[Foundation-l] hiding interlanguage links by default is a Bad Idea, part 2

[Gregory Maxwell]

On Fri, Jun 4, 2010 at 2:24 AM, Michael Peel <email at mikepeel.net> wrote:
>
> On 2 Jun 2010, at 22:51, Gregory Maxwell wrote:
>
>> A tiny benefit to a hundred
>> million people wouldn't justify making wikipedia very hard to use for
>> a hundred thousand
>
> Can you justify that the change has now made it very hard for users of those interlanguage links? Given that it's now one click away (click on 'languages' in the sidebar) the first time, and then it stays there afterwards (this menu does stay expanded after the first time it's opened, right?), I wouldn't have thought that would make it very hard.
>
> I would support it being expanded by default, though (even though I rarely use it myself) simply because it's a lot less intuitive to find the language links now, [snip]

I think you mostly answered your own question for the most part.   But
I think my statement was intended to be a more general statement about
comparing costs than really a statement that this makes wikipedia very
hard: OTOH,  if you don't read the language well and are depending on
the inter-language links to get you to the right article in the right
wikipedia, then the change did indeed make the site very hard to use.
This is the subject of Noein's car analogy.

I agree with the upthread comments on the roseate rectilinear
lego-hat.  It is as fertile a source of associations as any cloud
could hope to be, but "language" is not among them.

OTOH, I could make the same criticism for the watchlist star, which
has the additional sin of conflicting with the use of the star
iconography used for featured articles.

As far as the the dynamic hiding goes, I'd like to toss in my voice
against that:  Determinism is very important for usability.   Guessing
what the user wants is great when it works but terrible when it
doesn't.  Computers are often _stupid_ but at least they tend to be
consistent. The fact that you can learn to cope with their stupidity
without much effort is often their one redeeming quality.  Interface
choices should favour determinism except when the cost of doing so is
very high, the automatic mechanism is very very reliable, or the kind
of non-determinism is very harmless and non-confusing.

Anyone who has tried to get wolfram alpha to perform a specific
calculation and suffered through a half hour of swapping around your
word order knows of the frustration that can come from the computer
trying to be smart and failing.

In particular, that absence of a listing depends on an basically
non-deterministic guess of what you want  _AS WELL AS_ the article
simply not existing is likely to be confusing.  E.g. thinking an
article only has a german version when the german version is featured.

At the same time I think that changing the order, typeface, color, or
adding iconography based on automated smarts is far less likely to
result in confusion and is probably an OKAY thing to do.

On Wed, Jun 2, 2010 at 8:09 PM, Aryeh Gregor
<Simetrical+wikilist at gmail.com> wrote:
> . . . well, I can expand on this a bit.  Wikipedia's goals can be
> summarized as "Give people access to free knowledge".  This can be
> measured lots of different ways, of course.  But I see no reason why
> they shouldn't all scale more or less linearly in the number of people
> affected.
[snip]

Things like hiding inter-language links and switching to vector even
though it locks out browsers used by many people more or less
completely deny access to the site for people.  I think it's really
hard to justify effectively locking people out for the sake of the
soft benefits of a great number of people.

I'm not saying that there is a true hard incomparability. In general I
think that denying _one_ person the ability to effectively use the
site unless they understake a costly change in their client would
justified by a small improvement for the bulk of the users... but only
that it doesn't form a nice neat linear relationship where you can
directly trade readers to usability fluidness. ... and that, as you
described it, incomparability is a useful approximation much of the
time.  The approximation only really starts to fall down when you can
make a serious argument that there is a true like for like replacement
e.g. loss of life = actually saves two lives, as distinct from loss of
a life = makes 2000 people live 0.1% longer.


Sort of tangentially, ... am I really the only one that frequently
uses the Wikipedia inter-language links as a big translating
dictionary?  I've found it to be much more useful than automatic
translation engines for mathematical terms (both more comprehensive
but also in that it makes it easy to find the translations for many
related terms).  The hiding doesn't make this any harder for me, but
it would make me a lot less likely to discover this useful feature.