How we tried to post an article about a new sport on Wikipedia, and it was stolen by robots…

It’s hard to find a black cat in a black room, especially if it’s not there…

Expression attributed to Confucius

Hello, dear readers!

Today I would like to tell you a story that does not quite fit into my understanding of the world around me.

I want to make a disclaimer right away. I’m almost sure that everything stems from the fact that I don’t understand this issue well. Of course, for a computer technology specialist who has been involved in such things all his life, there will not be a hint of intrigue here, but looking at the situation from my point of view, anything can begin to seem…

However, let’s talk about everything in order.

In March 2021, a good friend of mine contacted me, who from scratch created a new, very dynamic and unusual sport called bandwrestling, with the simplest (from the point of view of a “teapot” like me) request – to post an article about this sport on Wikipedia. I agreed without thinking, because I was sure that no difficulties could arise.

Dear readers, for your understanding I will provide a link to one video from the YouTube channel of the founder of bandwrestling just so you understand the essence of the sport mentioned. I didn’t get on the platform myself, but from the outside it looks interesting and dynamic, especially considering that during the fights the organizers began to use fitness trackers to track energy consumption and heart rate in real time.

Since at that time I had just taken a new book to translate, I forwarded Roman’s request to post the article on Wikipedia to my friend, who was more knowledgeable in computer technology, and I continued working on the translation.

And then this happened.

We posted the article, but it was immediately removed because it “did not have encyclopedic significance.” In addition, there were no references to authoritative publications, peer-reviewed scientific journals, clinical studies, mentions of the subject of the article (namely, the new sport of “bandwrestling”) in the media, etc.

At that time, such a formulation of the question became an insurmountable obstacle for us, since in global terms, saturating the information field with references to the subject of the article (and even more so organizing publications in peer-reviewed journals) was simply an overwhelming job that required serious funding and all the time at our disposal .

Therefore, Roman had to refuse the offer to independently move in the direction of comprehensive information development.

We went about our own business, Roman did his own, we kept in touch, I watched band wrestling, because this topic was not indifferent to me, and, as you understand, by virtue of writing this article, it continues to arouse interest to this day.

Now, from the perspective of the experience accumulated over the past 3 years and communication on this topic with specialists, it is possible to formulate a number of theses that would certainly help us (and all other people who face a similar task).

Let me quote advice from a person who has been familiar with Wikipedia for a long time.

Wikipedia is the third layer of literature. There should be two more layers underneath.

First layer – research literature. It will tell you how band wrestling happens and what happens in it. Reports, research of documents, typical cases, individual events and reports on incidents. Simply put, everything is collected here.

Second layer – This is analytical literature. It discusses what is written in the first layer, summarizes various incidents, makes assessments and comes to conclusions. Simply put, here they separate the good from the bad, sort the good into good and excellent, this level involves serious intellectual work.

Third layer – it is a reference book like Wikipedia. It briefly summarizes everything that is written in the second layer.

Alas, without the first two layers you won’t be able to write an article on Wikipedia..

And a number of very useful explanations from the same specialist:

1. People without a name, without a degree, without certification write to Wikipedia. To write on Wikipedia, you do not need to prove your knowledge. That’s why Wikipedia has a strict rule. It is always necessary to provide a link to the literature so that anyone can check whether what is written here is true or nonsense.

2. Unfortunately, literature is very different. There is a stupid report in a newspaper, there is nonsense posted independently on the Internet, and there is an untruthful essay in an advertising magazine. Therefore, on Wikipedia it is customary to rely only on serious, trustworthy literature. As a rule, this is a book published under the supervision of a respected editor, or an article in a peer-reviewed scientific journal, or the work of a decent university, or a serious and long-standing thematic site. Moreover, one article will not be enough, because there is a danger that the author of the article is biased and has a personal interest. The text for Wikipedia should be based on several serious publications by different authors in different publications, about four or five.

   If you have difficulty collecting such literature, or if the topic is too new, the easiest way is to spend several years working with departments of different universities, so that students study the topic, understand it, make presentations in university collections, or so that department teachers give their own articles in review journals.

3. In fact, an article for Wikipedia is equivalent to a third or fourth year university abstract. Collecting materials for it takes one or two months, writing an article takes another one or two months. Moreover, these materials should not be from the Moskovsky Komsomolets newspaper, but from serious and reliable publications.

   It follows that a good Wikipedia article cannot be written in a week. And if it requires such serious research and work, then it will cost the same as serious student work. Estimate how much you would pay a student to work at your company, and set aside that money to develop an article for Wikipedia.

4. An already written Wikipedia article can be corrected and supplemented by other people. Therefore, it is necessary to correctly convey in it what is written in the literature, so that other people do not suspect you of subjectivity and bias. And, of course, you need to be prepared for the fact that other people will add and edit the article as new literature emerges in the field of knowledge under consideration.

5. As I understand, you are interested in a sports topic. The sports theme never exists on its own, separate from people. Sport is inherently inseparable from the vast community of athletes. Athletes compete in competitions, which are again reflected in detailed reviews, detailed comparisons, long historical essays. The further and “thicker” such a story is, the more important and serious the literature will be, the more important and serious your article will seem.

   For example, I know of a case where an article about an artist was erased due to the fact that the artist was insignificant. And it really was true. But six years later, this artist starred in films, received awards, so essays and reviews began to be written about him. Literature appeared that proved his popularity, his importance for art, and proved the serious interest of researchers in his work. The second version of the article about this artist no longer raised any complaints, because it was based on authoritative literature.

So, just think of a Wikipedia article as a student essay, which is subject to exactly the same rules in terms of writing and content. That’s all!

Adds further understanding of the essence one article about posting information on Wikipedia. Short, clear, succinct.

Let me quote just a few sentences:

There are companies offering to post an article for money and control it on Wikipedia. They have nothing to do with Wikipedia itself, even if claim the opposite (there is even a thread about this official message “Wikimedia RU” – Russian regional Wikimedia organization). Of course, they can post the article – the project is open for editing by everyone. But, again, there are no guarantees of its safety at all or in the same form – just as in the case of posting an article by employees of any organization. Spending money on this is useless.

Just like collecting group of employees or create several accounts in an attempt to create the illusion of public support for his opinion. Wikipedia only counts arguments, not the number of votes, and such a move only compromises those who defend this opinion.

And especially there is no need to threaten Wikipedia and its contributors with court and by any other means – the result of this will only be permanent blocking threatened.

It is almost impossible to order anything from Wikipedia, because a single entity with this name simply does not exist. Wikipedia is a lot of very different people, different gender, age, nation, religion, citizenship, and most of them are anonymous. This set changes dynamically with each arrival of new and departure of old participants.

About a year passed and in June 2022, Roman officially registered the Bandwrestling Federation of the Udmurt Republic. Below I attach a scan of the certificate.

This fact does not in any way affect the further content of the article, however, I want to say that the creators of bandwrestling carried out organized and systematic work towards achieving their goal.

And six months later there was a boom in neural networks. I will not dwell on this in detail. In one sentence, this event (“neurobum”) can be described as follows: if in the second half of 2022 there were only two neural networks in the public domain and rumors – ChatGPT for text communication and MidJourney for image generation, then already at the beginning of 2023 The number of neural networks for various purposes began to experience almost exponential growth.

At that time, I was translating a book through ChatGPT (more details here) and naturally, I was eager to “test” ChatGPT for logic and the amount of accumulated knowledge. Here it is necessary to make a note regarding the fact that at that time ChatGPT did not have direct access to the network and its information training base was limited by date September 2021. The neural network did not know everything that happened after the specified date, as a result of which there were corresponding gaps in its answers.

We started asking the neural network about band wrestling. And here we have an ambiguous situation.

If at the beginning of communication on this topic ChatGPT seemed to voice more or less correct answers (see images No. 1 and No. 2 attached below), then suddenly it suddenly forgot everything (see images No. 3 and No. 4 attached below)…

I immediately apologize for the quality of the images, these are just screenshots, at that time ChatGPT itself was just developing, the version of the neural network from January 30 did not have a separate functional button “Continue answer”, this command had to be written “by hand” over and over again, but even with In this case, the neural network could go astray from the initial logic, start answering the question again, etc.

I apologize for attaching such images-screenshots, instead of the logs that ChatGPT stores for each question asked by the user, but as a “digital tidy”, quite scrupulous about the issue of order in the data, I deleted the logs, and managed to restore the screenshots from personal correspondence with Roman, the father of bandwrestling.

But MidJourney – if I’m not mistaken, at that time it was the 4th version of the neural network – coped with its part of the job just fine. Below will be presented several images on the topic of “bandwrestling” in the state of primary generation (i.e., images not processed in Photoshop or similar programs). I will not publish the information through which the below images on the topic of bandwrestling were obtained, since I believe that this information is not of decisive importance.

I understand what to score
a certain phrase-prompt into one or another neural network and can get a good result
everyone, and it seems even indecent to publish pictures generated
neural network, however, I apply them, since each of them contains
a piece of my inspiration and my soul. Sorry for the pompous expressions,
but it’s true. By the way, I liked one of them so much that I
added it to Home page of my sitededicated to
mainly translation activities.

Why is this part with the generator from neural networks present in the article? In order to explain to you, dear readers, that throughout all this time (3 years of communication with Roman) and to this day, the topic was interesting to me, but I could not deal with it due to the fact that my main area of ​​activity is This is a translation of literature.

I hope I haven’t bored you yet, and the climax of my story will come literally in the next sentence.

So, in the summer of 2023, I
writes Roman with a simple question, namely, whether I or my friend published,
who dealt with the issue of posting an article on Wikipedia, somewhere the same
article about bandwrestling (see screenshot of communication with Roman via VK below)?

I immediately replied that neither I nor my friend had anything to do with the appearance of this article on the Cyclopedia resource due to our general employment and suggested that Roman ask his friends. I was sure that, while promoting, Roman turned to several people and someone with a proactive type of thinking (from English – “taking measures in advance”, “acting in a proactive mode”), which I respect very much, especially in working relationships, played proactively and published the article.

Below is a screenshot of the correspondence with Roman:

It should be noted that the original version of the article that was published on Cyclopedia was identical to the article written by Roman. Why am I focusing on the original version?

All because according to stories articles on Cyclopedia, edits made 25 times.

The denouement of this detective cyberpunk story turned out to be extremely trivial… 😉

An article on Cyclopedia was written by a bot (AI, artificial intelligence) which was created as parser articles from Wiki (including unpublished ones) for auto-filling the Cyclopedia site.

I started googling what a parser is, but I only vaguely understood the concept.

The founder of the article is a certain cyclopedia user (but he has no data in his profile), and further editing was also carried out by users with zero data in their profile. They were created so that they would upload articles on their own behalf, but in fact they are bots.

Brief summary.

Due to the unambiguous operation of the principle Occam’s razors (when your doorbell rings in the evening, of course, you can assume that it’s Joe Biden, but in fact the caller turns out to be a neighbor who just came in to ask for salt), an extremely trivial and relatively understandable solution to this strange story was found.

However, in contrast to the principle of Occam’s Razor, we can put the results of the well-known Jung’s experience (the observer gives birth to the observed) and assume that what happened was the result of the unity of goal setting of an unlimited group of users and this led to the network becoming aware of itself and solving problems known to it alone.

I like the second option better, despite the fact that the first one is certainly simpler and more logical.

Please don’t judge harshly. Perhaps in 2024 we will learn a lot more…

And I want to say something else from myself…

The cyber god is coming. And he, like any other deity, has his own heralds.

So, I recommend being guided by the saying of Confucius “Don’t do anything bad even in the dark.”which is usually interpreted as a proposal to nip bad thoughts in the bud.