wikipedia-and-truth-on-the-internet.md (5903B)
1 # Truth on the Internet 2 3 ## Claim 4 5 Wikipedia is an okay source for purely factual information like mathematical topic, CS topics, etc., but it is not a good source for other kinds of information. Similarly, there are other internet sources that are good for academic information, but there is a lack of rigor in most other topics. 6 7 ## Reasoning 8 9 Wikipedia has been a good source of information on computer science and math topics for me. In general, articles are correct and I appreciate the formatting of the site working in my preferred web browser, lynx. When it comes to more subjective topics, it has not been a very good source for me. This also applies to the broader internet as well, albeit it is not consistently good on academic topics. 10 11 ### Gemini 12 13 Consider the Gemini protocol article on Wikipedia [1]. Most of the information in the article is acceptable, discussing actual information about the protocol, its history, and other related concepts, but the reception section is quite problematic. I will allow you to read it yourselves. 14 15 > Gemini is praised for its simplicity but criticized for "excluding people who use ordinary web browsers". Gemini's 16 > usefulness has been said to be "dependent on the kinds of content available on Gemini and whether it appeals or not". 17 > Stéphane Bortzmeyer has said Gemini is retro but with modern features. Daniel Stenberg reviewed the 0.16.1 protocol spec, 18 > and criticized it as weak on security (Trust on first use) and slow in performance (short-lived bursty TCP connections) if 19 > it was ever used to transfer resource heavy HTML pages.[13] Gemini pages are usually downloaded as gemtext only without 20 > requesting fonts or linked resources such as images. 21 22 Let's break down a few things. 23 24 #### Gemini Excludes People 25 26 > Gemini is praised for its simplicity but criticized for "excluding people who use ordinary web browsers". 27 28 A correction to this may be "it was once criticized by an individual for being exclusive, but said individual was misinformed". The only exclusion that is happening is browser makers do not support the protocol. There is no reason a normal web browser can't support Gemini, and no one who is developing the gemini protocol pushes back against its adoption in major web browsers. 29 30 Such a criticism is similar to criticizing HTTP because none of the Gemini browsers support it. Gemini is an exceptionally easy protocol to implement and to criticize it for exclusivity is a bit silly. 31 32 #### Security Concerns 33 34 > Daniel Stenberg reviewed the 0.16.1 protocol spec, and criticized it as weak on security (Trust on first use) 35 36 I view his first criticism as a selling point of Gemini. TOFU is how I believe the internet should work, or maybe DANES, but certainly not CAs. CAs are antithetical to the spirit of the internet. The internet is supposed to be free of authorities, but when CAs are considered the authority on who is able to host a website with HTTPS, which is functionally a necessity to have a voice on the internet, we do have arbiters of who can speak, and we lose freedom. 37 38 I think the term CA is a misnomer because there is nothing authoritative about CAs, and there is no authority on the internet. Furthermore, there have been many CA incidents in the past [2] [3], which is to be expected when authorities are appointed. 39 40 #### Slowness 41 42 > slow in performance (short-lived bursty TCP connections) if 43 > it was ever used to transfer resource heavy HTML pages.[13] Gemini pages are usually downloaded as gemtext only without 44 > requesting fonts or linked resources such as images. 45 46 I seem to be missing something here. A sportscar is slow when it is towing a tree up a hill, but that doesn't make a sportscar slow. If the Gemini protocol were saddled with the burden of HTML then yes, it would be slow, but that's the thing, it's not. It's made for gemtext not HTML/JS soydevery. 47 48 ### YouTube Video Uploading Metrics 49 50 YouTube video upload metrics are what prompted me to write this post. I was doing research for a project where I wanted to determine how long YouTube's business model of not deleting any videos on the platform, apart from those breaking ToS, could be sustained [4]. I started with a duckduckgo search and the top search result claimed that in February 2025 2.6 million videos were uploaded every day [5]. This cite then linked to another site [6] which made no such assertion. I returned to the original site to realize I'd been tricked. I should've known the domain SEO.ai was an AI slop SEO farming site, but it was the top one. I then went to Google to see what results I would get there. The same result showed at the top, but this time I got an AI summary which stated the exact same hallucinated claim. 51 52 Truth can't be found on the open web anymore. There are too many layers of nonsense. We have an AI summarizing an AI that summarized an unsubstantiated blog post that didn't even make the claim the first AI summary thought it did. 53 54 ## What Can Be Done 55 56 At present, we can stop, or limit, our usage of search engines. I plan to replace search engines with going directly to sites that I know are useful, using RSS, and trying to reference books as much as possible. It's modestly disappointing, but good options are lacking right now. If you have any thoughts on this, feel free to email me! 57 58 ## Sources 59 60 => https://en.wikipedia.org/wiki/Gemini_(protocol) Gemini Protocol Wikipedia Article 61 => https://en.wikipedia.org/wiki/DigiNotar CA Hacked 62 => https://sslmate.com/resources/certificate_authority_failures Even More CA Issues 63 => gemini://blog.laack.co/sustainability-of-youtube.gmi The Sustainability of YouTube 64 => https://web.archive.org/web/20250814122654/https://seo.ai/blog/how-many-videos-are-on-youtube AI Summary of an AI Summary 65 => https://web.archive.org/web/20250304100048/https://photutorial.com/how-many-videos-on-youtube/ AI Summary of a Poorly Researched Blog Post