSD

[Daytona955i]

Well considering many of the developers would not work under a BSD license I say your argument is flawed. There is a BSD licensed UNIX that runs on x86. I think not as many developers work on it because what incentive does a large comany have to return it's modifications back into the free version as opposed to just saying FU, I have my fork.

Granted it does happen (as in the case of apple) but it is kindof enforced under the GPL and I think that gives the developers some solice that they wouldn't have under a BSD style license.

How would you feel if someone took your work, made a change or two and sold it as their own? I'd be pretty pissed.

10 Billion ++

[HexaByte]

According to an IBM rep who spoke to our LUG 2 years ago, IBM promised to invest 1 Billion in Linux because they had estimated it would take them $10 Billion to get it to where it was then. That was in the 2.4 stage. Now, with the 2.6 kernel, it should be worth another Billion or so. Of course, I'm willing to sell copies of the latest kernel to all comers for only $500. That includes a year of my "Platinum Support".

Re:Why

[Brandybuck]

The 3-clause BSD licence is poisonous, because it allows someone effectively to turn an open-source product into a closed-source one, just by not distributing the source code.

Not "poisonous" at all. Keep your FUD out of this. While one can take BSD licensed source code and create a binary closed source product, this is not "poisonous". The orginal source code is still there. The orginal project is still there.

It would be like someone taking one apple from a free apple tree and locking it up. Are people going to be screaming "he poisoned the tree" when he locks up one apple? Of course not!

FreeBSD *is* Linux! :)

[Xtifr]

For some very very very very very loose definition of "Linux", FreeBSD *is* Linux! No, really. I saw an ad once for a vendor that was selling Linux distributions, and there was FreeBSD on the list of Linux distributions....

Heck, who am I to tell 'em different? I used to refer to Abiword as "my version of Word", as in, "My version of Word seems to have problems with your file, could you try resending as RTF?" Nobody ever questioned me (which just shows how overrated the notion of Word as a "standard" is). :)