The Care and Feeding of Open Source Software

An anonymous reader writes "You might find The Care and Feeding of FOSS (Free Open Source Software) interesting. This article debunks a lot of the myths and misunderstandings about the open-source software development process."

Comments

The real reson

[j_heisenberg]

... FOSS has flourished in recent years is a tiny nuicance up in Seattle. Microsoft crushed (almost) all competitors in their main markets, OSes, productivity suits and browsers. The only way to avoid this fate was to produce free software, using the same tactic MS has employed.

Another explanation would be: A lot of highly trained, intelligent and creative people have rather dull jobs, maintaining or servicing existing technology. They want to "realize their potential", and they do open source.

but what about the programmers?

[Anonymous Coward]

they need care and feeding aswell
ideology behind open source is nice of course, but some of us have to make living by writing code
open source isn't always the solution for everything

Re:but what about the programmers?

[DrLZRDMN]

Do we need to make a living writing code? If you are a hobiest programmer than no. There are other ways to make money such as selling hardware, support, or installation services. You are also assuming that open source programmers neveer getg paid, this is not true either.

Re:but what about the programmers?

[saigon_from_europe]

You are missing the point. Instead of fruitles multiplication of people doing the same job, now we can code things that are really different.

I used to work in company that produced wireless routers (for our internal usage only; unfortunately we never reached the stage to sell them). They were based on Linux. In propriatery world, we would have to buy licence for OS. Then, we would be limited in changes we could made to them. And many many related things (no documentation, no source of drivers...). Company with 3 developers would have no chance to make anything similar.

So what is the point: Instead to write OS from the scratch; to write drivers and similar things we used existing free technology. My main task was to write bash scripts, to patch kernel and similar. Other guy wrote sofware for automated control.

Basically, we did not try to reinvent flowerpot hole (i.e. writing something that already exists), we focused on new value - i.e. things that were specific for own company.

I can say, without FOSS, what we tried to do would be totaly impossible. In this case, FOSS created 10 new jobs.

(Unfortunately, they lasted only one year, but it was not related to FOSS.)

In more general terms, FOSS enables you to move from general things to company specific things. All companies calculate taxes same way - but many many companies have their own in house built software for that. At the other side, there are no two same companies with exactly same business logic. FOSS, in theory at least, allows you to spare resources from coding tax logic (common tasks) and to transfer them to coding support for your business logic (specific taks).

To be honest, this happens also in propriatery world. SAP has same tools for all companies, but their (expensive) consultants will (hopefuly) customize their software for your own needs.

Re:Paints a pretty picture

[salesgeek]

Its main effect is to take out all the small/medium players and polarize the market into FOSS and the commercial giants which is unfortunate because the smaller commercial endeavours are where (as far as I can see) most of the innovation tends to come from

Maybe a more positive way to view this is that FOSS commoditizes markets that are way overpriced. This is especially true of:

* Compilers and software development tools.
* Operating Systems
* Productivity suites
* Web servers & Application servers
* e-Mail services

Now even the tiniest organizations can use FOSS to gain an advantage. I was able to start a regional online advertising network for less than $500 by using a commercial, open source ad manager that I've retooled to fit my needs.

FOSS also tends to lag behind the technology curve so by the time it starts to mature the market has moved on, creating new options for commercial software so FOSS will constantly be chasing a moving target.

I have got to disagree with this on many, many levels. A great many of the innovations you take for granted now were invented via FOSS (try basically, the internet as a platform, web browser, email, etc). FOSS has nothing to do with commercial or not. It's about the app coming with the source with rights to modify (oss) or source + right to mod + right to redistribute (free). There are plenty of commercial apps that are oss. There are also commercial apps that are Free as in speech. I think what you are comparing, is for example, OpenOffice vs. MS Office, which compares a commercial closed source with a semi-commercial FOSS one. Even so, OpenOffice has capabilities that MS office does not and MS Office has capabilies that Open Office does not. The technology curve is more of a 3d wave than a 2d curve - it's possible to lead in one area and follow in another.

Don't assume that because an application is FOSS that it is somehow inferior to a closed source product.

Can OSS survive?

[nodehopper]

I think this Article includes some points that I see developing.

"Both FOSS and commercial versions may coexist, but commercial forces usually dominate the innovative process at this point."

As Corporate America embraces the OSS culture it will take what was once an "IDEAL" and find ways of turning it into something that is good for the Corporation.

I think a great example is in the Music Industry. A genre of music will develop in some obscure place creating something living and evolving and vibrant. Then the big record companies will discover it and then it starts. They milk it for every cent available until that great sound becomes lost in the watered down mass of "product" that they churn out. Soon the creative subculture has been bought and sold and cloned so many times it losses everything that was good about it. Could this be starting to happen with OSS. Red Hat has become a corporate entity. Suse is on it's way. When the cash offers becomes so large that OSS developers start to sweat and their knees get weak....will the "IDEAL" be enough any more? Will the culture survive or will it sell out?

Not sure about this

[GebsBeard]

I'm sure I'll get modded flamebait but this article strikes me as deluded. It assumes the following:

1. The market leader once gaining dominance will just choose to gravy train their product, ala IE.

2. The flow of new features goes from the commercial version to the FOSS version. The commercial entity apparently isn't capable of absorbing new features from the FOSS version.

This is a crock. A commercial version can remain hyper competitive even in the face of the "FOSS onslaught". There are numerous tricks than be pulled. Obscure or constantly changing file formats (ie .doc), cross integration with other products such that the whole "web" must be replicated to be truly competitive. And let's face it, not all software is an OS. Or an Office suite or a web server. Especially complex client server products can have dozens or even hundreds of interoperating processes; this stuff is a nightmare to replicate. You may find these tactics unpopular or even unethical but that's just the way it works in the real world. Large successful companies can and do make it painful in the extreme to erode their market share. The author seems to ignore the economics of market dominance.

Conclusion mostly correct, reasoning wrong

[wrook]

I have to agree (mostly) with the author's conclusions (or at least the ones I didn't ignore :-)). However, I disagree with the reasoning.

FOSS dominance in most fields is inevitable. I'm not talking market share, I'm talking stable features. In a prorietary world, features cost money. Bug fixes cost money. Even worse than that, they cost resources (one can have money to do something, but not enough people to pay to do it).

Currently, proprietary providers live with a difficult economic reality. They invest 10% of their expenses on R&D and require at least a 5% total profit. This means that you need a 10.5:1 return on R&D investment in order for a feature to be worthwhile. Since not all development works out, successful features need to be even more profitable.

When you first start, there is lots of low hanging fruit and even a 20:1 return on R&D investment (what most VCs demand) is quite reasonable. As the product matures, features get more difficult and more and more of your R&D gets used up by support (bug fixes etc). Not only that, but by Brook's law we know that adding R&D resources slows things down (due to communication overhead). This means that's there's a practical maximum of resources that can be added to the project. At some point, development slows down to a crawl.

FOSS, by nature of the fact that there isn't an expected return on R&D investment, runs into no such problem. In theory, there are infinite resources availble to the problem.
If a feature is desired enough by someone, it will be implemented. The return on investment (benefit vs cost) only has to be worth it for *one* person. The benefit does not even have to be returned as money.

Not only that, but FOSS operates in an evolutionary way. HUGE numbers of resources are expended on projects which yield no results (just check the abandoned projects on sourceforge). But it doesn't matter. Those people eventually migrate to the successful projects. If I develop A and then discover B is better, I can abandon A with no cost to myself. Eventually the more successful projects end up getting more and more resources.

FOSS generally doesn't need to worry about Brook's law, because wasted effort is irrelevant. FOSS projects can afford *not* to communicate thereby duplicating effort. The popular version will win out and everyone can migrate with no loss (generally speaking) to themselves.

That is why FOSS feature/stability dominance is inevitable (generally speaking). So why doesn't it always work that way? Well the first problem is recruitment. At the beginning, there may be many different competing projects. Until the consolidation period (in the author's paper), there may not be enough resources in any one project to compete with a proprietary provider's VC backed investment. This is not always the case. A good example where FOSS was *way* ahead of the curve is window managers. I think most people would concede that OSX temporarily tipped the balance, but not for long.... FOSS dominance is inevitable. Good ideas will be copied, bad ideas abandoned and new ideas will be forthcoming.

The other area where FOSS does not dominate is in projects where there isn't enough interest to do development. Word processors used to be a good example. Nobody wrote one because nobody used one. Now there's are a few groups of people with a bee in their bonnet about office suites. Given enough time, most areas will probably be dominated by FOSS. They may lag behind the curve, though, if developers don't see the reasoning early enough.

Finally (whew!), feature/stability dominance != market dominance. I personally don't agree that IIE *ever* had feature dominance over Apache. However, market share is market share. The thing about FOSS, though, is that it never goes away. You can try to kill it, but it will just resurface a year later. With proprietary software, all you have to do is cripple or buy the producer. With FOSS, you are faced with the prospect of a never ending propoganda program (making your product even *more* expensive).

The *only* way for proprietary companies to successfully compete against FOSS is to make FOSS illegal. We need to be vigilant.