Wikipedia – The Sound of a Million Monkeys Typing

Britannica finally shut down its print version, and of course pundits blamed Wikipedia. They might have blamed Britannica online, on which you can search for free and read longer pieces. Just like Wikipedia.

We, the Weberati,  have been entranced by the rapid growth and apparent cooperative organization of the free-to-read Wikipedia.  Academics and commentators have largely accepted Wikipedia’s superiority as a “matter of fact” and many present the unimaginable volume of the resource as proof of the viability of the open, online wiki system.

I’m reviving an old issue  that has never been resolved and remains live for me. CAN Wikipedia get better over time?  How would we know?  How would we measure improvement?  Would we ever agree?  I’m not convinced that it will improve beyond its current state. It is still quite an accomplishment, but is still not the miracle we are expected by current social norm to respect.

Last year, Time published a fair overview of the history of Wikipedia. They revise the Britannica controversy by claiming the reverse now, that Wikipedia is almost as correct as Britannica. Still, even with a thousand human reviewers evaluating a sample subset of articles each, the comparison will remain unsatisfying. These publications have entirely different methodologies. We know WHY Britannica is good. We hope our Wikipedia article is correct and are pleased when we learn something.

A Million Programmers Programming

Good work spread thin is still thinly spread. I generally agree with Jaron Lanier’s observation (and his outlook in 2006’s Digital Maoism) that online collectivism us not yielding better quality. We have given open source enough time, and there haven’t been hugely disruptive innovations. Perhaps the good products get sold to Google, where they can try (again) to employ their global platform to innovate by throwing acquisitions. into the ad-supported market.

Open source, Linux, and Open Office are proof that collectives can debug pretty well. But as a software product and web services designer, I am skeptical that open source yields innovation. Collectives tend to fix things and groupthink takes over if there is no real ownership. Small groups that intensely care about a product yield innovation.

Open systems do not just “generate” a better quality innovation. Open Office is still nowhere near Microsoft Office, and Wikipedia has had time (over a decade) to prove its quality. If its not obviously better than Britannica, then small differences are not convincing to me. As long as there is a Britannica, it is “citable” as an authority. But as an academic, I will always distrust a Wikipedia citation and will discount them when presented in research.

Much has been made of the death of Britannica in print. But its online service is authoritative (i.e., trustworthy) and the articles reflect the “deep tail” of Britannica’s experience. Select something you really know about – systems engineering in Britannica and compare it with an equivalent systems engineering article in Wikipedia. The Wikipedia article is oddly unsatisfying.

This is the kind of article Wikipedia excels at, according to both Nature and Nicholas Carr. It is factual, has citations, images, and can be vetted by practitioners. It is just – OK. I would not waste my time updating it, because there are more “passionate” authors who may change what I add because they believe I am biased (otherwise, why would I be changing it?)

The Britannica authors are known experts, they reveal their names, and I know who they are. They have context, authority, and trust, all qualities of people as well as information. Their article is both deep and broad, and they are fair to all aspects of the field.

When searching in any corpus, relevance algorithms give either better precision or recall, but not both. In a similar way, Wikipedia has great recall (odds are there IS an article on your topic) but its precision is a crapshoot. In fact, the precision article I just linked to here was in fact good science. But try something fuzzier – and with multiple variants.  Sensemaking for example. It will often show the biases of the primary volunteer writer and – in this case – its is not reflective of agreement in a field, even if referenced.

The Horrible, Awful Editorial Process

I would never try to fix articles in Wikipedia – because their editing process is as closed as Britannica’s and I have no idea whether my labors will be useful or rejected. Britannica is a merit-based editorial process, it is not “open,”  but if I wrote for them, my efforts would stand. But Wikipedia is only superficially open. In reality it is closed to sincere and informed participants because it is now a club. The commons are selectively enclosed by the ranking of editors who have invested their time and have higher standing than authors. Because there is back-room dealing behind the History tab, its apparent openness is anything but inviting. There is no recourse for factual or style problems with a piece. You are on your own to fight it out with the club’s pseudonymous editors.

The editorial process can be vindictive and prejudicial.  The process for dealing with discrepancy or article takedowns is so labyrinthine, it reminds me of Terry Gilliam’s Brazil. I have seen my colleagues’ articles re-edited and even deleted by “editors,” those who have made themselves little kings in the Wikipedia editors club.

There seems to be no patience for a newbie’s article to be improved – for the rest of us to learn – unless the roving panopticon happens to miss an article nobody cares about, in which case the crap lasts forever. The criteria are unclear and not evenly enforced.

What Would I Do?

I intend to wean myself from the habit of using Wikipedia. Even for me, it has become a brand operation. A sub-attentional Google-like reflex, which we perform as a “success to the successful” habit of curious reinforcement. I actually like reading the authoritative Britannica – it reflects the wisdom of its select authors, not the normative blandness of a legion of undergrads.

Articles will revert in quality to the social mean, as they are not code and will not be debugged. It will always be handy and usually effective, but never truly authoritative. That in itself is reason for me to use an alternative.

My immediate solution is to encourage the old media, Britannica, to wiki-up. Add a commentary tab to every entry to enable controversies and counterpoints. Then, Wikipedia would be left trying to revert to the mean, a constant fight among editors to refine the ultimate page. Britannica could post a rich authoritative piece of knowledge and allow anyone to contribute to an informed commentary. A good fight between the sources might push Wikipedia to change its editorial practices.

Until then, I’m afraid Wikipedia is a reflection of the knowledge of our general society. It is quite often thinly researched, weak content with a strongly enforced pseudo-objective writing style that constructs the facade of authority. It can be good at times, but will never be as good as intentional articles written by experts AND public commentary.

Comments are closed.