Re: Debating difficult development issues in essay form
Wouter Verhelst <firstname.lastname@example.org> writes:
> This is probably true. However, I'm not convinced your proposal solves
> more problems than it introduces:
> - First, I find it extremely difficult to follow a discussion on a wiki
> page. Yes, there is a diff feature in most wikis (including ours), but
> that requires you to remember when you last read the position on the
> wiki page in question; this makes it prone to losing out. In contrast,
> when I participate in a mailing lists discussion, I simply have new
> information marked as "unread" and old information as "read". That
> makes it much easier to figure out what's new and what isn't.
The goal is not to have that sort of discussion on the wiki. The goal,
indeed, is to have the wiki pages *avoid* that sort of discussion in favor
of more comprehensive statements of position. Frequently, I expect those
statements of position to converge on implementable proposals as the
For back and forth, while wiki comments are available and may be
convenient for some purposes, I expect that most of the real discussion
will continue to happen on debian-devel and similar fora. However, the
*results* of that discussion, as opposed to emerging nebulously from
back-and-forth posts and watching who stops talking first, but rarely
being stated outright, can be recorded in this format.
My hope is that someone who was interested in the outcome but not horribly
interested in the process would be able to skip the debate entirely and
just read the resulting statements and still have enough data to make an
informed decision. The debate will continue to be important for refining
> - In my experience, when discussing controversial subjects, it is a
> mistake to believe that the number of 'sides' in a discussion is
> significantly smaller than the number of participants to that
> discussion -- or indeed that it is even possible to distinguish which
> 'side' one is on. I've often experienced during such discussions that
> I may fully agree with someone else on one detail of the matter at
> hand, but vehemently disagree with that same person on another detail.
> With your proposal, this would probably mean we'd either need to write
> smaller "essays", one for each part of the matter at hand, so that
> people can sign off their own combination of details, or we'd need to
> write multiple mostly-but-not-quite similar essays. Both pretty much
> defeat the purpose of your proposal.
For most of these discussions, we have to, at the end of the process,
converge on a single decision. For example, we're only going to have one
release process. Therefore, while it's certainly true that each
participant starts as their own "side," and we need to provide room for
that to evolve and change, I would strongly encourage authors to trim out
the parts of their proposals that aren't reaching consensus and thereby
create proposals that have a broader base of support.
One of the advantages of the wiki pages is that they can record what parts
of the argument people find essential and keep them separate from the
inevitable digressions and debates about surrounding issues that, while
interesting, don't need to be taken into account when making a decision.
I think it's very difficult to tell, right now, what positions someone
holds about a topic after a 50-post debate in debian-devel and (more
importantly) which of those opinions they consider essential and which
they consider incidental.
> - Most importantly, if you write down an opinion that multiple people
> then sign off on, it becomes much harder to change or restructure your
> opinion as a result of the debate. If you're discussing something in a
> mailinglist, it's okay to say "yes, you're right, you've convinced
> me", even if some people may (wrongly) see that as losing face. Once
> you've done that, people will understand that your opinion is no
> longer what it once was. If you've got an essay form of your opinion,
> should you then rewrite that? But what about the people who (used to)
> agree with you? Should they agree with the rewritten opinion, too?
> Probably not. But you can't sign off on it anymore. Should you then
> write a new version of that essay? That brings us back to the 2nd
> problem I pointed out.
In this case, I would check with the co-authors and see if they agree, or
if you can reach agreement. If not, indeed, I'd remove my name (while
leaving the document as-is) and either write a new document or indicate
support of a different document, or possibly just bow out entirely,
depending on the situation.
This isn't a voting system; you don't have to sign any document. The
point is to allow for co-maintainers to spread the work, not to have the
signatories be an indication of support. Support will be determined by
project consensus, not by counting co-sponsors.
> - Even if that wasn't true, after having put a lot of effort in an
> essay, I think many people will become entrenched in that opinion. As
> a result, they may be less likely to consider opposing arguments and
> change or restructure their opinion based on these arguments. This
> would result in less discussion, and more flames.
This is certainly a risk. I don't know what outcome to expect. My
personal guess, though, is the opposite of yours. Based on my personal
experience with writing up position statements, I think that the
back-and-forth discussion is much more likely to create emotional
entrenchment in a position than writing a document, even if the latter is
more work. At least for me, what creates that sense of entrenchment is
much less about the amount of work I put into writing something and much
more about social issues, such as whether I'm feeling threatened or
attacked by someone.
This system is designed to give people space they control to develop their
position without feeling under seige, the way that one can feel on a
mailing list, and I think that may improve people's willingness to change
their minds. I could be wrong.
Russ Allbery (email@example.com) <http://www.eyrie.org/~eagle/>