Do you just use your system to track day to day minutiae? It is similar to a system I developed for rote memorization (except mine was on paper, back when all computers still had hand-wound cores). I learned to take notes as q-a pairs directly, and I still take notes that way. It fosters breaking a detail down into unambiguous components.
When I went back to school to become a tech writer, I wrote a simple computer learning program, in C for one of my assignments. The program took a collection of q-a pairs and randomly presented them, tracked hits and misses, re-presented missed items, etc. If you wanted to extend your system to help to memorize stuff, you're more than halfway there.
No, I don't track day to day minutiae with this system. This is for bigger picture answers. I still log my late nights in VoodooPad.
I'm also good at memorization but have decided that my limited brain capacity is about as full as it can handle. A system like this keeps me from having to memorize things. If it's not a daily-use type of answer, I just relegate it to the info bucket and let my search tools find it when it comes up again. That's why the q/a system is necessary, too often I forget the question; it's hard to find the answer without the question :).
Thank you for your thoughtful and thorough reply.
I think what you said, "Don’t try to fit everything into a single app. Consider a metadata system that allows you to use the appropriate app and be able to think of the collection of apps as the “bucket”." is key and will take your advice, re-read your other posts on the topic and build the "bucket".
Please write the follow-up post you alluded to if you find the time. Can't wait to read it and find out even more details perhaps.
How do you deal with questions where the answer is more conveniently provided by a picture?
I find that plain text works for me for almost everything BUT the knowledge base. Your needs and mine are quite similar but I find myself lacking the discipline to input everything in plain text.
Let's take the example you used on the project page at GitHub, the favorite cleaner : wouldn't it be easier to just take a picture of the product label? Many things I need to remember are like that: settings for a GUI tool, a son's favorite cereal, the model of printer ink, the refrigerator water filter, a cover of a book that drew my eye, hours of operation for a store etc. Sometimes the product name is not sufficient and I need to remember the size, model.
Do you transcribe them? Another example you provided: "best h.264 settings". What if the settings are from Handbrake preferences? Do you transcribe the settings in plain text?
It's a genuine question as I've been struggling with this riddle for quite some time: How does one handle these situations and achieve a trusted system where everything is in plain text?
How do you deal with questions where the answer is more conveniently provided by a picture?
I use a picture. That's why Evernote continues to be relevant for me.
I find that plain text works for me for almost everything BUT the knowledge base. Your needs and mine are quite similar but I find myself lacking the discipline to input everything in plain text.
I should be clear: the "knowledgebase" I'm talking about is specifically for things that are text based. It's for quick notes about things I found on the command line, written answers to basic questions I'll ask in the future. I still use Evernote for things like grabbing a picture of the parking column at the airport, PDF copies of receipts, PDFs of websites that I was more interested in for their design than their content, etc.
It's not about discipline, really, it's just--to me--this stuff is just easiest in words. There are obvious occasions where it would be time-consuming and pointless to transcribe things. That's why I use a Spotlight-based tagging and relationship system: I can search my Evernotes, PDF files and my plain text at the same time.
Let’s take the example you used on the project page at GitHub, the favorite cleaner : wouldn’t it be easier to just take a picture of the product label? Many things I need to remember are like that: settings for a GUI tool, a son’s favorite cereal, the model of printer ink, the refrigerator water filter, a cover of a book that drew my eye, hours of operation for a store etc. Sometimes the product name is not sufficient and I need to remember the size, model.
In this example all I really needed to know was the brand. If my wife asks me to find a certain product at the store, I'm almost certainly going to take a picture of the empty product for reference.
Do you transcribe them? Another example you provided: “best h.264 settings”. What if the settings are from Handbrake preferences? Do you transcribe the settings in plain text?
I do my h.264 conversions on the command line, so it makes perfect sense to store the settings as plain text. In the case of Handbrake, I just keep backups of my settings files. If ever I lost my hard drive, my SuperDuper! drive, my Time Machine, my Dropbox files and my Arq (offsite) backups, I'd have lost all of my text notes, too. Neither note would do me much good in that case.
It’s a genuine question as I’ve been struggling with this riddle for quite some time: How does one handle these situations and achieve a trusted system where everything is in plain text?
Don't try to fit everything into a single app. Consider a metadata system that allows you to use the appropriate app and be able to think of the collection of apps as the "bucket". For me, that's nvALT, a folder of PDF files tagged with OpenMeta (nvALT tags can be automatically saved as OpenMeta tags on the text files as well), and Evernote.
Aside from basic Spotlight queries, I have HoudahSpot queries set up that will specifically search my input text as tag: (OpenMeta) and keyword: (Spotlight keywords AND Evernote tags) at the same time, as well as location specific searches that allow me to create contexts from my shallow folder hierarchy.
This is really turning into another post, so I'll leave it at that for now. In short, just go with what works most intuitively and build a system around that.
I use a different approach to solve the same problem, though it sounds like you're doing most of it already. I just keep a dated text file for each days random notes, ideas and activities. Each item has a single general tag -- readingnote, journal, JavaScript, etc from a slow changing set of about 100 tags, based on areas of interest, technologies, machines and projects. A small script can give me the last x entries from any tag, referencing it's file/date. Makes it easy to find most things, as well as to browse by topic and date. And because everything is in the one place, it's easy to grep when the tags aren't working for me.
It started out as a way of logging things that I'd later pull out and put into a wiki for reference-- but I find I don't need to do that; although it would be nice to have done, it would be a pita to maintain.
When a project needs it's own notes, the daybook just makes a reference to the file, so I can work out what I was thinking much later.
I started modifying qq, and ended up with: qq0.rb (capture and search single line notes) — Gist.
- Entries are stored one per line in a single file (`0.txt`) or multiple files (`0*.txt`)
- Lines are formatted like `question: answer` and are assumed to be between 10 and 140 characters
### Usage
- No args lists all questions
- If argstring contains `: `, add a question
- If argstring starts with `~`, search using a regex
- Otherwise search using `Amatch::PairDistance`, or common bigrams. (Needs a better algorithm.)
No idea what it's supposed to do yet. But I guess I'll use it as an imaginary version of twitter.com/#!/lranta/status/118112920432422913:
Web app idea: Twitter-like site for posting answers to solved problems. Post titles could look like potential search engine phrases.
Another one: q — a Ruby script for capturing and searching "key: value"-style notes in a shell — Gist
<ul>
<li>Uses a matching algorithm that counts common 3 to 8 character substrings</li>
<li>Can also be used to copy entries, like holman/boom - GitHub</li>
<li>q @somenote something searches for something in qsomenote.txt</li>
</ul>
Using amatch/regex for search is a really good idea. I used amatch for fuzzy project name matching in an OmniFocus script and was really pleased with the results.
Remarkbox