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

FOI rights, rejecting Debian (was ...general-purpose workstation

At 2003\04\24 11:32 +0100 Thursday, Lane Myer wrote:
>Hello all,
>I am very close to replacing another BSD on an old laptop with OpenBSD.
>Aside from issues of running on laptop hardware, what makes a spartan,
>server oriented OS like OpenBSD good for a general purpose workstation?

Below I write on Debian somewhat uncertainly. There could be some
sort of U*nix problem or just some scandal in Debian of getting the
leader picked wrong, that is not going to spread widely.

At the bottom I give some impressions on OpenBSD.


OpenBSD has out of date documentation. That is a big selling point for
me, since there is an understanding that eventually they will copy from
FreeBSD and complaints might be fixing documents that would be discarded.
I.e. it reduces the chance of the project not really being in the right
if down and out in some argument with me. A lot of the need for tiny
comments can be deflated here. I can't see what is happening about
keeping the whole OS up to date.

I installed OpenBSD in the last month.

I have a theory that FreeBSD does not have rights for staff members and
Debian is similar, and OpenBSD is not.

The Debian project uses a Condorcet method to pick the leader of the
whole project. Last month that found Mr Martin Michlmayr to be their

The results were these (after eliminating a very small fraction of the
 * Martin Michlmayr [current leader], 32.7503%;
 * Branden Robinson,                  35.0137%;
 * Bdale Garbee,                      32.2359%.

Last month they should have picked the 2nd, i.e. Mr Robinson, since
monotonicity implies a 1/3 quota to lose against. Their debian-vote
mailing list seems to want to know nothing and also they want to believe
Condorcet is best while debating or proving that false view.

I did not see Debian saying that the winner was right. They said it would
be bad to dump him and reaffirmed the false idea that Condorcet is best.

It is not possible to walk in, swing the door against one, hang a coat
onto a second, and get the third to think "I am not bright enough to
solve these algebra puzzles". A passing expert may expecting a rule
saying that the final thought has to be spoken. Nowhere here do I see an
OpenBSD opinion or rule on that topic. The best could stay after arriving
with FOI rights. A fourth could brew some tea for the guest while there
is a wait.

Here is the mailing lists at one of its most relevant moments, where
it actually considered the idea of corruption in a preferential
voting method:
At 2003\04\22 12:13 +0200 Tuesday, Matthias Urlichs wrote to
   Subject: Re: Robonson wins [...]
   Date: Tue, 22 Apr 2003 12:13:52 +0200
>Craig Carey wrote:
>> The ballot paper
>>     44444
>> is a special paper that gives the voter a power equal to 50,000 times
>> the power of all other ballot papers. Only Mr Urlichs knows that.
>> The method can be used to elect the leader of the Debian project,
>> but due to DCMA encryption issues it a widely trusted black box.
>Oh wow. My net.kook meter just went off the scale.
>Few people manage to do that.

The form the Condorcet algorithm is presented was legalistic &
algorithmic, hence no expert could understand it to be good. Experts
would see voting programs as special type of programming where the
"if" statement, sort of, promptly leads to an unfair algorithm.

>Oh yes, welcome to my killfile. You should find the company in there
> quite acceptable.

That is as good as it gets in Debian. A passing expert can't interact
with something so dumb as the Debian-Vote community. It actually threw
away a lot of votes last month because their Mr S did not know what to do
with kth-equal preferences. I guess that Mr Rob L might be transferring
cash to keep Debian using Condorcet. However a group that can't identify
a topic (why is the winner wrong and the method unfair this year too?,
and how can pairwise comparing be more fair than a competing equation
defining fairness exactly ?).


A hypothetical OpenBSD: official information is made available unless
there is good reason for withholding it, etc.

In preferential voting, a simple princple of duality is that preferential
voting methods pass this test: the same method results if winners
switched with losers adn the vote counts are negated.

  True = (All X)[Winners(X) = (A - Winners(-X))],
  A = "Set of all candidates"

Just as clearly, there can be a Freedom of Information style principle of
THE DUAL OF SECURITY that provides a right to get the damaging
information out. [A spike of the fish need to be used or else I might as
well hear the story on how the mascot/emblem was selected or wrongly
chosen. Maybe it should be changed if not quite right].

A request could be sent to DARPA. A post-hoc analysis would be better
than confident statements from USA asserting that Americans do not have
statements of reasoning. The US journalists apparently have to hint that
the DoD was wrong. 

USA has a mailing list on FOI rights: grim reading: FOI-L:

    Homepage: http://www.nfoic.org/FOIL.html
    Archives: http://listserv.syr.edu/archives/foi-l.html

    Owner: National Freedom of Information Coalition

I was having a problem with Jonathan Bresler, the FreeBSD postmaster who
was deleting e-mail between FreeBSD's mail server and not replying to my
e-mailed questions except insufficiently. When I figured out that ISPs
were the target under some bad anti-spam offence I quit. Would not wish
that for others: associating with a damonic project where the red staff
daemons lack a power to demand that the black elementals 'daemons' answer
questions. I can't think of a name of a principle making it sound like
its best for staff members to have no right to know the smallest amount.


In case the question was hoping for some plain comments on whether to
avoid OpenBSD:

That OpenBSD 3.2 installer is unfriendly. OpenBSD 3.2 can't see NTFS and
it can't see Ext3 (any fix coming?) and there is a maximum of 10-11?
partitions per HDD or something (16 in total). (I don't know about more
on disk #0 using a 2nd OpenBSD partition while staying with the
/dev/wd0*** naming scheme). OpenBSD with host networking ran in VMware
easily (not tried X windows).

A restriction of OpenBSD seems to be having only 80 columns in console
windows (and with FreeBSD after a kernel compile, there was
90 columns).

Craig Carey