Recasting Wikipedia as an AI Assisted global open Worksite
I managed to save my comment:
Wikipedia Survey Coordinators, Analysts, Managers, Board, Stakeholders,
I started the survey. When I clicked resume later, it cleared the survey and said I had cookies blocked. It mentioned an email, but would not go back to the email page again.
I have been using Wikipedia from the first days. For the last 25 years I have been studying the development of global collaborative communities on the Internet. When the original Internet Foundation was cancelled, I took over the domain and task. Setting policies, finding and encouraging best practices, reviewing sites and issues.
Several times I have downloaded Wikipedia to study it as a corpus. I have made recommendations to Wikimedia to standardize the format and content of Wikipedia at a fine scale, in line with tokenizing the whole Internet, particularly for search engines and GPT efforts which now use arbitrary tokens.
The equations on Wikipedia are NOT equations with universally accessible tools for symbolic mathematics and simulation.
The topics on Wikipedia are NOT the best that can be done with any of the topics. The authors of the works are not clearly identified.
Letting “anyone” make changes is not a good idea. Particularly, some really excellent approaches and presentations on a topic are erased or broken by later amateurs. When groups like Google use Wikipedia content to show in search results, they need to show the authors and give credit. If the material comes essentially from the references, it needs to be cited and linked to the specific material, not the general citation. For overall time and effort efficiency of the roughly 5 billion Internet users.
The topics on Wikipedia are NOT the best that can be done with any of the topics. The authors of the works are not clearly identified.
Letting “anyone” make changes is not a good idea. Particularly, some really excellent approaches and presentations on a topic are erased or broken by later amateurs. When groups like Google use Wikipedia content to show in search results, they need to show the authors and give credit. If the material comes essentially from the references, it needs to be cited and linked to the specific material, not the general citation. For overall time and effort efficiency of the roughly 5 billion Internet users.
I have a lot more I can tell you. The 25th Anniversary of the Internet Foundation is 23 Jul 2023. For the last 10 years or so I have been discussing internet policies and methods that encourage open sharing. When I review websites and topics, I try to find the people who can actually change policies and methods. Talking about things is usually a waste of time. It takes work. In fact the recommendation overall is for the Internet to be recast as WorkSites – where content is meticulously recorded and traced. Wikipedia can be restructured to allow collaboration between authors, reviewers, commentators, researchers, and all the categories of users.
There is massive duplication on Wikipedia. For instances ( site:Wikipedia.Org “Fourier Transform” ) has 10,600 entry points today. That kind of massive duplication and fragmentation causes huge time losses on the Internet, up to substantial portions of global GDP. During covid and now, there are 5 Billion humans looking for answers and about 7.5 Billion entry points for ( “covid” OR “coronavirus” OR “corona virus”). The same is true of about 18,000 deep topics that I have checked in detail in the last 25 years, working every day.
( site:wikipedia.org “speed of light”) has 17,600 entry points (Google, 11 Jun 2023, GMT 2:00 am)
There is only one node for “speed of light” value and all the places it is used on the Internet should point and link to all that is known about it. Starting with its precise CoData value.
“speed of light” has about 39.5 Million entry points on Google now
( site:wikipedia.org “Joules”) has 4300 Entry points. Wikipedia does not properly and consistently code units, dimensions, and conversions. You let people type in anything they want. But there are rules and standards they should follow. They mostly do not. I would say that Wikipedia is not curated properly, or at all. It could have an AI assisted (not the current GPT AIs) interface for browsing, collaboration and posting knowledge. I recommended several changes for Grammarly. Going beyond a few English grammar rules to how to properly code all knowledge for sharing and open collaboration for all humans.
Science Technology Engineering Mathematics Computing Finance Governance Other (STEMCFGO) covers most of the quantitative disciplines on the Internet. All the domain specific terms, units, dimensions, models, algorithms, reference constants and measurements, instruments, data streams, purposes and applications — need to be translated to SI Units and Internet global codes and tokens. It is NOT hard, just tedious. Requiring meticulous effort by authors, publishers, content creators and anyone adding anything to the Internet. Wikipedia is only a tiny part of the Internet.
The classification of Wikipedia is poorly supported, the handling of symbolic mathematics and the whole of tools and methods from STEMCFGO is poor. It is not hard (on the scale of the Internet as a whole, to change that and create open, collaborative, verifiable, (AI assisted), traceable, auditable worksites where people using the Internet can work together.
The whole of Wikipedia is “paper”. The tables are not data in global form, they are not linked to raw sources. Charts, graphs, models, simulations are not tied to globally accessible tools for everyone to afford to use them. The words can be compiled and translated to all human languages. That is critical for using the AI language models, after they open and properly tokenize and index their training sets.
Rather than asking for a survey that too few can access and process, you ought to create a collaborative site where anyone can work on the topic of “who uses Wikipedia”, “What are they trying to do?”, “How does that work (even for personal or casual things, if time and effort spent, it is work) contribute to the humans and related species?”
There are about 8 Billion humans, and their time is 365.25*24 hours per year per person. The value of time on the Internet can be estimated at global GDP per capita, about $15 per hour. If people are sleeping, one day that time might be valued properly. If people are “working” on the Internet, that time is valuable.
Certain large commercial sites their programmers and managers and marketers can make tiny changes that waste Million of hours of human time. They do not value it at $32 per hour in developed countries, so they mis-allocate their time and priorities.
Talk to people openly and let them see the whole of the discussions, cases, activities and issues. People can work collaboratively. Many of the topics on Wikipedia are “sort of collaboration” but where people are not properly identified, not encouraged to work openly.
I would recommend all the pages use at least one level of classification – people, places, materials, devices, processes, models, events. I have several alternatives. But it requires recoding and testing to see what works best for the millions of groups who use knowledge on the Internet.
RichardCollins@TheInternetFoundation.Org
Laurent,
I wish I had known it was saved. Maybe you can not use cookies and let people give their email.
There is a lot more that Wikipedia can do in the world. For all people in all countries.
But users of the information should be allowed to ask questions, make suggestions, work together and ask for topics to be processed and summarized by GPT language AIs. Classrooms, topic groups, global issues, cities, places, interests – it is the use of the information in the world where things are happening. Not “write it and they will come”.
Richard
