[gentoo-dev] Re: Portage QOS

[Duncan]

Igor posted on Thu, 09 Jan 2014 16:44:02 +0400 as excerpted:

> There is no data to tell what happens with Gentoo (to give that data is
> one of the goals of the project). We only have some formal esteems from
> unreliable sources.
>
> According to distro watch:
>
> In February 2012, Gentoo distro was in 19th place.
> In December 2012, Gentoo went to 22nd place.
> In December 2013, Gentoo is down to 32nd place

There was some discussion of this previously.  The conclusion was 
basically that gentooers don't tend to be the trend-watching type, nor, 
really, are we really interested in the trend-watching type, as that's 
not gentoo's base or target user. Instead, our users tend to be 
independent customizers that aren't so interested in what the majority 
thinks, but, rather find gentoo's general support for letting them make 
of their gentoo installation what they will a very good match for their 
needs.  If that's not the best match or if their needs change, the fact 
that gentoo takes more work than many distros because you have to 
actually configure and build it, tends to have them quickly off to some 
other distro that's a better fit for their less time/interest, more 
cookie-cutter needs.

In a way, then, gentoo in the Linux ecosystem is what Linux itself is in 
the more general OS ecosystem, and gentoo tends to get only the self-
selecting independent thinkers who value the ability to make their OS 
what they want while never-the-less automating much of the process (thus 
we aren't Linux from Scratch), in the same way that the same group, but 
to a somewhat lessor extent, tend to be Linux users.

And just as a significant subset of those Linux users and devs value 
their (software) freedom and independence enough to not be willing to 
sacrifice it just to have Linux more popular and maybe exceed MS, so a 
lot of Gentoo users and devs aren't willing to compromise on gentoo's 
ideals of highly customizable individuality just to see us rise in 
rankings such as distro-watch.  If it happens, great, but it won't 
greatly affect the way gentoo is developed, and if it doesn't happen, no 
big deal anyway, since that's not something we consider significant or 
important, particularly /because/ we recognize that sort of user isn't 
what gentoo's targeting in the first place.

> According to Linux Counter
>
> In January 2012, Gentoo distro had 5.32%
> In January 2012, Gentoo had 4.04%
> In November 2013, Gentoo had 4,21%

I guess one of those January 2012s is supposed to be something 
different...

Same thing here, really, tho the reason is a bit different.

I know *I* certainly haven't registered with linuxcounter, and don't 
expect I ever will, either.  I see it as useless at best, and yet another 
way to be tracked at worst.  (I /do/ deliberately keep my browser's user-
agent string set to Linux instead of setting it to say the latest MS 
Windows version, and deliberately kept 64-bit back when 32-bit was the 
norm for similar reasons, so sites that I visit and thus care about can 
count that, but I most certainly do NOT let the various third-party 
tracing sites do their thing, using tools such as firefox plugins 
noscript, request-policy and cookie permissions, as well as privoxy, to 
help me keep that information out of third-party-tracker's hands.)

Tho interestingly, that does show percentage stabilizing or even 
increasing a bit between the second and third samples.  What it means, 
however, I'm not going to attempt to guess.  For all I know it simply 
means a few gentooers don't object to being tracked as strongly as they 
once did, which is actually slightly disturbing to me, tho it's their 
life so they get to decide, not me.

> And from my experience of Gentoo forums, gentoo.wiki - I vote for Gentoo
> at least not gaining new users.

That would be a more interesting number, there.  But you don't provide 
stats for that one, and personal perception such as yours above for those 
constantly involved is notoriously inaccurate.  Someone who left for a 
couple years and came back tends to see changes much better, for the same 
reason you don't tend to notice changes in a friend as you grow old 
together, unless you're separated for a few years and then meet again.

I wonder what the forums stats counts are.  I know there's mailing list 
activity stats as I've seen them posted occasionally, but I'm not sure if 
there's anything like that for the forums...  That would give us some 
concrete numbers to work with.

> If in several years the number of users is not increased - we can tell
> about stagnation.

As I've personally argued about Linux, if popularity comes at the cost of 
loss of freedom, etc, it's not worth it.  There's worse things than 
seeing numbers stagnate, and losing our ideals in a likely futile pursuit 
of popularity (what's the chances of gentoo topping Red Hat even if we 
forsook all that makes gentoo gentoo and tried? that's not what we're 
good at or care about) is one of them.

> It's all not very well thought after at this stage but immediate goals
> are like this:
>
>
> * Knowledge of [... multiple suggestions for tracking various things
potentially objectionable to gentoo users.]

I suspect that the various gentoo stats efforts failed for the same 
reason I suspect fewer than normal gentoo users are registered with 
linuxcounter... Gentoo users tend to be the independent sort, and have a 
distinct aversion to being counted or tracked.  A few might opt in, but 
not enough to get particularly good or reliable data, and if it were opt-
out or worse-yet hard-coded, we'd likely lose a lot of gentoo users over 
it, not just because of the tracking objection itself (many can patch 
that out if they have too, as I did gentoo/kde's hard-coded semantic-
desktop stuff here, when they tried to dump the flag in early kde 4.11, 
tho fortunately they returned it before 4.11 stabilized), but because 
that sort of hard-coding would be a betrayal of everything that a lot of 
gentoo users have come to gentoo FOR, so were it to happen, it'd be time 
to leave.


Meanwhile, those of us who have been around gentoo for a few years have 
seen the "gentoo's stagnating/dying and here's what it must do to be-
popular-again/survive" thread several times over, by now.   Gentoo's 
still here; I'm still here.  Those ideas... aren't... until they come 
around for another round, as they seem to do every couple years...

And FWIW, gentoo's number of devs rose until it hit something around 300, 
then it fell back a bit (I think it reached 350 or so before they started 
actively retiring devs who had disappeared for quite some time, but I 
believe the number of active devs has never much exceeded 300, if that), 
but has remained relatively steady around 250-ish devs, 200 or so active 
depending on definition of "active", for several years now.  Sometimes it 
goes down a few, then it goes up a few, but overall it remains about the 
same.

Which actually fits various organizational/group dynamics models, 
apparently, too.  There's (apparently, this was posted one of the other 
times a discussion like this came up, and it makes sense, but I've not 
looked into it further) some studies to the effect that there are several 
group size thresholds.  IIRC (and I may not) the maximum effective size 
for small groups was 20-50.  At about that point, conflict goes up and 
groups either adapt and change their practices to grow, or drop down 
below that number again and tend to stay there.  There's another such 
threshold at 250-350, forcing further group practices adaption to grow 
further.  A number of FLOSS groups reach that one and never pass it, and 
gentoo has been right there for some years now.  But if that threshold is 
passed, the group can grow relatively unrestrained again, to a size of 
several thousand.  I believe Debian is one of the few all-community 
examples of passing the 250-350 threshold, but that they're stuck at the 
next one, 2500-3000.

I think the kernel has passed the 250-350 barrier now too, with git and 
the distributed hierarchy of kernel lieutenants, codified signed-off-by 
practices, etc, helping with that.  Before the distributed git and 
bitkeeper before that on the technical side, however, and before the 
hierarchy of kernel lieutenants was established, there was a real crisis, 
as Linus really was becoming the bottleneck, and nobody really knew how 
to fix it as there's only so much one man can do.  But they worked thru 
the issues and busted that cap, and the kernel's moving faster than ever 
thought possible, before.

If that is indeed the case, then in ordered to grow, gentoo needs to 
figure out how to get past that 250-350 developer threshold, and until/
unless we do, we'll "stagnate" in at least active developer numbers.  IOW, 
it's primarily a social/organizational problem, not a technical problem, 
tho as with the kernel and bitkeeper and then git, the right technical 
tools can help.

Actually, in that regard it's very possible that gentoo's long planned 
and worked toward cvs-to-git conversion will help finally bust that 
barrier for gentoo as well.  Time will tell I guess, but that's one more 
reason to try to help make it happen.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman