Well, this is the ugly truth. Nobody can visualize fast enough for their reading speed. That is if we visualize each and every word. However, why would we visualize each and every word?
For some tasks we define the goal as 100% retention. It could be learning a new language, or learning a poem or a joke, or remembering definitions. In paretto of things we do, there is always this annoying 1% of tasks that requires our 100% attention. The best way to approach this within our framework is high-level visualization. Creating a single consistent highly accurate scenario and placing within this scenario detailed visualization for each and every word really works. It is also highly entertaining. Notice that middle-level visualization may also work for shorter texts. Check out middle-level visualizations on technical material . In case of detailed visualization, the time/reading speed is not a constraint. It is not reasonable to create such a detailed level of visualization within seconds.
For most tasks (in my case around 80%) we can do with very few (0-5) markers per document. For example, when reading blogs you are not expected to find A LOT of new information or generate A LOT of markers. So when reading the article, if relevant, generate a couple of markers, but when finished reading do spend some time to link them properly. This method we call “hyperlinking” is harder than it seams, since you need top-notch analytical capabilities to link only the relevant information to the relevant information. You can read blogs and some books at 2000wpm and generate good properly linked markers, but your markers will cover the text sparsely [you may miss many details].
And then there are those tasks that are well balanced between the reading speed and retention [like some wikipedia articles]. These tasks are the tasks the original superlearner course was built for. The idea is as following: read the document at the regular speed (~1000wpm) and try to generate markers as you would generate free associations. You get a lot of markers, some of them less accurate than others [based on amount of training], some of them linked [based on content structure], some of them subvocalized [names etc]. It is quite a mess when reading at first. When done reading stop and try to recreate the text using the markers you have. If you need specific details, read the text again very fast, pausing only to remember the details you need [like dates and places] with high-quality markers. Then read a third time to unify the experience – have all the details accurate. Each time try to recreate the article from your memory when done reading. The resulting average speed will be around 600wpm with 80% retention, but you will remember all the details you really need [dates, places, people, reasons – everything needed for a history or neuroeconomics test].
Nobody I know can visualize every detail and read at his or hers top speed. By choosing the right strategy some of us can read VERY fast and yet remember sufficient amount of details to impress the peers.