User:Carrite/Stolfi

While I don't agree with every word or idea, I think that Brazilian Wikipedian Jorge Stolfi is one of the most provocative dissident voices among us. Here is some of his best stuff... —Tim /// Carrite (talk) 17:10, 3 November 2012 (UTC)

Stolfi on deletionism
Deletionism is not and never was a "consensus", not even a majority opinion. It is the stupid and destructive ideology of a small minority, that prevailed by a combination of robot power and a broken "consensus" mechanism that, in any other context, would be called "ballot fraud". It is stupid, because its goal is to move Wikipedia backwards, towards obsolete standards of paper encyclopedias. It is destructive, because it has led to the loss of tens of thousands of good articles and good editors, and earned Wikipedia some very bad press — which, this time, was quite deserved. In conclusion, Wikipedia will soon change, in spite of all shrugs and so-whats. If it does not change course now, radically and quickly, it will just die in a few years.

To save itself, Wikipedia must set as its top goal the recruiting and keeping of new bona-fide editors. That includes banning deletionism and any other unnecessary practice, rule or feature that may drive those editors away, no matter how dear it may be to its inventors and users. That includes, in particular,
 * scrap the notability rule,
 * delete and ban all editorial article-side tags, and
 * stop the paranoia about unsourced BLPs.


 * All the best (with a bit more hope) --Jorge Stolfi (talk) 18:51, 28 February 2010 (UTC)

Thoroughly disappointed
When the unreferenced tag was developed, a straw poll was held *among the editors who had designed it* about where it should be placed. There were about 30 votes cast (out of a universe of perhaps 10,000 regular editors). These comprised 9 votes for for "top of article page", 10 votes for "bottom of article page", and 13 votes for "talk page". Needless to say, the obvious fourth alternative "nowhere" was not even in the ballot.

So, if that tag is now showing at the top of hundred of thousands of articles, it is because nine editors wanted it there, twenty-three did *not* want it there, and 9,970 editors did not have a chance to give their opinion.

A similar story applies to the Notability guidelines. I found a straw poll in the Notability talk page about a dozen or so specific questions. The questions were all in jargon (like "PROD" in this RfC) which I was unable to decipher, so presumably only the people who had been involved in the writing of the guidelines voted. There were less than 200 votes, and some of the items in the ballot passed with a tight majority — that is, less than 1% of the pool of active editors. Unfortunately I could not determine whether the final declared "consensus" honored these votes, or — as in the case of the tag — the minority opinion prevailed anyway.

As for this RfC, I see that 400 editors took part in phase I, 40 took part in phase II. The honest thing to do would be to declare this RfC hopelessly bungled and start all all over, beginning with the basic questions — like "are unrefernced BLPs a real problem?". Instead, it seems that this RfC will follow the same path as the other straw polls: the proposers stubbornly insist with their thesis, ignoring all data and arguments to the contrary, until all oposers get tired and leave; and then they will declare the "consensus" to be whatever they like.

In the summary to Phase 1 it was stated that all participants were concerned with the welfare of Wikipedia. I beg to differ. People who really care about Wikipedia should want to know, first, whether the unsourced BLPs are a real problem, and second, whether the proposed solution will do more good than harm. I don't see this worry among the proposers of the RfC. Indeed, it seems that the surest way to end a thread in this discussion is to post concrete numbers and examples. Instead of debating that data and what it means, the proposers merely shift to other threads.

It is clear to me that the original purpose of this RfC was not to find the best way to deal with the "problem" (or to find out whether the "problem" was real), but merely to obtain some legitimacy for what was a predetermined decision, namely that unsourced BLPs are to be deleted. If there is one thing that is clear from this discussion, is that unsourced BLPs are harmless and deleting them solely for being unsourced is extremely harmful.

The only explanation that I can find for the persistent wish to delete unsourced BLPs is psychological, namely the "lust for power" of editors who are tired of being just "workers" and want to be "bosses". In academia, were I work, this sort of thing happens all the time: people get tired of being just ordinary professors or researchers, and try to move to a position where, insted of working, they direct and control the work of other people.

How can one rise to be a "boss" in Wikipedia? Certainly not by editing contents: even if you edit 10,000 articles over several years and create a handful of "featured" ones, you will be just a "worker" like any of the other 10,000 regular editors. The same applies to any work (such as sourcing) that requires reading each article and thinking about its contents: no one can do that on more than 50-100 articles per day, the same top rate as for contents editing. Moreover, in that sort of work you often have to justify your edits to other "workers," and that puts you in the same "social level" as them.

A "boss" must do something that affects hundreds of thousands of articles, and does not require interacting with "workers" at their same level. It must be something definitive that an ordinary "worker" cannot stop or undo. It must be something that clearly put the "boss" on a higher level than the "workers."

That is the only explanation I can find for why we got the editorial tags at the top of articles. Robot-assisted tagging does not require thinking, so one can easily tag 1000 articles a day. The tagger is clearly "boss" because the tags are not "work", but "comands": every editorial tag says "I want this to be done, so some worker had better do it". A tagger is clearly above ordinary editors, because (by definition) the only way these can remove a tag is by complying with the wish of the tagger. Article tags have also the "advantage" that they violate the basic rule, "all editorial comments must go in the talk page": that is an advantage because (as in real life) one's social status is measured by the rules one can violate with impunity.

And that is also the only explanation I can think for this RfC and the way it was carried out. The real "problem" of the unsourced BLPs is that the "bosses, after sticking hundreds of thousands of tags, realized that they had been largely ignored — that is, the "workers" did not rush out to comply to their commands. That was doubly frustrating: not only it negated the authority of the "bosses", but made them look silly for wasting all that tagging work for nothing.

Enter then the idea of deleting all unsourced BLPs. Like tagging, deleting is something that can be done very quickly en masse, without having to read the articles. Like tagging, deletion cannot be undone by ordinary editors. Even if each deletion has to be voted in the AfD, the place and timing of the vote ensures that voters will be mostly "bosses", and the final decision is made by a "boss": if one or two "workers" happen to see the AfD all in time and cast their vote, they can be just ignored.

That explains why no one here seems interested in statistics that prove that unsourced BLPs are harmless, or in the damage that deleting them might do. That explains why the proposers adamantly refuse to allow an editor other than the tagger to remove a tag without complying with its command. That is why they adamantly refuse to extend the AfD voting period beyond 7 days: for, if more "workers" get a chance to vote, they may out-vote the "bosses". After all, a Master of a thousand Slaves is not a Master at all if he lets even one Slave disobey his commands, or lets Slaves vote on whether to obey them.

Five years ago, Wikipedia could be defined as "three million encyclopedia articles which anyone can edit". I am afraid that today it has become "a decadent social networking site with 10,000 members who have three million articles to play with". One just has to look at the pages in the "User talk:", "Wikipedia talk:", and "Template talk:" to realize that most Wikipedia decisions are being made by a small minority of "bosses" who seem to derive more pleasure out of social interaction (and, in particular, the sense of power that comes from "bossing" over other members) than on making real substantial contributions to Wikipedia.

At the root of the problem is that Wikipedia's decision-making mechanism is thoroughly broken. As we saw here, and in countless other cases, any clique of ten editors can write a rule or standard, vote it among themselves, and declare it "consensus". Almost every guideline in Wikipedia:* was decided in this way. No country could survive more than a few years with such a "randomcratic" government; and it seems that Wikipedia cannot either.

All the best (if still possible to hope), --Jorge Stolfi (talk) 22:43, 23 February 2010 (UTC)

The cost of uniformity
Too much edit-energy is currently spent in trying to make certain groups of articles conform to group-specific standards, such as order and title of sections, level of detail, style of figures and tables, mathematical notation, and so on.

Much of this effort has come to be associated with the Wikipedia 1.0 project and the many WikiProjects that were set up later. Many disputes betwen editors were apparently caused by conflicts between standards of overlapping Wikiprojects, or isolated editors disagreeing with the project standards, or by project standards that turn out to be inadequate for some articles.

The perceived need for uniformity is also responsible for a substantial slice of Wikipedia's complexity. It is one of the main reasons behind the absolutely ridiculous inflation of the Wikipedia style manuals and the creation of hundreds of project-specific guidelines, as well as the invention of infoboxes, the templates and dozens of other formatting templates, the Wikipedia table syntax, and much more.

The notion that uniformity is important is largely a legacy of the printed-paper world. When someone considers buying a paper encyclopedia or subscribing to a paper magazine, he will typically browse through the entire book or issue, and his decision to buy or not will be influenced to a large extent by the purely visual aesthetic quality of the object as a whole. He will hardly notice the quality of the information contained in individual articles. Even if each article, seen in isolation, is nicely and consistently formatted, the variation from article to article will make the whole thing seem ugly and shoddy.

Moreover, uniformity of style and layout across the entire book or issue implies the existence of a tightly-managed editorial body with sufficient human and financial resources; and the buyer will unconsciously assume that the same strict oversight that is evident in the style of all articles was applied also to their content and accuracy.

Uniformity of style, layout, language and notation is also important large works which will be extensively read, either serially (e.g. novels, textbooks, chronicles) or randomly (e.g. dictionaries, manuals, travel guides, catalogs). Uniformity then helps the reader because he has to learn the section structure, notation and terminology only once; and helps the writer because he does not have to redefine terms and notation again in every chapter or section. A uniform layout also helps the reader to quickly find specific information in each section (dates, theorems, part numbers, open hours, etc.)

Wikipedia is not paper
However, Wikipedia is not sold in bookstores. It is not a paper encyclopedia, nor a magazine or scholarly journal; it is not a manual, catalog, textbook, travel guide, etc. etc.; nor a library of such things. Therefore, none of the above reasons apply to it.

Wikipedia does not have to "sell" itself, and readers never have to decide whether they should "buy" it. So Wikipedia does not need to try to look nice to casual browsers.

Most readers will come to Wikipedia in search of very specific information, and therefore will read only one or two articles; and those are the readers Wikipedia is being written for. Some readers may browse Wikipedia for pleasure or general curiosity, but these should not be our primary targets; and, anyway, these "surfers" are likely to jump haphazardly from dinosaur to Dyna-Soar to sore throat, so they will hardly notice whether Dyna-Soar and X15 have a uniform structure or not. Moreover Wikipedia does not have a tightly managed editorial team nor hired editors, and will never have; therefore it should not try to pretend that it has them.

Wikipedia is also not a manual, catalog, textbook, travel guide, etc. etc.; nor a library of such things. It is unlikely that a reader will want to go through all Wikipedia articles on "racing cars" or "christianity" in a single session, much less in any specific order chosen by the editors. Readers who want general information on those topics should (and generally will) read the appropriate master or overview articles, and then perhaps they will read one or two of the subsidiary articles. But they are just as likely to jump instead to related but off-topic articles, such as "gasoline" or "Nero".

As for readers who want to learn calculus or convert to Buddhism, they will find that a Wikibook or a regular book is much more effective, and much more pleasant to read, than any collection of Wikipedia articles.

Finally, the drive for uniformity across all articles of any given theme is both a symptom and a tool of "tribalism", the tendency to fragment the pool of editors into "tribes" of people with similar interests, backgrounds, tastes, etc. Tribalism should be discouraged in Wikipedia, because it runs against the fundamental unity of knowledge, and hence of Wikipedia; and because is encourages biased points of view, and leads to a situation where the larger groups can impose their views on smaller groups and isolated editors by sheer numerical strength, than by the strength of arguments. These costs negate any advantages that the formation of tribes might have. But that problem is the topic of another essay.

Good uniformity
A standard that aims to achieve uniformity across articles is important only to the extent that it might substantially help the bulk of Wikipedia readers. If most readers generally access Wikipedia sporadiclly and randomly, as I believe, then the only standards worth enforcing are those that can be enforced for all articles, independently of topic; and which can be unconsciously learned by those casual readers after reading two or three random articles.

Very few style standards will pass the above criteria. Examples of good and important standards are the use of bold in the lead section only for the article's topic, and the rule that the first sentence must be a definition of that topic. These rules helps the reader have a clear idea of whether he got to the right page, and also give him the most important information about the topic. Examples of somewhat helpful but not so important rules are the consistent placement and contents of the "See also"/"References"/"External links" sections. These standards are generally desirable, but alternative layouts and section titles should be accepted if they are expected to be substantially more helpful to the readers of a particular article.

Examples of standards that should not be established, even among a small set of related articles, are the level of detail, table layout and colors, the number and arrangement of the other sections, names of variables, style of figures, etc. It would be nice if they were uniform, but the value of such uniformity to readers is insignificant compared to the cost of implementing that standard across 10-20 articles.

-

Stolfi: Wikipedia Is Dying...
Wikipedia Is Dying — and What Can Be Done About It