Google Goes After Wikipedia

On 21/12/2007 1:22 AM, "Johann Beda" <st-tidbits.combeda.ca> spake thus:

> Fortunately for most of the science and technology (and popular
> culture too I suppose) there is less problem with subjective interpretation
> of events that things in the history areas struggle with.

I've been told by a mathematician/physics friend that most of the Wikipedia
articles in these fields are quite good. Certainly the ones I've looked at
were very good. Again, not surprising given the fairly black and white
nature of the subjects for things that aren't bleeding edge.

--
Nigel Stanger, Dunedin, NEW ZEALAND.
http://xri.net/=nigel.stanger

While directly citing Wikipedia is probably inappropriate in an
academic setting, the wiki can be a fantastic jumping-off point,
especially for well cited articles. (And the standards of Wikipedia do
encourage citations for all asserted facts.)

At 04:27 12/20/07 -0800, Lukas Mathis wrote:
>Click on "history" and you'll see exactly who has
>changed what at which point in time. [...]
>I see that you yourself have researched some edits on Wikipedia and
>found out who made them, so I'm not entirely sure why you'd say that
>there's no clear mark of authorship.

"Clear" and "possible to track down some internally consistent source
reference" are not the same thing. It takes significant technical
capabilities to track down "who" did what to an article. Geez, to even
discuss an article you have to learn an arcane markup language -- not even
one that resembles most god-awful web forums. Just the length of the phrase
I had to use to describe what passes for clear in the Wikipedia world
should give pause. One accustomed to reading in the wider world quickly
gets the sense that Wikipedia contributors are hiding something.

And when you're done, you still in most cases have no idea of the identity
of the "who" that you're looking at. Look at my edits and you see
"paleolith". This means nothing to people in the Burroughs/Unisys world
closely related to the ALGOL article I linked to. My own name does mean
something to a lot of those people. I am not my Wikipedia username.

So the "identity" of Wikipedia contributors is a navel-gazing exercise. You
know that this contributor to awikipedia also contributed to bwikipedia
and to cwikipedia, but between that and the rest of the world is a vast
chasm. While the rest of the world is embracing a network of human
knowledge-bearers, Wikipedia is curling into isolation.

Edward
--
Art works by Melynda Reid: http://paleo.org

At 1:48 PM -0800 12/17/07, Tomoharu Nishino wrote:
>I think this is partly a result of the lack of positive incentives to
>participate, as Klaus describes. But I would go further. There are
>significant disincentives for real experts to participate. After all,
>what expert would be willing to spend the time and effort to craft a
>well thought out and articulated article on a particular topic if he
>knew that the moment he posted it, it would be edited, altered,
>contents added or subtracted to suit a particular perspective, and
>perhaps mangled beyond recognition by people who regard themselves
>rightly or not as his equals? To those who have actual expertise and
>care about the *integrity* of their work, I suspect that what goes on
>in Wikipedia would be intolerable.

What confuses me about the approach of experts to wikipedia is that
they expect that their word should be taken as valid because they are
experts on the subject. An expert who cites well and explains
lucidly should have their work hold up. Sure it'll be tweaked and
altered, but if it is well cited it should stand.

~Nick
http://www.inmff.net

At 6:25 AM -0800 12/18/07, Edward Reid wrote:
>See http://en.wikipedia.org/wiki/ALGOL. Unlike the chaparral article, this
>is one that I know enough about to contribute to. There's a paragraph which
>appears to be entirely incorrect, but I'd still prefer to ask the original
>author where the information came from before replacing it. But who wrote
>it? How do I contact them to ask? It appears that they are not watching the
>low-activity discussion page, as the request I posted has gone without
>response for almost a week now. Perhaps there's an easy way to discover who
>originally added text, but the only way I've found is by binary search on
>the edit history, and this one has had a few hundred edits. I think I
>finally figured out who added it, and it's one of those people with a huge
>edit count (of necessity mostly things they haven't carefully researched),
>Danakil, who has now disappeared from Wikipedia. I did locate one comment
>disputing the paragraph in question -- but it was on Danakil's discussion
>page, not the ALGOL discussion page. There's no proper sig link from that
>posting. I think I know who did it, but that person is also preternaturally
>active on Wikipedia. At this point, I have to assume that all of the twenty
>or so edits which Danakil made to this one article are all suspect. Anyway,
>the point of all this is that just figuring out why something is said,
>before even attempting to correct it, can be extremely time-consuming. This
>leads to most edits being done by people who don't care about the
>background, either because the edit has little or no affect on the content
>or because they simply aren't contemplative enough to consider the
>background. I hope I don't have to marshal arguments that this is a Bad Thing.

This is more a matter of an academic disposition versus a wikipedian
disposition. The academic looks to engage in discussion. The
wikipedian fixes it first, then engages in discussion. Admittedly
this causes problems, but an academic can easily edit these by citing
their own peer reviewed work. Wikipedia builds upon peer reviewed
work, it doesn't displace it. Anything and everything on wikipedia
that is not prima facia valid on wikipedia is fair game to delete.
Wikipedia is not a sacred document, instead it is just the opposite,
a working draft, that just happens to be useful at the moment.

I honestly think Google Knols is too little too late and won't really
work. But I also said that wikipedia would never work, and I tend to
think of it as generally successful.

~Nick
http://www.inmff.net

At 4:27 AM -0800 12/20/07, Lukas Mathis wrote:
>edward claimed:
>>>there is no systematic lack of credit to authors on the Web in general,
>>>with nearly every article, blog post, comment, and home page providing a
>>>clear mark of authorship.
>>has an obvious and major exception: Wikipedia.
>
>This is the impression one gets from reading the article. However, it
>is not the case. Click on "history" and you'll see exactly who has
>changed what at which point in time. If the user is not logged in to
>Wikipedia, his IP address will be listed.
>
>It's always clear which changes were authored by whom on Wikipedia;

I'm a huge wikipedia addict and editor, but I think the history
features leave features to be desired. The biggest problem is trying
to figure out who actually put the initial assertion into the
article. Many edits are of a housekeeping nature, cleaning up
spelling, punctuation, formatting, etc. (Indeed this is probably a
majority of my edits.) What is needed a tool that says "show me all
of the editors of this block of text"

>so I'm not entirely sure why you'd say that there's no clear mark of
>authorship.

I think from an academic perspective it is more a matter of
pseudonyms vs. actual names. I can understand this, but there is a
matter of just mapping them appropriately. For example almost
everything at http:///www.inmff.net was written by me. (There are
self-documenting exceptions, that are clear) I then use this to
document ownership of my other accounts. (e.g. wikipedia.) But how do
you know that someone with the username of a famous academic on
wikipedia is really that academic? Herein lies a problem, not a hard
one, but one none the less.

~Nick
http://www.inmff.net

On Dec 17, 2007, at 1:48 PM, "Adam C. Engst" <acetidbits.com> wrote:

At 5:19 AM -0800 12/16/07, Michael Fleming wrote: > Google's knol concept seems to me to be very similar in some ways to > today's conventional publishing world.

In some ways, except that the conventional publishing world attempts to vet authors and ideas to avoid publishing work that's inaccurate, wildly biased, or simply uninteresting. Google won't be doing that at all, instead relying on PageRank to separate the wheat from the chaff.

Adam's post leads one to conclude Google is on the wrong track. Wikipedia has a great collection of knowledge and sticks to that. G will lose its way in diversity as with so many other Behemoth companies.

Dave

On 1/1/08 at 8:22 AM, shawnyourmaclifeshow.com (Shawn King) wrote:

>On 12/20/07 7:22 AM, "Ken Hallenius" <halleniuup.edu> wrote:

>>It is interesting to me (and I too am guilty!) that people who are
>>distracted by inaccuracies in articles where they have some
>>expertise will nonetheless read and "trust" Wikipedia articles on
>>other new-to-themselves topics as a reference to familiarize
>>themselves with that topic...

>Agree completely. I "rant" about this on a regular basis. 2
>examples:

>1) Consumer Reports being cited. I've read their computer reviews
>and they have historically been poorly done regarding Macs. But My
>wife wants to "trust" them regarding other things. If they got the
>computer stuff wrong, what else have they gotten wrong?

Since Consumer Reports, Wikipedia and any other source of
information are human creations, there will be some level of
error present. The real question isn't what else they have wrong
since it is virtually certain there are additional errors. The
real question is what fraction of there information is wrong and
how severe the errors are. Looked at from this perspective, it
is quite reasonable to regard a source of information as being
accurate in some areas while having known errors in other areas.

I am actually quite shock that Wikipedia works as well as it does. I'd
never use it as an authoritative source, but I have found the
references an excellent place to start. Plus, there are some topics
that only Wikipedia covers. For example, the Current TV show
Infomania. There's not only a Wiki article on that, but on the various
items in the introduction to the show.

And, from what I can see, Wikipedia can be fairly accurate. The secret
is that it is not a true Wiki. There's a hierarchy of people who
believe in Wikipedia, and do their best to make sure people quote
sources, etc.

The big problem is that Wikipedia has no authoritative voice. Anyone
can contribute to any article, and you simply don't know who is behind
the article. Yes, you can trace down the discussion which *might* give
you some information, but then who is wiiportal and is he (or she) an
expert in the Ming dynasty?

I think Google might have something although I am not sure whether
"The Market" may be the best way to determine expertise. At least I'll
be able to see who is behind a particular article, and do a cross
check on them.

Will it work? If Google appoints someone to see it off the ground and
that person finds people to help write articles and convinces them to
write these articles, I think it will go somewhere. If Google merely
sets up the infrastructure and says "That's that!", it will fail.

--
David Weintraub
qazwartgmail.com

It's been interesting reading all the criticism of Wikipedia, much of
which is undoubtedly deserved. However, I think a few important
points are being missed:

1) Google Knol is no more an "academic" project than Wikipedia.
Anyone can create a knol, and they'll be created primarily on general
subjects (purely from the reason that most subjects are general). So
the criticisms of Wikipedia's accuracy from an academic standpoint
will likely apply to Google Knol as well, albeit without the
capability to change things that are wrong. That will instead, it
sounds, be replaced with the possibility of appealing to the author,
who may or may not acknowledge the comment. And authors will lose
interest, change fields, and die, raising the question of what
happens to inaccurate but popular knols that have no one even
available to make changes.

2) Although I haven't heard if there has been an update, the last
time accuracy was compared, Wikipedia and the Encyclopedia Britannica
were comparable. That said, I believe Wikipedia will always have more
errors, simply because it's accepting a vast amount more information
(and even types of information) than any encyclopedia. Personally, I
believe the accuracy of knols will be at best comparable to
Wikipedia, if not notably lower when taken as a whole, due the
fallibility of individuals.

3) For the most part, I'd argue that the errors in Wikipedia, such
that they are, aren't a particularly big deal. When I go to Wikipedia
to read about, say, Andrew Jackson, 7th president of the United
States, I don't much care about the ultimate accuracy of specific
details because I'm just there to get a broad overview of the man's
life. That's true of nearly everything I look up in Wikipedia,
because if I care deeply about the accuracy of something, I'll want
multiple different sources for the information.

4) Wikipedia is meant to be an encyclopedia, and as such, it's
entirely reasonable that it shouldn't be considered a primary source
for academic research. No encyclopedia should be - they're all
secondary sources. The question is if Google knols would be
considered acceptable sources, and the answer to that is the same as
the answer to the question of whether or not any random Web page
published by an individual is considered an acceptable source.

5) I'll completely agree that Wikipedia is essentially anonymous,
despite the history of edits. But my point in the article is that
it's the exception, not the rule that the Google guy was claiming.
The vast majority of content on the Internet provides clear
authorship, so the fact that Google is raising this as an issue
merely shows that Google Knol is intended as a direct assault on
Wikipedia, rather than something that's solving a general need.

cheers... -Adam

At 08:28 01/01/08 -0800, Nicholas Barnard wrote:
>The academic looks to engage in discussion. The wikipedian fixes it
>first, then engages in discussion.

Well, sometimes -- you assume that all are willing to discuss their
changes. And you also omit the middle ground: those who seek knowledge and
improvement *via* discussion. More on this below.

>Admittedly this causes problems, but an academic can easily edit these by
>citing their own peer reviewed work.

Not always, not by a long shot. One of the cases I cited was simply an
erroneous statement, not a negation of a significant or meaningful true
statement. It said "A is derived from B". The truth is that "A is not
derived from B". The second, true statement is not worthy of inclusion in a
Wikipedia article. Furthermore, there is seldom any citation for such a
negative. To say that one can "easily cite" ignores this "negative
citation" problem.

Could I have been wrong about this statement being erroneous? Yes, of
course. That's why I sought to engage in discussion FIRST rather than going
on the offensive and changing it and waiting for feedback -- which might
not happen even then given that many writers don't watch articles to which
they have contributed. I might have learned something, unlikely though it
seemed in this case.

(In fact, I determined that the contributor of the erroneous information
was a highly active contributor who has since disappeared from Wikipedia
completely. I rewrote the paragraph to eliminate the erroneous information,
and have received no comments in response. It would have been a lot easier
if I hadn't had to spend so much time tracking down the non-source.)

Edward
--
Art works by Melynda Reid: http://paleo.org

Chris Page wrote:
> On Dec 15, 2007, at 03:51 AM, Klaus Beckmann wrote:
>
>> Witness Wikipedia's quality: while I admittedly use it every day as
>> a way to learn about things of general interest or to read about
>> history (where errors are of minor consequence to me), a large
>> number of articles on game theory, economics and other subjects I
>> can claim some expertise on have left me breathless on account of
>> their flaws. The community processes reputedly at work have clearly
>> failed to ensure sufficient quality.
>
>
> Become a part of the community and fix them. The brilliance of
> Wikipedia is that you can simply edit the pages to improve them. If
> you notice a glaring error, try to take a few minutes once in a while
> to fix some of them.

As others have pointed out, it is not "take a few minutes".

Am I the only one here with a "life". Wife, kids, school activities, etc... other than playing with computers? Marching band, drill team, baseball, etc... take time. Way more than just to be at the games. (Somewhat rhetorical as I know Adam has at least a son who occupies a non-trivial portion of his time.)

Plus the entire discussion which followed about the difficulties of editing an article mirror the process for the Mozilla foundation and posting bugs for Firefox, Thunderbird, SeaMonkey, etc... Create a "bug" id, figure out the bug system, figure out the abbreviations, figure out the classification system, get the attention of someone important, etc... And at some point someone regularly tells the crowd to "read the code and post a fix". Yeah, right.

Community things can work, but at times they are a PITA.

David Ross

>The big problem is that Wikipedia has no authoritative voice. Anyone
>can contribute to any article, and you simply don't know who is behind
>the article. Yes, you can trace down the discussion which *might* give
>you some information, but then who is wiiportal and is he (or she) an
>expert in the Ming dynasty?

I don't think this matters to hardly anyone outside academia and
investigative journalism, where people are used to existing in
relatively small communities of experts and don't mind tracking down
someone and reading their CV to determine their likely level of
expertise.

For the vast majority of people, I'm willing to bet that "wiiportal"
is equally as useful/useless as "John Samuels." Many people have
common names, and few would do the sophisticated searching necessary
to track down the right John Samuels and determine whether he's
likely to be an expert on the Ming dynasty.

What's actually happening is that people are taking a shortcut and
assuming that if something is in Wikipedia that a lot of people have
seen it and (hopefully) cleaned up errors. It's similar to TidBITS -
I have to make a decision as to whether an outside author has the
necessary expertise, but once I've done that, everyone reading
TidBITS uses a similar shortcut to assume that my decision means that
the author is indeed a sufficient expert.

cheers... -Adam

On Tue, 1 Jan 2008, Shawn King wrote:

> 1) Consumer Reports being cited. I've read their computer reviews and
> they have historically been poorly done regarding Macs. But My wife
> wants to "trust" them regarding other things. If they got the computer
> stuff wrong, what else have they gotten wrong?

True, but what I like about them is that, for the most part, you can see
what their biases are. Their selection criteria are by necessity kind of
"lowest common denominator", which forces them to make occasionally
strange evaluations, but if you can factor that in when parsing their
reviews, you're free to agree or disagree with those criteria.

So, for example, if you're looking at digital camera reviews, and they
penalize all the ones that lack a certain feature you don't care about,
then you can still get decent info from the other things they look at.
Likewise with Mac reviews -- if they're faulted for, say, not coming
with built-in antivirus software, you can take that with the appropriate
grain of salt and pay attention to the rest. You just have to go in to
the review with a rough sense of what you're already looking for, with
an eye towards which products will end up being the best fit for you.

My biggest problem isn't so much their bias -- as everyone has one and
at least they try to be honest, open, and quantifiable about theirs --
as the fact that a lot of the time, their reviews date quickly, and it's
common for the reviewed product to have been superceded by a new version
by the time you're in the market to buy.
 
> If they get so much stuff wrong I know about, how much stuff do the
> get wrong I *don't* know about? :(

Lots, but what are you going to do?

Bias is everywhere, unavoidable, and human nature.

If you really care, your best bet is to cross-check everything -- read
different reviews of the Macworld speech, a couple different reviews for
the product you want to buy or the movie you want to see, don't just use
one news site to learn about current events, etc.


--
Chris Devers
DO NOT LEAVE IT IS NOT REAL

At 8:01 AM -0800 1/2/08, Chris Devers wrote:
> > If they get so much stuff wrong I know about, how much stuff do the
>> get wrong I *don't* know about? :(
>
>Lots, but what are you going to do?
>
>Bias is everywhere, unavoidable, and human nature.

Bias and error are unavoidable due the fact that humans are fallible,
but in all too many situations, what's "correct" is a matter of
opinion. There's a verifiable (with effort) answer to the question of
"Who is buried in Grant's tomb?" but the phrase "Ulysses S. Grant was
one of the least effective presidents in U.S. history." has numerous
accuracy values depending on the context in which the statement is
made (and much else).

Writing in such a way (or avoiding the topics entirely) to avoid
these sort of statements is tough, though I believe Wikipedia tries
hard to encourage/enforce it.

Since Google knols won't have any of this sort of oversight, they'll
probably contain a lot more statements that can be interpreted in
multiple ways.

cheers... -Adam

--On January 2, 2008 8:01:36 AM -0800 David Ross
<drdavidrossconsultant.com> wrote:

> As others have pointed out, it is not "take a few minutes".
>
> Plus the entire discussion which followed about the difficulties of
> editing an article mirror the process for the Mozilla foundation and
> posting bugs for Firefox, Thunderbird, SeaMonkey, etc... Create a "bug"
> id, figure out the bug system, figure out the abbreviations, figure out
> the classification system, get the attention of someone important, etc...
> And at some point someone regularly tells the crowd to "read the code and
> post a fix". Yeah, right.

Actually making minor edits in wikipedia is a matter of minues. Creating an
account is not necessary, just hit the edit button and have at it.

Making substantial edits is a different matter simply because you
(hopefully) want to be sure of your own sources and facts, this is the
level that requires actual time to get involved in the wiki article.

Totally agree on Bugzilla. I've got accounts on Gentoo Linux's, and
Mozilla's bugzilla systems and I dread having to use them. Most bugs I've
found are already reported but don't seem to turn up in my searches. A
wiki bug reporting system would be nice (Trac has a wiki built into it's
bug system and Trac doesn't irrate me as much, but i'm not sure it scales
as large as bugzilla).

On 1/2/08 at 1:48 AM, acetidbits.com (Adam C. Engst) wrote:

>2) Although I haven't heard if there has been an update, the last
>time accuracy was compared, Wikipedia and the Encyclopedia
>Britannica were comparable. That said, I believe Wikipedia will
>always have more errors, simply because it's accepting a vast amount
>more information (and even types of information) than any
>encyclopedia. Personally, I believe the accuracy of knols will be at
>best comparable to Wikipedia, if not notably lower when taken as a
>whole, due the fallibility of individuals.

I can see how the same observation (more contributors to
Wikipedia than for Britannica) could be used to argue Wikipedia
is likely to have fewer errors than Britannica. The argument
would be since there are more contributors it is possible there
are more expert contributors looking over the article making it
more likely errors are caught.

What is missing here is any measure of expertise and how many
contributors with more than some threshold of expertise there
are for a given article.

And there is an additional problem in trying to measure
accuracy. That is the selection of articles to compare for
accuracy. If I just choose articles in my field of expertise I
likely have a good measure of errors for those articles but they
may not be truly representative of the global error rate. But if
I choose articles at random, I am less likely to have an
accurate assessment of the errors.

> I can see how the same observation (more contributors to
> Wikipedia than for Britannica) could be used to argue Wikipedia
> is likely to have fewer errors than Britannica. The argument
> would be since there are more contributors it is possible there
> are more expert contributors looking over the article making it
> more likely errors are caught.
>
> What is missing here is any measure of expertise and how many
> contributors with more than some threshold of expertise there
> are for a given article.

In the 1980s I knew someone who wrote an article on mathematics for
Encyclopedia Britannica. He said that after the editors had worked on
it that it he asked to have his named removed as author. He said that
they totally ruined his article and that the edited version was simply
wrong.

Paul Chernoff

On 2-Jan-2008, at 02:48, Adam C. Engst wrote:
> 3) For the most part, I'd argue that the errors in Wikipedia, such
> that they are, aren't a particularly big deal. When I go to Wikipedia
> to read about, say, Andrew Jackson, 7th president of the United
> States, I don't much care about the ultimate accuracy of specific
> details because I'm just there to get a broad overview of the man's
> life. That's true of nearly everything I look up in Wikipedia,

Or, indeed, in ANY encyclopedia. I don't read an encyclopedia article
expecting to have the entire breadth and scope of a topic laid at my
feet, I expect a fairly in depth article that gives me a good idea of
the topic.

The nice thing about Wikipedia is that so often the sources ARE cited
and it is trivial to followup on some of them and get even more
knowledge. Much easier than with the Britannica, in fact.

On 04/01/2008 1:30 AM, "Bill Rowe" <readlistssbcglobal.net> spake thus:

> What is missing here is any measure of expertise and how many
> contributors with more than some threshold of expertise there
> are for a given article.

Yes, otherwise you're just pushing an "infinite number of monkeys" argument
(which oddly enough, turned up a couple of days ago in the book I'm
currently reading).

I rather like this take on it (frames 1--3):

<http://www.somethingpositive.net/sp10312006.shtml>


--
Nigel Stanger, Dunedin, NEW ZEALAND.
http://xri.net/=nigel.stanger