[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnu-arch-users] "tla commit" generates a patch-set even if there ar

From: Tom Lord
Subject: Re: [Gnu-arch-users] "tla commit" generates a patch-set even if there are no changes
Date: Mon, 17 May 2004 16:40:12 -0700 (PDT)

    > From: address@hidden (Julian T. J. Midgley)

    > >I'll raise a yellow card there on the rhetoric:  unsubstantiated
    > >(i.e., made-up) statistics combined with an ad hominem attack.

    > This thread itself contains the evidence that a significant proportion
    > of current tla users seem to dislike the current default.  

It doesn't.  That's a misconception.  You can't (directly) get much
about user demographics from -user mailing lists.  Most users are
silent and the ones that post are very likely to not be a
representative sample.

    > It is reasonable to assume that a similar proportion of future tla users
    > will dislike it, and it also seems reasonable that "thousands" more
    > people will use tla in future.  There's nothing "made-up" about the
    > "statistics".  

The "made up" thing is the axiom you're implicitly using that says
"g-a-u posters are a representative sample of the user community".

At best that axiom would need to be verified empirically but hasn't

But in truth, try to picture it the other way 'round:

You (presumably) use 100s or 1000s of programs, week to week.  I'll
assume you run a GNU/Linux box (or can imagine what that would be
like) --- all or nearly all of the programs you use have maintainers
and projects and mailing lists, and so forth.

How many of those project communities do _you_ speak up in?

Suppose you happen to be a reader of the list for ... i dunno ... say
`sed'.  And somebody proposes some really screwed up feature idea.  
Worse yet, 10s of others chime in and second the motion.   The feature
_sounds_, superficially, like a cool idea ---- but if you try to work
out the details, alas it would do more harm than good.

So, 10s of people clammer for this particular brain-damage that,
unfortunatly, is close enough to a good idea that it's hard to explain
_why_ it's brain damage.   And meanwhile, maybe 1 of the
co-maintainers or luminaries speaks up, says once why it's a bad idea,
and then ignores the thread.

Judging by the mailing list alone, and not diving deep into the
details of why the idea works or doesn't work, what do you see, as a
reader of this list?

You see maintainers making a not-obviously-correct unilateral
rejection of a "majority" of the user community.

So, I asked you to imagine that you're a reader of the sed list, not a
poster.  Now imagine that you recognized all along that the proposed
feature was a dumb idea.  10s of people clammored for it but a
recognized name batted it away.  That's sufficient, isn't it?  You
don't feel compelled to become a poster on this list just to help
correct the 10s of pros to the 1 of con?

An analogous scenario to that is perfectly consistent with the
evidence available on g-a-u.   There's no reason not to entertain the
hypothesis that (a) most users don't even read g-a-u;  (b) most
readers never post to g-a-u;  (c) there's little or no correlation
between the distribution of opinions on g-a-u and those in the general
user community.   In short, you can't conclude anything, with better
than random odds, about the democratic opinions of arch users by
counting posts on g-a-u.

I'm not writing this ramble to flame you, especially not you in
particular.  It's a pet peeve of mine, that's all.  It happens every
month or two that someone arguing for some feature on g-a-u says
something equivalent to "most users will...." or "most users
think....." or "haven't enough people posted about X so that you know
how most users think...."  The posts expressed that way are pretty
much never right.  It's always the argument people think of when
they're out of logical arguments.  It's a yellow-card condition -- a
kind of "oh, there's my cue to stop listening to this person for a

There is a much, much better substitute for counting posts on g-a-u:
the "trusted lieutenants" system.  There's maybe a half dozen or 12 people
who I know, from lots evidence, make heavy and thoughtful use of arch
in the context of doing real world work.   (I'm sure there are more
people than that using arch seriously --- I'm just saying there's 6 or
12 who I know about fairly well.)   I observe that many of those
people listen to lots of other users -- either here on the list or
within the context where they use arch.   It isn't just 6-or-12
individual opinions I notice -- it's like 5 people summing up the
opinions of this or that user community and 3 people just offering
their personal perspective.

That group of trusted users disagree on some issues, agree on others.
They're very good at figuring out and then articulating
_logical_reasons_ for their opinions about proposed features and

I don't know that that group is in any way statistically
representative of the user community's opinions --- but I do know that
they are a very good choice to be a de facto representative
legislature, their collective opinions reflecting the _best_interests_
of the general user community.

So, anyway -- don't say "most users think..." or "newbies will all
...." or anything like that.   It just plugs up my ears to anything
else you might have to say.

    > As it happens, you've convinced me that the flag is an inelegant
    > solution, and your log file changes will reduce the extent of the
    > problem to a manageable level, so I'll let it rest.   

Thanks.   Yeah, like I say, I'm sorry to get up on a soapbox about
your rhethoric.   It's not about you personally, really --- this
message about how to argue effectively on g-a-u is like a
once-a-quarter public service announcement I have to make and you
accidently gave me an excuse not quite flimsy enough to preclude me
from going on a rant.  So, thanks :-)

About the flag --- yeah, I was half-convinced (more than once,
apparently) that such a flag was needed.   The log file changes didn't
occur to me until just a few days ago when I posted them --- they
definately counter the rationale for the --force flag to commit.   
It that hadn't occured to me, I'd probably be (mistakenly and
begrudglingly) in favor of --force by now.

But I'm not, so .... Pffft.



Like my work on GNU arch, Pika Scheme, and other technical contributions 
to the public sphere?   Show your support!


address@hidden for payments.


        The cause of death is birth.
                        -- George Harrison

reply via email to

[Prev in Thread] Current Thread [Next in Thread]