Scrivener as an Idea Processor: an AI engineer cogitates

I do quite a bit of research on the internet regarding a number of topics. I am an inventor, so work across disciplines. (I think many writers do this also.) The open question I wish to introduce concerns Scrivener as an idea processor - its ability to link related material from the web into “Minestrone Soups” of coherent “dishes.” Not only can related items be grouped together arbitrarily by human choice, but also “hunches” can form links on a non-reductionist basis, where the evidentiary basis is not just cause-and-effect.
Here is where I stand on this project:

  1. Use of Scrivener for this as-is. Scrivener already pretty useful to do idea collection. I am a relative newbie at this use of S., as I think the intended use of S. is to just collect back-up material to support a writing project, which centers around the writer’s “thought train.” Different writers will have narrow or wide thought trains. I believe most writers try to have a narrow thought train to gain coherence and “punch.” In any case, the AI system must not try to be a “thought engine,” but rather support the various intents of the user.
    Take this example: I am writing a business plan for a complex computer-social system to treat chronic disease in medicine. This has many dimensions or intents: Marketing success, actual benefits, legal boundaries, social acceptance, computer platforms, media, wearables, micro-diagnostic chips, trademark-patents, competitive products, market data, funding sources, etc. That’s a lot of areas to keep track of!
    Now how would you use Scrivener to do this? Well, write a “chapter” for each area listed above. This can be just a one-page summary of the particular area’s relevancy to the intent. For a business plan , finding funding sources is key at the beginning. Then for each area, have an “evidence locker” which has relevant links. This can be currently done by adding folders to “Research” manually. Not bad.

  2. I contacted help desk at Latte, the a helpful person suggested something called “DevonThink.” In the Devon blog, there are posts with titles like “I am a squirrel.” They seem to have a tag-based system. So this option would have two softwares used in tandem - Devonthink to organize, and Scrivener to allow coherent writing about the buzzing confusion of data out there from original intent. I haven’t tried Devon yet, but would question how to maintain personal coherence and intent with such an arrangement. Just a very clever file system does little to maintain coherence of intent. But I am speculating, as I haven’t tried it yet.

  3. Scrivener plus an “intent coherency support engine.” This software works off stated intents gleaned from personal writing and perhaps filling out priority forms and questionnaires. (This is itself a whole area - determining intent and priority of supporting activities.) This forms the immediate priorities. For example, in writing a business plan, after it is written, perhaps the immediate intent is to explore various funding sources, then after 4 researched, develop a plan to approach, then approach one of them.
    So…You wake up one morning and ask your AI speech-reco intellibot, “What do I do today?” Among other things it says to go research funding sources and allocates 6 hours to do so. There may be other routines it reminds you of, but the biz plan intent is highlighgted here.

So that’s off the top a few ideas about integrating Scrivener as an idea processor with the pov to capture and keep intent. This is an antidote to the ADHD epidemic.

And yes, the “null” hypothesis: “This is all nonsense. Go back to pen and paper.”

Any ideas welcome, as this is an open discussion.

The historian and author Steven Berlin Johnson has written about his use of DevonThink as an “evidence locker”, for example here: Others of us have combined that idea with Scrivener’s strengths for composing and writing - as, for example, here: I’m not clear what an “intent coherency support engine” would be. A task manager, like Omnifocus? And I’m also not sure if these are the sorts of ideas you’re seeking - but they’re typical of what exists at the moment. I think, however, that I personally would beware of complexity. Otherwise, there’d be a risk one would spend more time tending the system than actually benefitting from its use.

@@@ Thanks, I’ll look at links. An “intent coherency support engine” - the idea of it at least - is to keep focused in an increasingly distracting world. That’s what business plans attempt to do. Particularly with failure, a “Plan B” is essential, because the human tendency is to fall into inaction on failure. The plan guides daily activity. And that’s the way the brain works - it locks into plans unconsciously. If there is no plan, or just distractions - it is content to just goof off.
It is difficult, with a complex intent particularly - to keep a coherent intent. Many things can make an intent incoherent, until it is lost in the noise.
The “intellibot” tries to keep you on track. Hugh, do you ever have an intent to do something, then get distracted, and wake up 4 hours later wondering where the time went?

I am not sure that’s the only problem here. If you are really focused on your intent, you should be able to filter out what’s irrelevant. The real problem, at least for people like you that are trying to create links that are not so obvious (and with links here I include also potential investors for a business plan), is to keep your focus while at the same time keep an open and curious mind in search for some evidence in support of hypotheses in support of potential links (that need to be tested)

As a scientist, I often face the same problem. I haven’t found a solution yet.

@@@ Finding potential new links probably depends on the subject. What kind of science do you do?
My main problem is I waste time with a wandering mind.

Don’t knock pen and paper until you’ve tried it.

Certainly paper-based organizing methods are, for me, the only ones that are both robust enough and easy enough to maintain.

The problem with electronic systems is that they don’t know anything unless you explicitly tell them about it. And so they have a lot of overhead: tagging, sorting, labeling, prioritizing. If you skip that maintenance – which I inevitably do when I’m busy – they collapse.

With paper, at least for me, it’s a lot easier to get the system back to a usable state.


I organise all my research in … wait for it … Scrivener. Research goes into the Project’s Research folder whether I’m working on an academic journal paper, text book, novel. I prefer to work in a single environment so while tools like DevonThink might be better Scrivener is the writing tool of my choice. Before I started with Scrivener I used Journler (and some research for non-Scrivenerated projects remains there) but as projects have progressed I have moved material out of there into a Scrivener Research folder.

I favour using Scrivener’s Research folder as an “evidence locker” - until the locker becomes so full that it slows down the saving and back-up processes. Then I start winnowing the research until I have a key sub-set, leaving the main body somewhere else.

That “somewhere else” used to be DevonThink. But now it’s - roll of drums - the Finder’s folder hierarchy, which with its tags providing an extra dimension, plus liberal use of Hazel for filing, fits the bill nicely (although I do still use DevonThink for indexing and searching - there’s nothing better IMO).

No, not really. Perhaps a lifetime of working to deadlines - real deadlines which if missed would have cost me my job in a very public way - have taught me to just get on with what I’ve promised to do.