Nintil posts these days are born in a fairly traditional way. Papers are stashed and read, following citations to find more things to add to the stash. The stash is worked through, meaning being read and having sections and figures copy pasted for later search and use. No note-taking takes place; when getting into new domains I don't know what's relevant so I may try to keep track of too many things. Instead, I rely on volume to figure out what's relevant, and that same volume of reading leads to a built-in SRS that helps me remember what I'm writing about: if I'm reading 10 reviews about the same thing, and I end up reading them twice (once at first and a second time when I write the post) the concepts are going to be revised over and over. Not in an optimal schedule, but it's still built-in into the process.
Nintil posts are likewise consumed in a traditional way, the only novel addition may be the occasional sidenote. This is not ideal, I can put a month of work into the post (which I may mostly remember) but you will not if you just read it once and you are not reading about the topic literally every single day.
We know there are better ways of learning than just reading that don't require extensive human intervention (tutoring): leveraging the testing effect (merely being tested and going through the effort of trying to answer a question helps you remember), spaced repetition (seeing the same fact repeatedly over time at set intervals). I talked about this in my Bloom's two sigma post.
So you may ask "Hey José why don´t you put some SRS in Nintil or something", the reason is that I have little incentive to do that, as some have noted I write to some extent for myself and time thinking about what to put in the SRS system is time not invested in doing the reading and writing. Unless of course there are easy readymade tools where I don't have to think about that, perhaps where I can highlight sentences and have those put in the system and have prompts interspersed with the main text, in the same style. In any case, given that I'm writing the content and that my memory is sufficiently good, if I need to know about X I know where to find X almost always; to some extent knowledge can be offloaded to your writings and Google.
From the writer's point of view there is more tooling that could exist. Why do I need to write >14k words long blogposts in the first place if I want to know about something? Why do I have to read hundreds of papers and piece together a puzzle?. Systematic reviews are not good enough, some miss what others talk about extensively, or worse they will say contradictory things. So I end up looking at a lot of things to try to produce something that if did it right, you shouldn't need to read anything else (for the level of depth I'm targeting), all the weird nuances and quirks should be there and every conflicting evidence being explained and made coherent. I don't always do that because of time constraints (And I want to write about other things) but that's more or less the goal. But imagine this existed already, that you type "Immunosenescence" and you are led to a single place that talks about that topic, showing first a simpler version and the user could select a more expanded one. This article wouldn't deal in maybes and perhaps, but in more solid assertions with curated wording around them; so if instead of saying "Protein A might be involved in process B" you say "Protein A is most likely involved in process B" and "most likely" is tabulated to mean 70% confident. And anyone could dispute anything there and have it added. This sounds like wikipedia, but I ignore if someone tried to do this (Setting some pages to be under these special rules) it would be allowed.
Perhaps something like this could be made to work if scientific publications were accompanied with metadata that states how much support their paper lends to certain claims, how each of the assays lends weight to the conclusion. You would then have disputes about whether a given paper actually is as strong evidence as it seems, but at at least you could make those disputes legible to anyone.
Another feature of a system like this would be browsing by concepts. As before, you may have come across Protein A. But where does Protein A come from? Has it been found to be implicated in this unrelated process? Perhaps there is evidence that protein A causes B and that B causes C, but no one has actually checked if A would lead to C, pointing to potentially interesting connections.
This would solve a problem I currently have, which is that I want to keep certain posts up to date (Like the Longevity FAQ) but at the same time I have more specialised posts (Like the immunosenescence one above), so if something new comes I may need to find out if I have mentioned that, and if so, check if it needs updating. Then repeat for all the other posts.
The nice version of this would be a centralized repository of science, but there could be a personal version of this one could make, so I could make Nintil posts would be linearized views of the underlying graph.
In academic work, please cite this essay as:
Ricón, José Luis, “Tools for knowledge”, Nintil (2020-08-17), available at https://nintil.com/tools-for-knowledge/.