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

David Levy lifeisunfair at gmail.com
Tue Jun 8 21:46:23 UTC 2010

Aryeh Gregor wrote:

> It would be perfectly fine if objections to the change stemmed from
> judgment, provided the judgment was sound.  In some cases (not all),
> I don't think it was.  I provided arguments for why I thought the
> result of my own judgment was better.

And that's fine (as I previously noted).  But you didn't merely seek
to refute arguments; you also dismissed them due to "the absence of
further data."  You then introduced arguments that were every bit as
speculative in nature, if not more so.

> If Domas says something about the DB, then yes, his judgment
> overrules almost anyone else unless they have data or other strong
> evidence.  Not everyone's judgment counts the same.  (But in this
> case, the usability team should be justifying its judgment
> explicitly and at length, so that the community comes to respect its
> judgment.)

I agree, but I don't think that anyone is challenging their expertise.
 Rather, we're questioning the application thereof.  It appears that
they made two key mistakes:

1. They assigned undue weight to the interlanguage link click rate.
The statistics were gathered strictly from the English Wikipedia,
which is not necessarily representative of Wikipedias in general.  And
even if it is, a mathematical measure of the links' usage does not
convey their actual importance within the project.  (As others have
noted, the "edit" link also receives relatively little use, but
Wikipedia could not function without it.)

This is what led them to believe that the change would not be detrimental.

2. They applied a general design principle without considering
atypical circumstances that might render it inapplicable.

This is what led them to believe that the change would be beneficial.

In summary, the issue is not that the team lacks expertise; it's that
it was exercised in a robotic manner.

> I agree with the "be bold, revert, discuss" methodology that's being
> employed now.  However, I don't think that data in favor of a change
> should be an absolute requirement in the face of objections (even
> strong ones), unless there's data against the change.  (Which there
> isn't really here.)

I strongly disagree.

Firstly, much of the necessary data (such as usage statistics from
non-English Wikipedias) simply hasn't been gathered.

Secondly, it's been explained that the links serve a mission-critical
purpose not fully reflected in raw numbers.

Legally blind people compose a very small percentage of our users.
(The rate within the U.S. population is ~0.3%, so the rate among
Wikipedia users worldwide likely is similar or perhaps lower.)  So an
interface change theorized to "reduce clutter" for sighted users
(while creating a stumbling block for the tiny number of legally blind
users) could be considered mathematically justified.  But from a
social standpoint, it would be unacceptable.

By all accounts, the interlanguage links receive non-zero use, and
there is clear consensus within the community that the importance
thereof far exceeds that which can be gleaned from numbers alone.

Unless and until a net benefit (and not a purely mathematical one) is
demonstrated, no reduction in the feature's prominence should occur.

David Levy

More information about the wikimedia-l mailing list