User: Password:
|
|
Subscribe / Log in / New account

Results of the Apache OpenOffice 4.0 Logo Survey

Results of the Apache OpenOffice 4.0 Logo Survey

Posted May 10, 2013 22:28 UTC (Fri) by rcweir (guest, #48888)
In reply to: Results of the Apache OpenOffice 4.0 Logo Survey by tialaramex
Parent article: Results of the Apache OpenOffice 4.0 Logo Survey

Long tail is overrated. I've looked at that claim in LibreOffice and they doesn't hold water. Yes, there is a long tail, but as a whole it contributes very little, and there is almost no progression from "long tail" contributor to "core contributor" over time. The development work there is still almost all done by professional developers, either ex-Oracle engineers now working for Redhat, or ex-Novell OpenOffice developers now working for Attachmate.

Remember, long tail has a cost as well, in terms of time to educate, to review, the hit to quality (which LO continues to suffer from), etc.

Put it this way: who creates the better product, 1 person with 15 years experience with the code base working 40 hours a week? Or 50 inexperienced people working 1 hour a week? It is a rhetorical question and the answer is obvious. Of course, we all hope that some in the long tail progress, gain experience and spend more time hacking. It sometimes happens also. But not as often as it does in open source fairy tales.


(Log in to post comments)

Results of the Apache OpenOffice 4.0 Logo Survey

Posted May 11, 2013 3:49 UTC (Sat) by BrucePerens (guest, #2510) [Link]

Put it this way: who creates the better product, 1 person with 15 years experience with the code base working 40 hours a week? Or 50 inexperienced people working 1 hour a week? It is a rhetorical question and the answer is obvious.

You're kidding.

Of course, we all hope that some in the long tail progress, gain experience and spend more time hacking. It sometimes happens also. But not as often as it does in open source fairy tales.

OK, you're not kidding. You're just disconnected from what happens in non-corporate-operated Open Source projects.

We learned all of this way back when the Open Source Applications Foundation brought professional development processes and the commitment of full-time employees to Open Source, along with the management of known top-performer Mitch Kapor and his hand-picked team. That didn't work out so well. But they were very good at eating money that should have gone to more worthy projects.

The 15-year-experienced guy who works 40 hours a week has a queue of things to work on that's written by someone in management and he doesn't get a chance to do much that's serendipitous or even innovative. He started with StarDivision and he's been through so many ownership changes and general corporate nonsense on the way that motivation isn't easy any longer.

The 50 inexperienced people put in more than an hour, some of them much more. They are full of fresh ideas and they have their own agendas. They come up with things that nobody expected, but which gain a following rapidly.

We're all waiting for IBM and Oracle to return to their own kernels because Linux development just can't keep up :-)

Funny how those committers who are writing twice as much code as the other project aren't actually keeping up when it comes to user-desirable features. Are they doing stuff that IBM and Oracle want and nobody else cares about? Are they writing for a living and not for love the way the other guys are? Or is this just what should be expected from corporate-run Open Source?

Perhaps those Open Source folks are not the only ones who have bought into fairy tales.

Results of the Apache OpenOffice 4.0 Logo Survey

Posted May 11, 2013 11:18 UTC (Sat) by rcweir (guest, #48888) [Link]

My point, which I'll keep brief so you might read it, is that the LibreOffice long tail is a fiction. For example, earlier in this thread it was said that there were "only" 18 active committers on OpenOffice.

But if we look at the top 18 active committers on LibreOffice we see that they combined do over 80% of the commits. So the vaunted "long tail" is actually very, very thin.

And guess what, this core 80% of LO are almost all corporate-sponsored developers. And many of them have the same tired career path and the same corporate direction same questionable motivations that you accuse us of. So your disdain is a bit stronger than your logic here.

Lets check the numbers ...

Posted May 11, 2013 19:42 UTC (Sat) by mmeeks (subscriber, #56090) [Link]

Just a quick injection of a few facts; if people want to generate them, we publish a full affiliation database ( and welcome corrections ) here: http://cgit.freedesktop.org/libreoffice/contrib/gitdm-config/ Sadly there is as far as I know, no public, accurate affiliation database for apache committers; so it is hard to corporately credit their patches 100% correctly. > But if we look at the top 18 active committers on LibreOffice we see > that they combined do over 80% of the commits. So the vaunted "long > tail" is actually very, very thin. At LibreOffice we -welcome- -enthusiastically- contributions, no matter how small. We encourage people to get involved, and appreciate that the code-base makes large demands on them and their time. I was personally thrilled last week to see a new-contributor create a one-line patch that fixed bug with seven+ duplicates: that makes me happy. Since I did a gitdm run a few days ago here is the data for the top 18 guys (your cut-off point):
Developers with the most changesets
Caolán McNamara          8099 (14.1%) RedHat
Tor Lillqvist             4382 (7.6%) SUSE
Kohei Yoshida             3068 (5.3%) SUSE
sb                        2300 (4.0%) RedHat
David Tardon              2276 (4.0%) RedHat
Michael Stahl             2065 (3.6%) RedHat
Miklos Vajna              1788 (3.1%) SUSE
Markus Mohrhard           1663 (2.9%) Volunteer
Norbert Thiebaud          1467 (2.5%) Volunteer
Thomas Arnhold            1406 (2.4%) Volunteer
Michael Meeks             1278 (2.2%) SUSE
Luboš Luňák               1240 (2.2%) SUSE
Andras Timar              1103 (1.9%) SUSE
Fridrich Štrba            1090 (1.9%) SUSE
Bjoern Michaelsen         1040 (1.8%) Canonical
julien                     925 (1.6%) Volunteer
Matus Kukan                916 (1.6%) LibreOffice contract developer
Noel Grandin               828 (1.4%) Volunteer
Personally - I'm thrilled to see so many volunteers keeping up and out-pacing full-time payed employees (that are after slot 18) there - we have some simply awesome contributors ! Hopefully that leave some space for the gitdm output of changesets by employer:
Top changeset contributors by employer
Volunteers                17188 (29.9%)
SUSE                      17045 (29.6%)
RedHat                    14742 (25.6%)
Oracle                    5410 (9.4%)
Known contributors        1313 (2.3%)
Canonical                  708 (1.2%)
Assigned                   234 (0.4%)
IBM                        180 (0.3%)
ALTA                       158 (0.3%)
Lanedo                     106 (0.2%)
KACST                      100 (0.2%)
Igalia                      75 (0.1%)
Aentos                      68 (0.1%)
Collabora                   52 (0.1%)
SIL                         40 (0.1%)
Tata Consultancy Services   34 (0.1%)
Apache Volunteer            27 (0.0%)
Linagora                    26 (0.0%)
Openismus                   13 (0.0%)
Bobiciel                     8 (0.0%)
Nou and Off                  7 (0.0%)
Munich                       2 (0.0%)
Funky                        2 (0.0%)
CodeWeavers                  1 (0.0%)
CodeThink                    1 (0.0%)
Intel                        1 (0.0%)
Hope that helps the deliberations.

Lets check the numbers ...

Posted May 11, 2013 20:46 UTC (Sat) by rcweir (guest, #48888) [Link]

My numbers differ, since I don't count phantom developers who are not involved in the project in any way, such as the ones you claim from Oracle, IBM, etc. A little intellectual honesty would go a long way here. The fact that you take code and merge it does not make the author of the code a member of your community.

Lets check the numbers ...

Posted May 11, 2013 21:13 UTC (Sat) by luya (subscriber, #50741) [Link]

Would you mind to show your own stat then so readers can view them? Thank you.

Lets check the numbers ...

Posted May 11, 2013 21:45 UTC (Sat) by rcweir (guest, #48888) [Link]

Certainly. Here are the LO numbers for the core repo for this year. (I assume we're more interested in the present than ancient history.)

Here's for the top 18:

Commits Engineer % Cum%
780 Tor Lillqvist SUSE 9.50% 9.50%
716 Caolán McNamara REDHAT 8.72% 18.23%
521 Stephan Bergmann REDHAT 6.35% 24.58%
502 Kohei Yoshida SUSE 6.12% 30.69%
441 Michael Stahl REDHAT 5.37% 36.07%
399 Markus Mohrhard Volunteer 4.86% 40.93%
373 David Tardon REDHAT 4.54% 45.47%
295 Miklos Vajna SUSE 3.59% 49.07%
289 Julien Nabet Volunteer 3.52% 52.59%
228 Noel Grandin Volunteer 2.78% 55.37%
221 Lubos Lunak SUSE 2.69% 58.06%
218 Andras Timar SUSE 2.66% 60.72%
198 Thomas Arnhold Volunteer 2.41% 63.13%
186 Peter Foley ??? 2.27% 65.40%
180 Eike Rathke REDHAT 2.19% 67.59%
138 Michael Meeks SUSE 1.68% 69.27%
135 Matúš Kukan LibreOffice Contractor 1.64% 70.92%
134 Fridrich Štrba SUSE 1.63% 72.55%

Other time periods will obviously given you other numbers. If you back out the Apache codesets, which LO gets merely via their status as a downstream consumer of AOO, the relative contribution of the top 18 goes even higher. (Since we're talking about community trends, not code, it makes no sense to count code contributions that did not originate within the community. Otherwise we'd count every author of every 3rd party library we use, right?)

So however you slice it, the "long tail" may be long, but it is very, very thin. This is the way it has always been with OpenOffice, from the earliest days. Most of the development work is done by professionals, and the community mainly works on translation and marketing. LO has not really changed this fundamentally. The "long tail" is just a myth they tell their SUSE and RedHat executives to convince them to continue pouring in cash to prop up the project. But with declining PC sales, and the niche Linux desktop market, I really don't see that continuing for much longer.

Lets check the numbers ...

Posted May 11, 2013 23:27 UTC (Sat) by Trelane (subscriber, #56877) [Link]

What are both your methods, please? Code is best.

Thanks!

Lets check the numbers ...

Posted May 12, 2013 0:35 UTC (Sun) by shmget (subscriber, #58347) [Link]

so the arbitrarily set at 18-core is 72.55% not 80%, and
and you own number shows volunteer = 16% of the total + part of volunteer out of the rest : the 27.45% 'tail'.

ok so now, what are the similar numbers for your project ? (that is code change, not web-page edit)

Lets check the numbers ...

Posted May 12, 2013 1:50 UTC (Sun) by rcweir (guest, #48888) [Link]

If it is not clear from my other posts, I don't think these comparisons are worth the effort spent gathering them. I know LO supporters think otherwise. They expend time making gathering numbers, publishing colorful charts, etc. They love comparing their numbers to AOO. In fact, when Ohloh was not updating AOO numbers I received quite a few emails from LO supporters complaining that the AOO numbers were not being updated. They even complained to Ohloh! (Oh no, how can we create our colorful charts if the AOO numbers are not there!)

But in the end, I think what counts is what comes out, the product that gets into the end of users, the features and quality that they receive. That is the value of an open source project. That's the difference between an open source project and a political party.

Maybe you don't see that in the insular world of the Linux desktop, but in the broader world Apache OpenOffice, the product, is doing very well. Our 4.0 release will be out soon, and we're soon going to hit the 50 million download mark in a few days. We're not just hanging on. We're advancing. Download are growing. Name recognition is growing. On the commercial side (and I apologize for not being able to give very many details here) AOO is also doing extremely well. I'm solidly focused on results, not on claims of community size. What did the Bard say, "The fewer men, the greater share of honor" ?

Of course, that's my personal view, and I realize there are other views as well, but those who promote those views have not (IMHO) delivered commensurate results.

Lets check the numbers ...

Posted May 12, 2013 4:01 UTC (Sun) by shmget (subscriber, #58347) [Link]

for reference, it is amusing to put the 'present' Rob in context with the Rob of 2 1/2 year ago:

http://www.robweir.com/blog/2010/09/libreoffice-newest-me...

" The key milestone I think will be if someday the Document Foundation can claim a headcount of developers that equals or exceeds that which Oracle has working on OpenOffice.org. In the end code talks, and developers write code."

I guess committers and commits where a 'key' milestone then... but not now...

http://www.robweir.com/blog/2010/11/the-legacy-of-openoff...
is interesting too. first it contain a graph, supposedly published in early November 2010, that reference Lo 3.3 and 3.4 and their date of GA in the future by 3 and 6 month respectively...
but also in the comment Rob put the emphasis once again on dev head-count:

"I’ve gone through the logs and the LO number shrinks from 150 to approximately 30 coders. That is a respectable number for a new open source project. I think LO should give a good, but realistic number, rather than trying to dazzle with the extremely improbably “150 developers” number. "

So back then, _that_ was an important metrics... and guess want, rob _could_ go to the logs again, and I bet you he did... but the numbers don't line up with his communication agenda... so let's pretend they do not matter anymore...

"it is not very surprising that Suse/Redhat/Ubuntu are distributing LO instead of OOo. Weren’t they distributing the “Novell Edition” (GoOo) all along?"

Remeber how Today's Rob talk about Linux distribution 'silently replacing OOo by LO'... well Rob from 2 1/2 year ago knew exactly what the situation was... he 'forgot' since then...

Rob from 1 year ago wrote a blog about such metrics...

http://www.robweir.com/blog/2012/04/free-software-marketi...

"Don’t just count those who are writing code. Almost anyone can be called a “developer” these days. Translators (“Localization Engineers”), build lab guys (“Configuration Management Engineers”), testers (Software Quality Engineers), etc. Include all of their contributions."

at first glance one may think that was a sarcastic blog, or maybe an April Fools joke... but no. 2 months later Rob applied his own advice and modified the 'subscription' in ohloh to get every modification of their website and internal wiki to count as 'dev' commit.
http://www.ohloh.net/p/openoffice/edits?page=6

"But be warned: use of these advanced techniques might open you up to criticism of promoting numbers that are meaningless,"

Well Today's Rob is meta-gaming ... he is now using the very criticism that he foretold about against his own shenanigan, to then declare that _any_ such numbers are meaningless... because he cooked his owns...
Genius!

"For example, suppose you have 400 developers, and 10 of them do 90% of the work, and they are employed by a single company. Avoid the naive mistake of saying that one company was responsible for 90% of the contributions. "

Which is pretty much the situation in Rob's project... except for the 400 dev part...
But he has preempted that problem quite a while ago, by declaring that IBM employee paid to work on aoo are 'volunteer' just like anyone else... and therefore it is irrelevant to figure out IBM % in the project...
(yes, yes he even went as far as arguing that since IBM employee are at-will employee and could quit if they wish, that make the case that they are 'volunteers.
but somehow that logic does not apply outside of his project... I guess he is taking the concept of 'fuzzy' logic to a whole new level...

"we're soon going to hit the 50 million download mark"
I guess it is best to let people see for themselves:
http://sourceforge.net/projects/openofficeorg.mirror/file...

Lets check the numbers ...

Posted May 12, 2013 13:18 UTC (Sun) by rcweir (guest, #48888) [Link]

First, your link to the AOO download stats was to the English version only. if you want the localized versions as well the URL us here:

http://sourceforge.net/projects/openofficeorg.mirror/file...

That link overstates downloads, since it includes Language Packs, as well as full installs, intermixed in the same directory. I have no idea whether LO includes these in their counts, since they have never disclosed their methodology. But we give full details of ours, include access to the raw data and Python script for getting the data yourself here:

http://www.openoffice.org/stats/downloads.html

Second, although it is good to remind readers about the development effort under Sun and compare it to what is occurring today, you make the error of confusing "headcount" with "counts of contributors". Actual headcount, the effort on the project is measured in Full-time Equivalents (FTEs). It is entirely meaningless to look a contributor counts to gauge the effort being applied to a project. That is like counting pieces of currency without looking at the denomination.

The rest of your complaints I have either addressed elsewhere.

And do let me know if you have any comments or questions related to the parent article, the one about the logo selection process for AOO 4.0.

Lets check the numbers ...

Posted May 16, 2013 11:51 UTC (Thu) by dag- (subscriber, #30207) [Link]

It must be the long tail that's spending all the time gathering numbers, publishing colorful charts, etc...

Oh wait, all that has not been accounted for in the statistics :-)

Lets check the numbers ...

Posted May 12, 2013 0:18 UTC (Sun) by shmget (subscriber, #58347) [Link]

"My numbers differ, since I don't count phantom developers who are not involved in the project in any way, such as the ones you claim from Oracle, IBM, etc. "

Since these developers are certainly not 'volonteer', that actually works against your argument...
I also encourage you to read LWN. They recently published a graph about another project that fail to produce anything of value -- based on your long tail argument --

https://lwn.net/Articles/547073/

Note that that project also has a very low rate of conversion to core-dev...
but more to the point, without a tail you can't convert any at all.

"A little intellectual honesty "

That is rich coming from you.


Copyright © 2017, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds