[aur-general] [arch-general] Fwd: [arch-dev-public] pkgstats: first results

[w9ya]

You guys asked for some numbers. Below I both answer Aaron and make a stab
at how lousy our data really is.

On Mon, Nov 10, 2008 at 2:04 PM, Aaron Griffin <aaronmgriffin at gmail.com>wrote:

> On Mon, Nov 10, 2008 at 2:44 PM, w9ya <w9ya at qrparci.net> wrote:
> > If there is a problem with the servers and the load becuase of the
> community
> > repo, please let's talk about that and how the community can help. I was
> > reacting to input that was already forming a basis for using a bot's
> output
> > to determine and regulate a TU's contribution. Further it was NOT clearly
> > pointed out that this was being discussed because of a server/trunk load
> > issue UNTIL it was challenged.
>
> Ok, let me put my entire reasoning quite plainly.
>
> You think "when the voting was added to the TU system the community
> and TUs were SPECIFICALLY told that the voting system would NEVER be
> used to make decisions about or demands concerning what any individual
> TU decided to add" (which, I might add, I cannot find recorded
> anywhere. Would you mind pointing me to some links, please?)
>
> I think that that's fine, but there is WAY too much garbage thrown
> around for no reason other than "oh, I wanted to try this new app, and
> put it into community! I stopped using it 3 minutes later, so I
> orphaned it" (read: WillySilly). This is unacceptable on the official
> archlinux server. It's not in the _spirit_ of the repo. We need SOME
> form of metrics, SOME form of control, and we have one.
>
> So no matter what "TUs were SPECIFICALLY told", I think that's a load
> of crap. You will always have people abusing a system. It is my
> personally opinion that the voting should matter more to people. If
> the voting system is flawed, propose another system. Don't claim "I
> should be totally free to do whatever I want on your server" - that
> reminds me too much of the Dave Chapelle skit where Rick James puts
> his dirty shoes all over the couch. You are not Rick James.


I am really trying to understand why you insist on using such limiting an
self-depreciating launguage and so forth. I guess to make a point. I wonder
about how it reflects on you.

In any event; Here are some specifics. There are 11,800 (approx.) registered
users and yet the most popular package gets
a mere 400 or so votes. Now this is an interesting number as it both
represents close to 90 % of the eligible voters AND only 3-4% of the
REGISTERED users !! How is this so ?

Well, only 500 or so users have downloaded the aurvote package. Further, and
this is EQUALLY important to consider, a user can download and NOT be
registered at all !!

In other words 3 votes can represent a large number (several hundred users).
If 3 votes represents many hundreds of users, is having only 2 votes, and
let's say a potential of 60 - 150 ACTUAL users ,not a good enough reason to
maintain a binary package ?

I could go on, but you should be able to understand that before the voting
system CAN be used for a metric, it has to reasonably be able to used as
such. The current system simply cannot be used as such because the numbers
have NO MEANING because they represent nothing of value. Do three votes mean
300 users, or only 10 ? Since we cannot pint to download data, and correlate
it to those who(m) vote, and whether those that vote are the most focused
users or that they represent the average user, or some other sort of user,
you have NOTHING real that can be taken away form the number of votes.

Further it should be noted that IF the problem is storage/cpu/networking
resources then doing this accomplishes merely a temporary solution, and one
that comes with a serious cost that I have touched upon earlier today.

I suggest that we look at ways to have a on-going fund raising program to
allow for proper resources for the "community" system.

*Nothing* I have said above is either new or something that hasn't been said
earlier today on one form or another. Yet a proposal has been offered, and
ONLY this one proposal has had your support so far today. If you are worried
about server resources as an issue Aaron, please let us offer other
solutions to you, after all a good fundraising effort was NOT something I
saw you get behind or discuss so far, so I am left to wonder if you would
support it ?

Best regards;

Bob Finch

P.S... The emails for the TU voting system discussion should be extant for
you to view, although I say "should be" in as much as there have been some
issues with users, passwords, and presumably other email list storage over
the past few years going "south", so to speak.

And as you suggest above, your input on this is based on the needs of the
future and in previous emails you have declared that that was history, and
"stupid: to be discussing as it was "regressive". Of course I disagree with
this assessment as history can tell us not only what was promised, but what
the reasons are behind previous discussions and courses of action.

To be blunt, I also reject the idea that looking backwards is "regressive:
and forward is "progressive". As any dictator from the 20th century has
ALWAYS done NEW "regressive" things, things called regressive in that such
thngs take away prerogatives from the public, all in the guise of
"improvement". The results are ephemeral at best, and yes you fell into a
trap of using the same language they used, Ala; "progressive".
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://archlinux.org/pipermail/aur-general/attachments/20081110/afd679ed/attachment.htm>