Re: Conlang Flag: Voting

Philippe Caquant wrote, quoting myself:

> > Anyone who is voting would have already looked up
> > the flags (at
> > conlangflag.htm) where they are fully explained.
>
> Ah, ok. Then I suggest a link inside the page: "More
> information about the symbolics ? Click here", getting
> to
> http://web.netyp.com/member/dragon/temp/conlangflag.htm
I'll link from the explanations page <conlangflag.htm> to the voting
page, but I really do think it's silly to do it the other way around.
The thing about symbols is that they stick in your mind. If a symbol
does *not* stick in your mind then that symbol has **FAILED**. So why
would you vote in favour of it? It's as simple as that.

Jan van Steenbergen wrote, sometimes quoting myself:

> I suppose the e-mail route is preferable after all... First, it is
> easier for you, second, the voting will be more or less anonymous
> (i.e. only you will know who voted what), third, multiple
> participation won't be easy.
Well, the web forms route has certain advantages. For one thing, it
gives me a chance to practice my PHP coding, which may lead to my
becoming incrementally more employable. For another, it means voting
can't begin until the form is complete, which guarantees people more
time for submitting flags. It's a little more work for me, but that
work is kind of fun, to be honest.

> > As Paul says, at real elections they record who has voted without
> > recording how they voted, but in this context there's a few
> > problems with that. Suppose someone *does* vote twice, but the
> > second time they enter a different name. AFAICS, you can't detect
> > that unless you have a list of people who are eligible to vote,
>
> Well, we dó have such a list, don't we? It's easy to get a list of
> those who are subscribed to Conlang.
I suppose that's possible if there are people among us who really feel
that absolute anonymity is important (i.e. people for whom such
anonymity is a prerequisite for their voting). Mind you, some people
are subscribed with more than one email address, but now we're getting
paranoid :-)

If we decided that preventing multiple votes was important but total
anonymity was not important, then what I'd do is simply get the code to
generate a text file like this:

    Vote registered by NAME1 at TIMESTAMP1: STRING-DESCRIBING-VOTE-1
    Vote registered by NAME2 at TIMESTAMP2: STRING-DESCRIBING-VOTE-2
    Vote registered by NAME3 at TIMESTAMP3: STRING-DESCRIBING-VOTE-3
    Vote registered by NAME4 at TIMESTAMP4: STRING-DESCRIBING-VOTE-4

Someone could read this if they knew or guessed the name of the text
file, but to the best of my knowledge they can't find the text file
without first knowing or guessing its name, so it's still fairly
anonymous (except that I'll know, of course). About as anonymous as
email. Less so if I call the text file something obvious like
"results.txt".

> What happens with flags who are very similar to each other? Like A-B,
> C-D, F-G, K-M, R-U? I would propose to work out the results per group
> instead of per individual flag. Like this (not looking at the flags):
No. If the flags are really so similar, people will put them next to
each other in their preferences anyway, so the voting algorithm will
take care of this automatically.

> Second question: what will be the practical use of this flag? I
> thought of that earlier, when you said that the thin gold lines in
> Jeffrey's Bable Tower were unpractical (a view with which I disagree,
> BTW). Do you intend to produce real flags, or will its use be
> primarily electronical?
The practical use will be whatever people think of. The option is
always open for someone to produce real flags if they wish.

> BTW, Roger's flags still show up as "Image hosted by Tripod".
Not a lot I can do about that just now. :-)

Paul Bennett wrote:

> In which case, you resort to the HTTP logs of the server to see at which
> time the bogus name was used, and which vote was cast during the same GET
> (or POST) operation, and discount both. If you like, you can get a good
> idea of which real person cast the bogus vote from the logs, too.
This is not something I know much about. Wouldn't I have to make a
special arrangement with my ISP to do this? And all for something that
probably won't happen anyway.

Jan suggests that I could obtain a list of subscribers and require
people to authenticate themselves by entering the email address they
use to post to Conlang. Probably that is indeed an option, if it is
deemed necessary and appropriate. (For my part, I had kind of hoped
that subscribers of other conlanging groups could be invited to take
part, so that the flag would belong to the entire conlanging
community, not just this list ... but that's just my opinion.)

Anyway, I'd like to hear more opinions about what degree of anonymity
is important, and how best to achieve it.

Adrian.

Reply