Fog Creek Software
Discussion Board




Study on Open Source and Closed Source Dev.

http://www.nature.com/nsu/030623/030623-6.html

They found open source seems to converge to less bugs quicker, but also they explain why.  It makes sense (to me anyway).

Enjoy & Discuss!

Raist3d
Thursday, June 26, 2003

Yes, make your code public and thousands of besserwissers will show up pointing out every flaw. For better or for worse.

Eric DeBois
Thursday, June 26, 2003

We'd need to read the whole article, not just the abstract.

One thing is that Open Source software is generally free whilst closed source is generally proprietory. In fact the Open Source/Closed Source and Public/Proprietory choices are completely independent.

So it is possible that in fact Open Source software is actually getting more reports and help because people feel they are contributing to the community.

I do suspect however that the main reason is that closed source alllows a cult of secrecy and hierarch to grow. I once pointed out in a meeting that it would take less formal steps for me to get to a meeting with King Fahd or Koffi Annan than it would to get to the Managing Director if I followed official guidelines.

We might all be only six steps away from Kevin Bacon but you can count on Management to add a couple of zeros to the figure!

Stephen Jones
Thursday, June 26, 2003

He wouldn't mention a slight difference in design approaches -

open source projects wouldn't go for features that
1) are not cool.
2) pose too much work, and too little job satisfaction (like documentation, or help files, or install packages).

The research doesn't mention thats a reson why GUI's are not often the strong point of open source projects (that's just the place where most of the bugs live ;-)
An open source project would just not invest into a fancy GUI,
problem solved.

Michael Moser
Thursday, June 26, 2003

there is just this order of magnitude difference between hobby projects and comercial systems

- some of this difference is made up of things like documentation, support for multiple languages, etc. etc (things that are not part of the core product, but you do actually need them)

- a lot of technical work goes into maybe 10% of functionality that makes up those 'missing features' that a comercial system would have to provide

Michael Moser
Thursday, June 26, 2003

A study of amateur (home) movies found fewer edits were needed than for movies by large movie studios. Clearly, this proves open movies are a superior method for producting movies.

.
Thursday, June 26, 2003

You know what?

I want to eat, too. I want to go to the movies. I want to have a nice vacation in some cool, relaxing location.

And if I program all day long, at least I'll do it for money and live a decent life.

What is the alternative? Write open source and live with my parents, as a poor person?

Nope. I need a decent life, so I shall only write commercial software.

Knight
Thursday, June 26, 2003

Iwould always assume that the "many eyes" hypotisis would actually favor big companies that apply dogfooding, since there are far more knowledgable eyes inside a company like Microsoft than most Open Source projects can ever dream of attracting.

Even so, I would refute the validity of a purely quantitative counting of "source viewers" since we all know it is the quality of the eyes that counts, not the quantity.

Just me (Sir to you)
Thursday, June 26, 2003

Michael,
I don't know how you backup your statement that it's in the GUI where most bugs were present. For me, another reason why some, but not all OSS provides a poor GUI, which makes more sense, is that the authors do not really want their software to be that professional-looking so that any crook could sell it as COTS.

Stephen,
I do not get the link between the licensing of software and it's management scheme. Actually, I do happen to know a bunch of OSS which is managed in quite a hierarchical way - e.g. Mozilla, which does have about 5 people, "product evangelists", which you cannot bypass by any means, and whose past decisions proved to be questionable -, and OTOH, I do happen to work for a company, working on closed-source projects solemnly, where you can talk to the CEO at any time and throw your 2 cents in.

Johnny Bravo
Thursday, June 26, 2003

Interesting...

Michael Moser - I disagree with the features argument. In open source, they tend to be necessary features first, then interesting features, and then minority features (those needed by a small group who build them.)  GUIs fall into this category.  Not because they are easy or hard, but because they were not originally necessary to getting the job done.

Not all OSS is equal, some could do better and some like OpenOffice are examples were it is being done at the commercial level. 

You then mention documentation and languages.  Compare the language support of most commercial versus OSS and you will find OSS can do better.  Microsoft must see commercial value in supporting a language (build time versus ROI), OSS need only someone willing to build the language interface.

If you have specific features, I would be interested in hearing them as there is always room for improvement.

Knight - Being unable to see how to make money under OSS is not the same as being unable to.  Did you write XP, or Unix?  VB or C++ or JAVA?  You can make money even though you did not create these with value add. OSS is no different. 

Mike Gamerland
Thursday, June 26, 2003

"the authors do not really want their software to be that professional-looking so that any crook could sell it as COTS"

Absolutely! Any competant developer would do the same. That's why I make my own software as buggy as possibly -- so that crooks won't want to steal it.

Ricardo Suarez
Thursday, June 26, 2003

Not intended to be a flaimbait....

How many IIS-worms/virii can you name?

How many Apache-worms/virii?

Urby
Thursday, June 26, 2003

Not intended to be a flaimbait....

How well does Windows work?

How well does J My Number 1 Project work ? ( get in from sourceforge. )

.
Thursday, June 26, 2003

.
Um. I think Apache vs. IIS is a fair comparison. The primary difference between the two is how they are developed. I think  apaches strengh is due to the development model.

Comparing Windows to someones first project is not fair. Try Windows vs. Red Hat or something.

Urby
Friday, June 27, 2003

Urby, open source pushers always fall back on a relatively small number of well-known and good products as arguments for the merit of a development methodology, yet that methodology produces tens of thousands of pieces of junk.

An accurate evaluation would compare all professionally developed products against all OSS products, then draw conclusions. That, by the way, is only one part of the total set of issues.

.
Saturday, June 28, 2003

*  Recent Topics

*  Fog Creek Home