|
There is a way to avoid "related results". Search from the command line using a simpler HTTP client to retrieve the first batch of results and use the public API to retrieve subsequent results as JSON into a results file. Using this method I can retrieve the full number of possible results, e.g., hundreds or thousands, not simply the first batch, e.g., 15, 25 or whatever. I call this "continuation" searching. No Javascript is required. When people use highly complex graphical browsers and Javascript to make these requests it is sometimes triggered by a "button" at the bottom of a page called "more results" or something similar. Alternatively the requests may be triggered automatically when down scrolling, producing what some people call an "inifinite scroll" effect. Javascript is usually what produces the annoyances people experience. It is also used for tracking and telemetry. The token for the public API to retrieve the next batch of results is the results file. Using a simpler HTTP client, I can (a) search YouTube very quickly and comprehensively entirely from the command line, (b) download videos without ever visiting a YouTube page in a graphical browser and being exposed to annoyances, tracking and telemetry (c) switch from retrieving via search string to retrieving via channel name, (d) mix the search and channel results into a single results file, (e) output a TSV table from the results file. Currently I include the following fields in the TSV table: YouTubeID Title ChannelName Duration SearchString/ChannelPath Search/Channel The last field is just an indicator of whether the result is from a search or a channel. Videos are browsed and selected using a TSV table instead of an HTML search results page.^1 Because I am interested in videos of a certain duration, including "unpopular" videos with low view counts, I have found this is an optimal search method. No distractions. YouTube wants people to view popular, low quality, "viral" videos, e.g., "fake news", extremism and the like, because this "content" is optimal for their advertising business model. Hence automatic "recommendations". This is what makes YouTube search so horrible. Advertising as a "business model" for websites can influence design and have very harmful downstream effects. 1. The HTML pages for videos are where the "recommendations" come from. Thus I never see the recommendations as they are not part of this search/retrieval method. I could extract them from the HTML into a table if I wanted to see them, but I choose not to. I only extract the video download URLs from the pages for the videos. |