>>>>> "Barak" == Barak A Pearlmutter <ba...@pearlmutter.net> writes:
Barak> On Wed, 21 Apr 2021 at 16:57, Sam Hartman <hartm...@debian.org>
wrote:
>> That's a big jump, and I don't think I agree. At least not when
>> you phrase it that way. Why should my preference matter less
>> just because it's weaker? It's still my preference and I'm
>> attached to it very much:-)
Barak> There are two ways to approach this kind of question.
Barak> First: we can use our intuitions. What makes sense? And we
Barak> can discuss that, explain why some things seem intuitively
Barak> fair and others don't. We can make analogies. How does a
Barak> group decide on a restaurant? What do we think is fair? What
Barak> doesn't seem fair?
Barak> Second: we can get scientific about it. This means we define
Barak> some performance metrics for voting systems, then measure
Barak> their performance. Such measurements can be done
Barak> theoretically, or in simulation, or in practice. It can use
Barak> various assumptions about the environment, and even various
Barak> performance metrics. This might include difficulty of filling
Barak> out a ballot, or understandability of both the ballots and
Barak> the system as a whole, as part of the performance. When we
Barak> try it on real people, factors like what fraction of the
Barak> eligible voters actually bother to vote, or what fraction of
Barak> them can correctly answer questions about how the system
Barak> works, might be things to measure.
Thanks for bringing this up. This was the one part of the conversation
I didn't get to touch on in my last message, and I think it's the last
lose loose end of the conversation.
I actually find that the scientific part of the conversation is not
helpful to me in determining what the requirements should be--what the
desirable properties are.
As an example, I am finding this conversation very difficult to follow,
because you are always phrasing things in terms of alpha, beta, and some
generic options.
I appreciate given the past few weeks why you're doing that--the
discussions have been charged.
While I find that our track record with intuition is bad, it's very easy
to get into the mathematical land with bad requirements and have high
confidence in a system that doesn't meet our needs.
Let's take an example of something you brought up early on: in our
voting system, the outcome can change when a ballot option is added.
If I think about that in terms of generics, it sounds like a horrible
property.
But as I start to think about that with specific analogies, I realize
that it's actually related to situations that come up in consensus
decision making.
And when I restate it as something like the following, it's much less
clear that it is an undesirable property.
Sometimes new information from the voters can influence the outcome of
the decision making process. If we end up asking about an outcome that
fits into the middle of a cycle between other outcomes, we have more
information, and this information can change the result.
And once you look at things that way, it becomes a lot less clear
to me at least whether that's an undesirable property.
And no, thinking about strategic abuses isn't going to help much. Are
those really strategic abuses, or are we saying that people who can
introduce options that allow us to better determine the voter
preferences can successfully influence the election?
That is, is it strategic abuse, or strategic examination of the voters
desires?
The math has its place, and may even help us think about that question,
but it's not going to answer it for us.
The math certainly helps. We can easily see that even if we think that
kind of strategic exploration is not an abuse, it clearly would be an
abuse if some privileged category of people got to choose the ballot
options.
So, for me, I've been finding participating in this discussion difficult
because of the mathematical emphasis. It's not that I can't follow the
math. It's that divorced from the analogies, I cannot reason about
whether our initial requirements are any good nor reason about trade
offs between them.
--Sam
signature.asc
Description: PGP signature
