Andrew C. Oliver wrote:
This is going to be another one of my long answers to a short question...

Good! (I crosspost to community. I think it really belongs there ;-)

Some context:

Howard M. Lewis Ship asked about Tapestry/POI usage:
People keep asking me "how many people are using Tapestry"
... and I honestly have no idea. Insufficient feedback.

Do you have a way of determining the user base of POI? Any guidelines based on downloads?

Andy answers:
I don't really attempt to measure this. It would be trivial to measure the number of downloads from the access logs; however, I prefer to mesure it subjectively.
Note that its documented on the Jakarta site that Opensource is not about units shipped. I'd look up the page but I'm sure that if I don't someone will do it for me so why bother.

Specifically in server side applications. For instance, as Andy hints in my next quote, a single download from a intranet server in a big corporation can lead to tens of thousands of (unsuspecting) users.

(...big snip, not that I don't like it, but please read it in the archives)

First, POI attacts mail from some of the largest banks in the word, financial institutions, governments, millitary institutions, nuclear power plants, etc. There is even a large Apache backer flirting with the idea of using it (while its irrelevant to me whether they do or not, it is relevant that they are considering it).

Next, I measure the success of it by two other things: Microsoft's flirting with open file formats (I'm sure it will be "open" in that Microsoft sort of way) and the final crux will be the day this goes out of business. The first clue to eventual success is that Tidestone has re-emerged as a seperate business entity instead of just a redirect to a page on Actuate's site. The second is that they have lowered the price from 15k per processor to 5,000k per server (I'm sure there is a big astericks) This is after an extensive advertising campaign including full page adds in Dr. Dobbs. This is despite some functionality that we do not yet have.

I don't agree that it is a good metrics, since it's a crisis situation and a lot of other factors could be involved into pricing (product life cycle, etc.). Also, we are not trying to make anybody unhappy, that would be (at most) a side effect of our approach being successful. But the post goes on:

My final measure is how much money I'm making and how many other POI developers I'm able to cut in on it. Thus far (this year) I'm able to derive 35% of my income from opensource efforts (a percentage which is up about 800% from last year). I suppose all of those are directly or indirectly related to POI. I'll undoubtably be flamed for this unique viewpoint, but its a measure which I find important. I've managed to pass on some of this work to two other POI committers thus far. (no one bother writing me offering to do this work, I only pass this work on to contributers to the project)

So to me how many people are using POI and not contributing to the project in any way is totally irrelevant. I measure it in actual benefit to myself and the other contributers. To me any other mesure is trivial.

This is the point I think merits further exposure/discussion. I'm not going to flame Andy on this, since I fully agree with it. If we cannot extract actual benefits from our involvement in Apache projects, the projects will not work/scale well.

Each and everyone involved in Apache projects should benefit in terms of:
* better career opportunities
* being better known in the industry
* having better tools in our daily work toolset
* higher productivity in integration
* knowing where technology is moving
* __fill more here__

The Apache licensing model is oriented towards consultancy/system integration rather than product sales. This is in opposition to other licensing schemes like GNU:

* If you hold the copyright of a GNU licensed stuff, you can re-license it as closed source (a lot of GNU-licensed projects are doing this, see Aladdin or Transvirtual with ghostscript and kaffe)
* If you hold the copyright of an Apache, BSD or Artistic licensed stuff, it is far more difficult to do this, because everybody is free to do the same.

This introduces an asymmetry I don't like WRT GNU licensed projects: the person transferring copyright looses rights WRT the person holding it. I don't critizise this approach with the FSF proper, but I don't like, for instance, kaffe benefiting from my patch and I being unable to benefit in the same way.

Thus, I find that people doing system integration and consultancy, both in big and small companies will naturally prefer Apache-like licenses:
* you don't need to care about your customer wanting closed modifications, as they can do them --> less overhead
* you don't need to care if your customer wants to redistribute the output --> less overhead
* if you happen to find a niche where closed source or just integration with closed source makes sense, you can do it

This, IMO, is one of the big "assets" and branding elements of the ASF (together with quality). YMMV applies here.

Some people will find that this would enable commercial companies to "cheat", by keeping their patches private. It doesn't work in the long term:
* Open "variant" evolves faster, so maintenance of patches is a nightmare (Brian has an interesting essay on this (look for software as a liability), I know this from my fingers :-) ).
* Private extensions/functionality requires inmense ammounts of money to get people to know about it and use new APIs, while exposing a public API and a RI in Apache can give this with far less investment
* Open knowledge flows ordins of magnitude faster than closed knowledge

Conclusions? not many:
* Community success is community (user and developer) benefit, not downloads or size. This is what stroke me of Andy's post
* I find Apache licenses to make us "freer" than free software, removing significant business opportunities from our portfolio
* I like being a small part of such a big movement.



P.S.) You know I'm in flux when I nest parenthesis in my writings (inheritance of my LISP past, I noticed re-reading this rant) ;-) BTW, I don't know if it is legal to do this in natural languages.

P.P.S) It's not polished, but I *needed* to express this. It's just what I think, and I *don't* want a licensing flame war.

--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to