Who's Writing Linux These Days?

cold fjord writes "IEEE Spectrum reports, "About once a year, the Linux Foundation analyzes the online repository that holds the source code of the kernel, or core, of the Linux operating system. As well as tracking the increasing complexity of the ever-evolving kernel over a series of releases from versions 3.0 to 3.10, the report also reveals who is contributing code, and the dominant role corporations now play in what began as an all-volunteer project in 1991. While volunteer contributors still represent a plurality among developers, over 80 percent of code is contributed by people who are paid for their work. ""

Comments

Patrons

[jones_supa]

While volunteer contributors still represent a plurality among developers, over 80 percent of code is contributed by people who are paid for their work.

This. I've said it before and will say it again. The open source projects with most bugs and slowest development time are the ones without proper sponsors. That's why I also use a lot of commercial closed-source software myself, but do not have any particular grudge against OSS either. Just pay the developers properly, because complex, properly quality-assured modern software is impossible without that.

Re:Patrons

[RabidReindeer]

While volunteer contributors still represent a plurality among developers, over 80 percent of code is contributed by people who are paid for their work.

This. I've said it before and will say it again. The open source projects with most bugs and slowest development time are the ones without proper sponsors. That's why I also use a lot of commercial closed-source software myself, but do not have any particular grudge against OSS either. Just pay the developers properly, because complex, properly quality-assured modern software is impossible without that.

Before you get too comfortable with that assertion, recall that Linux Torvalds wasn't being paid to develop Linux in the beginning nor for long after. Nor were his earliest assistants.

It's certainly easier to develop good-quality software if you aren't distracted by the need to earn a living doing something else, but it's not essential.

Re:Patrons

[Bill, Shooter of Bul]

I don't see how that follows. If you knew that sponsored open source projects were superior in quality to non sponsored open source projects, I don't see how that applies to non open sourced sponsored projects. It might, but its not a logical inference from the previous statement.

Re:Patrons

[JoeMerchant]

There are many paths to not needing to earn a living - while 99% of the world may not have the luxury of "room and board, including broadband, covered without working for a paycheck" - that still leaves tens of millions who do.... now, whether or not these are the people you want developing your OS kernel is another question, but if you throw another 99% filter on them, there may still be thousands of people out there who have the means, ability and disposition to write high quality FOSS, and, thankfully, the internet has given them the means to collaborate.

Re:Patrons

[NatasRevol]

There's a saying.

"He's the exception that proves the rule."

Re:Patrons

[Anonymous Coward]

I don't see how that follows. If you knew that sponsored open source projects were superior in quality to non sponsored open source projects, I don't see how that applies to non open sourced sponsored projects.

Most closed source projects are backed by a team of full-time developers.

Now, I did say "most". Some companies are cheapskates.

Re:Patrons

[jones_supa]

Before you get too comfortable with that assertion, recall that Linux Torvalds wasn't being paid to develop Linux in the beginning nor for long after. Nor were his earliest assistants.

I'm still mostly comfortable with my assertion. :) I am talking about modern software, which is significantly more complex than early Linux. Indeed, it is the complexity and lines of code which makes it day by day harder to make meaningful software without it being a full-time paid commitment.

Re:Patrons

[jones_supa]

I don't see how that follows. If you knew that sponsored open source projects were superior in quality to non sponsored open source projects, I don't see how that applies to non open sourced sponsored projects.

Closed source projects can simply directly sell their product.

Re:Patrons

[Drew617]

But... corporations!

Let me explain to you how this works: you see, the corporations finance Linux, and then Linux goes out... and the corporations sit there in their... in their corporation buildings, and... and, and see, they're all corporationy... and they make money.

Re:Patrons

[DaveV1.0]

That saying does not mean what you think it means.

Re:Patrons

[RabidReindeer]

Before you get too comfortable with that assertion, recall that Linux Torvalds wasn't being paid to develop Linux in the beginning nor for long after. Nor were his earliest assistants.

I'm still mostly comfortable with my assertion. :) I am talking about modern software, which is significantly more complex than early Linux. Indeed, it is the complexity and lines of code which makes it day by day harder to make meaningful software without it being a full-time paid commitment.

I'm pretty sure that if software had gained some sort of magical properties in the last 30 years I'd have noticed it.

Yes, the codebase contains more components than it used to - although having smarter and more standardized hardware has reduced the number of unique drivers. But Torvalds is still "Penguin-in-Chief". He just delegates a lot now since there are more components to ride herd on. And now it's his primary job.

The fact that a lot of the components have full-time professional teams working on them is generally an indication that they can see a benefit from having control over an item on their personal agenda and on their own schedule instead of waiting for someone to come along on their own time and in their own way. Which is natural, since the essential systems were worked out 2 decades ago. Since then, we've seen the addition of virtualization support (in large part created by academic, rather than commercial developers), abstractions in block I/O, new network features and filesystems, clustering and other things that are typically of commercial interest.

Along the way, a lot of these items were originally developed by unpaid developers who then leveraged what they had done into careers for themselves. Even Red Hat itself wasn't a major commercial endeavor at first.

Not to say that IBM and Oracle haven't been major contributors, but Linux has many roots and many parts and they each have their own characteristics.

Re:Patrons

[Anonymous Coward]

Personal opinion, unrelated to my work:

You said:

1) " The open source projects with most bugs and slowest development time are the ones without proper sponsors.

Hum, let's take that at face value, just for the sake of argumentation (I personally don't think so, but let's put that aside for the moment)

2) "That's why I also use a lot of commercial closed-source software myself, but do not have any particular grudge against OSS either."

2 does not follow 1.

Open source with sponsors is better, you say. Closed-source is not an extrapolation of that.

The fact that open source is favoured in many situations where cost is not a factor shows its quality to be superior to that of closed-source -- not the opposite.

While surely people must have a means of living, in my opinion business pressure leads to poorer products -- either by rushing incomplete versions out or simply by design.

I do not have data to back this up other than the massive use of open source in everything.

Re:Patrons

[Tough Love]

The open source projects with most bugs and slowest development time are the ones without proper sponsors.

I know, right? Take Samba for example... oh wait.

Re:Patrons

[http]

No, there isn't a saying.
The tradition in law is that the existence of an exception may be used to infer a rule - e.g. "No Parking: 4pm - 6pm" means that parking IS permitted at other times. Mr. Torvalds may be considered unusual, but his efforts are not part of (or exception to) any rule I can think of, except maybe "being polite and rude consistently gest things done in the long term"

Re:Patrons

[Anonymous Coward]

Well that was a long time ago, minux and the first Linux were way smaller. There is a quote from the mailing list that goes like this," i don't see any reason that linux would ever need to support anything but MFM hard drives. Linux is big now, with a lot of users, to keep achieve that linux needed paid support. Not every one has the luxury of living in a european welfare state and doesn't have to worry about things like paying for insurance, self and kids college, grad school. Most of us cant go directly from bachelors to grad school and phd with out first spending time in the commercial world.. also the experience of the commercial world helps a lot.

Thins still cost money simple, this isn't star trek..yet.

Re:Patrons

[Anonymous Coward]

Are we just shouting opinion now without any backing? You say something without actually arguing your case. You don't manage to defend any points. I shout in your general direction that I don't like your post. At least I managed to communicate why.

*Ploink*

Re:Patrons

[Bill, Shooter of Bul]

Still does not logically follow. You are making the impicit assumption that the sole determination of quality in software is the percentage of people that are paid to write it. You are totally ignoring the closed/open source variable.

Re:Patrons

[Bill, Shooter of Bul]

And London is the Capitol of the UK. What does that have to do with software quality in closed source applications?

Here are a few more random things that have nothing to do with what we are talking about, devoid of any context, in case someone else wants to see them.

Trees are made out of wood.
Bill Clinton didn't found Microsoft.
ARM processors are more common in smartphones than MIPS processors.
Debian supports multiple kernels, not just Liinux.
Last year eneded on Dec 31st, 2013.

Re:Patrons

[icebike]

Before you get too comfortable with that assertion, recall that Linux Torvalds wasn't being paid to develop Linux in the beginning nor for long after. Nor were his earliest assistants.

It's certainly easier to develop good-quality software if you aren't distracted by the need to earn a living doing something else, but it's not essential.

Linux moved very slowly (Glacially) while Linus was working for a living and building linux evenings and weekends. He was still in the University at the time if the initial release in 1991. He graduated in 1996 and took work with Transmeta, (Crusoe) which lasted till 2003. Transmeta gave him wide scope to spend significant time on Linux on the company clock.

From 2003 on, he has been essentially paid, allowed, and encouraged to work on Linux with a free hand.

So he spent 8 years at University (interrupted by a year of military service). How those years were financed is not public knowledge, but I suspect his Parents and the Finnish government played a part.

From graduation in 1997 on, he was on the payroll of companies that had the good sense to let him do pretty much as he wanted.

And that's not unusual. A lot of these early contributors were in the employment of companies that allowed and encouraged them to work on linux. You need only dig through early archives to see the email addresses used.

Re:Patrons

[Kjella]

It's certainly easier to develop good-quality software if you aren't distracted by the need to earn a living doing something else, but it's not essential.

And it's certainly easier to develop good-quality software if you aren't distracted by the need to earn a living off it, you're free to reject bad code or go back and refactor until you get it right no matter if it has a "business case" or not. And you're not going to get a CEO who's read too many trade magazines and wants to replace you with half a dozen Indians. I guess that's not true for "scratch an itch" projects but if it's your baby you don't want it to be just "good enough", you want it to living proof of your skill. At least that's my impression of several dedicated project leads/core contributors.

Re:Patrons

[Anonymous Coward]

Just pay the developers properly...

People in low wage jobs usually don't get paid for all the work they do, but do get pressured to do more work. Why should developers be any different?

Re:Patrons

[Anonymous Coward]

"Before you get too comfortable with that assertion, recall that Linux Torvalds wasn't being paid to develop Linux in the beginning nor for long after. Nor were his earliest assistants."

Linux was paying himself. This was very smart. If you want to be a great programmer, and get paid a lot of money in the future you have to challenge yourself and write great programs early without getting paid. It is the same in any high skill arena. Take a sports example: how many times did David Beckham kick the soccer ball before getting paid to do it. Millions probably. But without doing all of that FREE work he would never had made any money playing soccer.

Re:Patrons

[Anonymous Coward]

Trees are made out of wood.

I thought wood came from Trees. I thought trees grew naturally and weren't made at all.

Re:Patrons

[Anonymous Coward]

Tridge and Jeremy have been working on Samba for twenty years now. Which is a very long time.

I remember hearing a story about how they went to a conference and they were so excited to meet the Microsoft SMB devs and ask them questions. But when they got there they found that the Microsoft devs had only been working there for a couple years so they were still newbies.

And when they asked how many people work on SMB at Microsoft, they realized that the Samba team had more full time professional developers than the Microsoft team.

Re:Patrons

[Pav]

Apparently "glacial" development was still good enough to steal market share from the commercial Unix vendors.

Re:Patrons

[Anonymous Coward]

The tradition in law is that the existence of an exception may be used to infer a rule - e.g. "No Parking: 4pm - 6pm" means that parking IS permitted at other times.

Well, we have a sign on the loo door saying 'This toilet is cleaned by a female cleaner after 5pm' but, being more rigorous type when it comes to semantics, I still check before flopping out my todger at any time of day.

Re:Patrons

[Pav]

Mod up... Part of what is so refreshing about F/OSS is that it's self directed by passionate people. There's something obviously wrong with the premise of the article. If patrons are the real source of vitality in Linux and F/OSS why are the most successful distros clustered around Debian and not Redhat?

Re:Patrons

[Pav]

Doh... premise of the COMMENTS

Re:Patrons

[Anonymous Coward]

"Exceptio probat regulam in casibus non exceptis", or "The exception proves the rule in cases not excepted". However, Wikipedia does suggest that "the exception that proves the rule" also has a loose rhetorical sense, where the exception, being an exception, proves the commonality of the rule, which is taken as a 'rule of thumb' rather than a strict requirement.

Re:Patrons

[7-Vodka]

Sure, mod up the moron who doesn't know the difference between Free Software and open source.

Re:Patrons

[jones_supa]

It's still better than insulting other people by calling them morons.

Re:Patrons

[RavenLrD20k]

I hope I can make the links you need to see the logic that other posters have been attempting to convey.

With non-sponsored open source software in most cases the developer is creating the software mostly for their own personal understanding of the concepts of software design or to develop their own personal tools for whatever other personal projects they are working on. At this level the software is unpolished, providing only enough of an interface for a single person to utilize. You can start to have other developers joining the project for their own personal growth and understanding. An interface design major might put a better GUI front end to the project. Someone who needs to learn about database design might put more robust SQL statements into the code, or make the code behave better with the APIs of a certain database engine. Regardless of who comes in and adds to the project, with non-sponsored open source, there is no vested interest beyond what a person puts in for their own personal growth and understanding. Also, everything with the development of an application is a serious might. At this level, it's not very often that people training in QA have any interest in getting involved as college classes on System Analysis tend to focus on Use Cases and hardly ever suggest students go out and find an open source project to test. Not that they would get anywhere with the egos that tend to be involved in many of the non-sponsored open-source projects. Coders at this level can, at times, have very strong emotional attachments to their code and take a criticism as a challenge to duel.

The next level is sponsored open source. At this level, you're going to have people putting money into a project to help make it as good as it can be. This incentives developers into producing cleaner code with better functionality that would appease a more wide spread audience. As the project grows and gets more monetary involvement a lead developer may look over the system he's building as a whole and try to identify areas that need to be improved and features that would be good to add. For this he may start looking to hire on people to start fulfilling the roles where he isn't necessarily the best fit in the interest of putting out a better project. Depending on the project lead's ability to take criticisms in the face of this goal, he may actually hire on some Analysts to perform QA and testing on the project to help it appeal to a wider audience as well as some form of tech support to help the end users in utilizing the application.

The last level is "sponsored" closed source (quotes denote redundancy). If a person or business is developing closed source, it's because they live or die by the code they're writing, and they have every intention on living by it. This code is either going to be in a product that is to be distributed to end users, or it is an internal application that is going to handle some aspect of internal business processes. In the former case it absolutely has to be easy to use, well documented, well tested, patched and maintained, and have support teams available to help the end user troubleshoot any problems that may be cropping up. If any part of this does not live to some kind of market standard, the product will fail. In the latter case much of the same philosophies apply with the addition of mission critical reliability. If internal software is regularly choking and losing customer records to a buffer overflow three times a week one can assume that if it isn't fixed in short order the entire business will very shortly wind up on the rocks. For both of these cases, companies that develop closed source software will set up an entire department of Quality Assurance and Testers to make sure that what gets put into production is the absolute best it can be.

I'm sure there may be some other aspects of these three tiers that I have completely missed the mark on, but this is what comes to mind the quickest for explaining the reasons why closed source software has the greater likelihood of being higher quality applications

Re:Patrons

[Anonymous Coward]

London is the *capital* of the UK. The *capitol* of the UK consists of the Houses of Parliament (the House of Commons and the House of Lords).

Re:Patrons

[Anonymous Coward]

What about OpenBSD?

The results of my inverstigation

[For a Free Internet]

Dear Slashdort:

After painstakingly invesgingating Linux and other FLOSE operatings systems, I have comr to the conclosion that they are all copies of an alien intelligence left by secret underground network s of masonic monks five hundred feet underneat the piramods in Earjip. So it is naturil that there are some bufs, because the softwart is very oled. Sincde it is FLOSE you are not allowid to do anything about that. Beter to use 100% Italian sowftare from ADOBE, it has style and it has CLASS.

Re:The results of my inverstigation

[Anonymous Coward]

Beter to use 100% Italian sowftare from ADOBE, it has style and it has CLASS.

It has DRM too.

I.e. By People Who Know What They Are Doing!

[Anonymous Coward]

Because if you ain't paid for it, you ain't nothing but an armature.

Re:I.e. By People Who Know What They Are Doing!

[Sponge Bath]

Because if you ain't paid for it, you ain't nothing but an armature.

The subject is software, not electric motors.

Re:I.e. By People Who Know What They Are Doing!

[inasity_rules]

There happens to be a fairly large amount of software used to control large synchronous machines. Heck, even on smaller scales a modern VSD probably has some form of basic embedded OS.

Basic functionality of an OS is there

[jfdavis668]

The constant need for change and improvement are generated by companies making software and hardware for Linux. So they are motivated to extend it. There are no longer major pieces of the OS to develop. All that code developed by companies is then reviewed and tested by part timers, who don't get the credit they deserve.

Re:Basic functionality of an OS is there

[DerekLyons]

All that code developed by companies is then reviewed and tested by part timers

[[Citation Needed]]
 
The linked article suggests the opposite and shows that signing off on code submissions is overwhelmingly corporate.

Re:Basic functionality of an OS is there

[Monkey-Man2000]

All that code developed by companies is then reviewed and tested by part timers

[[Citation Needed]] The linked article suggests the opposite and shows that signing off on code submissions is overwhelmingly corporate.

What, you don't browse the Linux repository's weekly commits every Saturday evening like me? People get paid for that?!

Re:Basic functionality of an OS is there

[Anonymous Coward]

Schedulers? File systems? Security? Buffer bloat? The Linux kernel is so far ahead that it's not even fun to compare it to proprietary OSs, yet it keeps improving.

Re:Basic functionality of an OS is there

[DerekLyons]

In other words, you don't have a citation to support your claim.

Noted.

Corporate Linux?

[IanGrant604]

Wasn't Linus Torvalds the one who railed against, in so many words, corporatism? Oh yeah. He was. (http://www.theregister.co.uk/2013/07/16/torvalds_potty_mouty_fight_back/)

Aye well, so much for the "volunteers" behind Linux, right?

Re:Corporate Linux?

[Anonymous Coward]

Yeah, so much for them. They only contribute about 20% of the code. That's much less of a percentage than is developed by volunteers in any other enterprise-class operating system in wide use today except, well, all of the others.

Re:Corporate Linux?

[Anonymous Coward]

Do you have the link to the report for OpenBSD handy, I can't seem to find it.

Re:Corporate Linux?

[Anonymous Coward]

Do you have the link to the report for OpenBSD handy, I can't seem to find it.

Wait I found it. It was in the freezer next to my sheet of LSD.

Re:Corporate Linux?

[Anonymous Coward]

Be careful, you're making it very clear you're a regular Register reader (and I don't mean that in a good way you troll).

D'oh!

[Anonymous Coward]

"Basic functionality of an OS is there"

Gee, you think? Hold on, the phone is ringing...

What a surprise, it's 1995! Just a second... you want what back? An accomplishment? What accomplishment? Something to do with software? Can anyone shed some light on this?

Another facet

[Anonymous Coward]

1991 ! back then was a different time. Today's corporations have a few developed reasons to make open source code. and if so why not. But I bet that 80 % number just reflects a change in business practices among corporations and far less about the free as in free/pay work. or it's quality, or what linus did with his free time, or obama care,.... taxes!!! death..!!! ummm aliens?

Unknown?

[INT_QRK]

Just curious. Does TFA, when ascribing contributions to "unknown," really mean "anonymous"? I can't imagine such a significant contribution by any truly "unknown." But why would a corporation or other non-governmental institution wish to be anonymous? On the other hand, I can imagine why certain government entities might. Rand(thoughts).

Re:Unknown?

[NotBorg]

There are a number of developers for whom we were unable to determine a corporate affiliation; those are grouped under “unknown” in the table below. With few exceptions, all of the people in this category have contributed ten or fewer changes to the kernel over the past three years, yet the large number of these developers causes their total contribution to be quite high.

The category “none,” instead, represents developers who are known to be doing this work on their own, with no financial contribution happening from any company.

"Unknown" means they don't know if the author's work is sponsored.

canuckistani

[Anonymous Coward]

One of the drawbacks of corporations becoming major Linux kernel developers is that corporate priorities start becoming the things that get the most attention. If something is important to regular users, but not so important to big business then it doesn't necessarily get done.

Contributions do NOT equate with usefulness

[Anonymous Coward]

A number of years ago, I bailed out of LKML because I was not interested in the pissing wars that happened. Yes, personalities on the part of "sanctioned" and "official" contributors got in the way.
Personally, I believe the kernel needs to have multiple forks:
- mainframes - the scheduler requirements/bias are much different than for the rest of the computing universe
- large servers - stability, a rather large memory footprint (256GB), and large attached storage are important
- web servers - 8-16GB memory, robustness of network features
- desktops - assume 8-16GB memory, focus on sw development. "make -j" and "scons -j 128" need to work well. Do NOT, repeat NOT try to replicate the stupidity of Win8.
- notepads and smaller form factors. Usability is all. OS irrelevant - it is all about interactions with whatever set of apps. UI HAS TO BE WONDERFUL. If not, it is dead.

Some sub-factors: very talented engineers put in a lot of effort to provide profiling capabilities for the HPC universe. As a broad rule-of-thumb: Ignore whining by "Linux Kernel Developers" who (I refuse to get into a pissing war with professional ass-hats) are intransigent and incapable of understanding that someone else might actually be capable of delivering useful performance monitoring techniques other than the "required" perf mechanism.

Re:Contributions do NOT equate with usefulness

[Anonymous Coward]

You don't need a fork for that.

The current system is proof of that.

F*ck Beta!

[sexybomber]

Who's writing Linux these days? One can only hope it's not the same people who are responsible for the abomination that is Slashdot Beta.