Message-ID: <20200313031947.GC225435@mit.edu>
Date:   Thu, 12 Mar 2020 23:19:47 -0400
From:   "Theodore Y. Ts'o" <tytso@....edu>
To:     "Bird, Tim" <Tim.Bird@...y.com>
Cc:     Laurent Pinchart <laurent.pinchart@...asonboard.com>,
        Laura Abbott <laura@...bott.name>,
        "ksummit-discuss@...ts.linuxfoundation.org" 
        <ksummit-discuss@...ts.linuxfoundation.org>,
        "tech-board-discuss@...ts.linuxfoundation.org" 
        <tech-board-discuss@...ts.linuxfoundation.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [Ksummit-discuss] Linux Foundation Technical Advisory Board
 Elections -- Change to charter

On Thu, Mar 12, 2020 at 09:28:09PM +0000, Bird, Tim wrote:
> I agree with Laurent.  I'm not sure how to solve this problem, but
> I think you need something to indicate the voter approval policy
> besides "the TAB will decide it, and can change it when they like".
> 
> I suppose the pool of voters has been decided historically by the Kernel
> Summit invitation committee.  Some randomness was introduced by
> allowing voting by attendees from whatever event the Linux Foundation
> co-located with the Kernel Summit.  I think in practical terms,
> this means that recently the voting pool was self-selected (somewhat), but
> was skewed towards people who could travel, or had employer support.
> But in any event, the selection of the voting pool was done by people outside
> the TAB (or at least not necessarily inside the TAB), and without any eye towards
> skewing the election results.  That is, I don't think the kernel summit invitation
> committee, or the LF event staff, ever considered TAB voting in their KS attendee
> selection or event pairing choices.

(Speaking personally for myself)

The choice to include whatever LF event the Kernel Summit was
colocated with was a choice that was made by the TAB on an ad-hoc
basis.  There is nothing about that in the TAB charter at all.  So
we've *already* been doing things in a way that is not consistent with
TAB charter --- for years and years.

Starting last year, we experimented with electronic voting.  We didn't
change the composition of who could vote (it was anyone attending
Plumbers), but one of the reasons was that we didn't want to change
two variables at once.

We haven't made any final decisions yet about how the pool of voters
might be expanded.  But it might include (for example) people who have
user accounts on kernel.org.  Or historically, one of the pools from
which the kernel summitte attendee list would be drawn included
everyone who at least N Signed-off-by:, Reviewed-by:, etc.  Since
there was a Kernel Summit program committee filtering who got invites
(and there was non-trivial overlap between the program committee and
the TAB), in theory that very much could influence the TAB elections.
Expanding the pool to those who were interested and who were attending
the colocated event very much decreased that effect, but there has
always been a human filtering element.

One of the potential failures that having a human filtering element
prevents is the Sad/Rabid Puppies scenario:

https://slate.com/culture/2016/04/sad-and-rabid-puppies-are-trying-to-game-the-hugo-award-shortlists-again-in-2016.html

So for example, if we set some rule like, a single Signed-off-by: is
enough to get a vote, what would happen over a few months, a thousand
spelling/whitespace "trivial" patches, all from different sock puppet
accounts, show up and get absorbed into the tree?  With a human
program committee, it's easy for humans to say, "Ha, ha.  No."  But if
we use a mechnical rule, and badly chosen criteria, it might be really
easy for some mischief makers to carry out a Sad Puppies style attack
on the voting system.

On the flip side, having humans deciding who can and can't vote has
other really bad effects regarding the election's legitimacy.  It
worked 5+ years ago, because it was simpler times, and the formal
reason for the selection was attendance to a closed technical meeting,
and we later decided to hang the TAB elections off of it.

So that means we need to be smart about how we pick the criteria.
Using a kernel.org account might be a good approach, since it would be
a lot harder for a huge number of sock puppet accounts to meet that
criteria.  We don't have a final proposal for something which can be
objectively measured, but can't be easily gamed by someone who is
trying to subvert the system.  It is pretty clear, though, that we
need to have that clearly articulated, in writing, *before* we start
the nomination for the next round of TAB candidates.

I will also point out that we may not have much of a choice about
switching to something besides "people who attend the colocated event
where the Kernel Summit is held".  The program/organizing committees
for LPC, KS, and MS, are continuing to make plans in the hope that the
COVID-19 pandemic will have subsided enough that it will be safe to
hold an conference of 400-500 people in Halifax.  However, nobody
knows if that is the case.  If you look at this article from the
Lanclet medical journal article, "How will country-based mitigation
measures influence the course of the COVID-19 epidemic?":

https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(20)30567-5/fulltext

The takeaway is it's really not clear how long it will take for the
COVID-19 pandemic to run its course --- but some of their sample
curves extend out for of 5-6 months or longer.  So while we are
continuing to plan that LPC will take place, it's only responsible to
consider what we should do if in fact health and safety restrictions
are such that we might not be able to hold *any* Linux systems
conferences in 2020.

In that case, we might be forced to either keep TAB members in place
beyond their original elected term, or we might have to go to a pure
electronic voting for the upcoming TAB election.  I very much hope
that won't be the case, but we need to be prepared for that
eventuality.

(I'll keep silent about what I think of the current US
Administration's competence in steering us through this crisis, except
to note that in South Korea, they are testing 10,000 people a day,
using drive-through centers, and we haven't been able to test that
many *total* so far in the US, and there are Biogen employees in
Boston and Life Care Center employees in Kirkland, Washington, who are
still waiting for COVID-19 test availability....)

						- Ted

Powered by blists - more mailing lists