[erlang-questions] FOP (was: Re: Trace-Driven Development)

[Michael Turner]

On Tue, Jun 12, 2012 at 7:02 AM, Richard O'Keefe <ok@REDACTED> wrote:

>> Again, you're falling back on "wikis suck because I found something
>> bad."
>
> I have never made any claim that general.
> I say that *documentation* wikis are in my experience bad;
> far far worse than the existing Erlang documentation.

Again: if all the existing Erlang/OTP documentation were suddenly in a
wiki, word for word, would it be "far, far worse" than the existing
documentation?

That, and the quality of the ensuing administration, are the only two
things that matter for this discussion. NOT that somebody added some
incorrect information about you to your Wikipedia biography.

> I've actually pointed to Wikipedia as unusually *good* in my
> experience.  I may be wrong about that:

It's incredibly bad in some places, and very good in others. That,
again, is a function of an administrative choice: Wikipedia can be
edited by anybody, even anonymously. High-interest articles tend to
have a lot of watchers to help maintain quality. Low-interest articles
can have crap in them that nobody notices for months.

> http://catless.ncl.ac.uk/Risks/26.79.html#subj6.1
>
>        60% of Wikipedia entries about companies contain errors
>        When respondents attempted to engage editors through
>        Wikipedia's "Talk" pages to request factual corrections
>        to entries, 40 percent said it took "days" to receive a
>        response, 12 percent indicated "weeks," while 24 percent
>        never received any type of response. ...

This is ridiculous. If you think a factual correction is required to a
Wikipedia article, DO IT YOURSELF. If you have the editing skills
required to alter a Talk page, you probably know enough to make the
change you think is necessary in the article itself.

Afraid to touch an article? Then don't get involved in the first
place. One of the Five Pillars of Wikipedia is "Be bold."

   http://en.wikipedia.org/wiki/Wikipedia:Be_bold

If you approach Wikipedia in something other than the spirit Wikipedia
itself strongly encourages, you shouldn't complain about the results.
Editors with a strong interest in articles set up watchlists mainly
*for* the articles, not so much for the article Talk pages. You want
attention to a problem in an article? Apply your own attention first:
make what you think is a fix. Who knows? You might actually be right!

>  ..... Of those who were
>        familiar with the process of editing Wikipedia entries,
>        23 percent said making changes was "near impossible."
>        Twenty-nine percent said their interactions with Wikipedia
>        editors were "never productive."

This is also a ridiculous criticism of Wikipedia when, if anything,
it's to Wikipedia's credit. Go look at Articles for Deletion

   http://en.wikipedia.org/wiki/Wikipedia:Articles_for_deletion

And that's just just the tip of the iceberg of the generalized
egomania out there, an obstacle that Wikipedia is always trying to get
past while staying as open as possible. A dismayingly high proportion
of those who get "familiar with the process of editing Wikipedia
entries" are people who familiarized themselves in order insert
self-serving information of one kind or another, including entire
articles about themselves, their non-notable music careers, their
non-notable companies, their non-notable self-published novels. Even
those who actually DO clear Wikipedia's bar of notability (and their
fans) will insist on trying to add information to the resulting
articles that doesn't qualify as properly sourced or written from a
neutral point of view.

   http://en.wikipedia.org/wiki/Wikipedia:Neutral_point_of_view

And how likely is it that a survey of *PR* "professionals" (corporate
flacks) who tried to edit Wikipedia would have tried to do it without
violating

   http://en.wikipedia.org/wiki/Wikipedia:Neutral_point_of_view

and guidelines about Conflict of Interest?

  http://en.wikipedia.org/wiki/Wikipedia:Conflict_of_interest

This is a function of Wikipedia's access control, which is pegged at
the most liberal end of the spectrum. It's also possible to set up a
MediaWiki-based wiki that permits nobody except a tiny coterie of
administrators to contribute edits. And there is a range of
possibilities in between, including customized access control
algorithms for those who know PHP/MySQL. (Yeah, yeah, I know: that
choice of technologies is a bug in itself. Whatever.)

> The title of the RISKS article is a little misleading:
> out of the people surveyed, 60% of those whose companies had a Wikipedia
> entry, who knew it had, and who had checked it, had found factual errors
> worthy of correction.

Wikipedia has a number of areas where it's weak. In my experience,
except for the more prominent companies, business is one of them. Bad
information about companies is one syndrome in that topic area.
Another much more serious one, however, owes to the PR people surveyed
in the paper you cite: companies using Wikipedia for self-promotion.

None of which is relevant to this discussion -- unless perhaps
Ericsson wants to claim that it invented Lamport clocks, without any
supporting independent references for the statement.

No, we're talking about what's *theoretically* possible in putting the
Erlang/OTP documentation into a wiki, under access control policies
that would presumably be acceptable under the terms of licensing by
Ericsson, and that would favor contributions of reasonable quality.
I've already suggested one strategy for access control, and I've
repeated it several times since. Find it and tell me what's wrong with
it, if you want to criticize it.

> ... 21% of the people who reported errors [in business articles] had found
> spelling errors, which gives _me_ pause for thought about the utility
> of a Wiki as a device for correcting spelling.

That's probably because when you can see from the first paragraph that
the Wikipedia article is yet another puff-piece that got in there
somehow, you might as well just go to the company website. (And let
your eyes glaze over reading that.)

When people search on a general topic ("bees", "galaxies"), the
Wikipedia page often head the rankings. And the #1 result gets most of
the attention. When people search on a company, they usually they get
the company website first. Since so many Wikipedia articles about
companies are transparently nothing but plagiarized PR (often
self-plagiarized) from company websites, I'm not surprised that they
get so little copyediting attention, for any added information beyond
the PR.

> The paper has some sobering figures about corrections being undone.
>
> "Half of respondents who have either directly edited or used the talk
>  pages to make edits indicated that they believe the process of making
>  changes to a company or client’s Wikipedia article is typically time
>  consuming (n=275), but 27% stated that it was easy (n=152), while
>  23% said it was near impossible (n=127)."

"Sobering"? I'm laughing like a hyena.

I'm pretty sure that the overwhelming majority of those frustrated
23%-29% were trying to do something Wikipedia actually rules against:
stating "facts" without reliable-source backup.

  http://en.wikipedia.org/wiki/Wikipedia:Identifying_reliable_sources

And that's not even getting into "neutral point of view" and "conflict
of interest" issues.

Let's say that the Wikipedia article on Erlang/OTP were to say, on
nothing more than the strength of belief on the part of some Ericsson
veterans or current employees, that Lamport clocks were invented as
part of tracing, in the late 70s, and for the AXE series of switches.
I.e., that there was no citation to anything qualifying under
Wikipedia's Reliable Source rules. In that case, the statement should
be deleted. Anybody who insisted putting it back into the article
based only on their own personal opinion could, if they were
sufficiently insistent, end up being banned from editing Wikipedia.
Would they report that they were "frustrated"? Probably. SHOULD they
have been "frustrated"? Probably.

Even when something is true, it's not enough. In general, it has to be
reported elsewhere, independently. I did some edits on

   http://en.wikipedia.org/wiki/Brad_Cox

and contacted him about getting some reliable sources for certain
points in his resume (since he always seemed rather unsung to me.) He
asked, "What do you do when you yourself are the only source?" I
replied saying that I didn't doubt him personally, but that the rules
existed for good reason: the incredible level of dishonesty and
egregious self-promotion that happens on Wikipedia if you don't have
such rules.

>> You're not asking, "Why does this bad thing happen, and is there
>> anything to be done to prevent it?"
>
> No, and that's because I'm the wrong person to do so.
> That's *your* role in this debate.

I don't need to ask "Why does this bad thing happen?" Bad? I've
already seen it all. Fought it all.

>> In fact, these problems are quite
>> manageable, and Wikipedia, in the interests of gathering the largest
>> possible membership, leaves editing quite open. It also means some
>> information is unreliable at one time or another. Anyone who reads
>> Wikipedia without keeping that in mind is simply a fool.
>
> The thing is, the more you talk about "management", the less a Wiki
> sounds substantively different from the system we have right now.
> A restricted pool of editors?  We have that now

FINALLY you acknowledge a point I've made repeatedly: most of your
concerns about bad information getting into a wiki can be addressed by
MediaWiki access control, which Wikipedia only uses on a case-by-case
basis.

> the main difference
> is that you expect to become one of them.

Actually, quite the contrary, at least on the specific point of
getting the seq_trace documentation to mention that it implements
Lamport clocks if only I submit a patch. The internal corporate
self-mythologizing on this point appears to be quite strong, if the
views of two former employees are any indication.

> Low barrier to entry? We have that now.

Eye of the beholder again.

In the Erlang/OTP documentation I find errors in grammar, punctuation
and spelling that never seem to get fixed. There are points in the
text where a link would be enormously helpful, but those links never
seem to get added. If the barrier to entry for fixes is so "low", as
you claim, why weren't these things fixed long ago?

> Nothing stops you setting up an Erlang documentation Wiki right now.

Unless if violates the Ericsson license terms.

And then there's this: the discouraging impression that highly
intelligent people can nevertheless reject ideas for reasons stemming
almost entirely from their own ignorance and prejudices, and who will,
having taken such a stand, go cherry-picking all subsequent evidence
they present, from of irrelevant contexts if necessary, and who will
quickly change the subject to one those picked cherries when shown
evidence that they are wrong on a given point.

Case in point: you.

-michael turner