Archiving workflow

Last edited by alastor933 on Wed Apr 09, 2008 1:36 pm, edited 2 times in total.

Here's my workflow for archiving. I tried to make it as friction-less as possible (or you won't do it, as GTD evangelists keep telling us).
It consists of these components:
⁃ Automator plugin to set metadata and move items
⁃ Hazel rules to keep the Archive sorted, and to update metadata
⁃ Smart folder to triage old items

The container
I created a folder ~/Archive, with an alphabet of subfolders inside - this script made the alphabet.
Why the alphabet? Well, this is a relic of analog GTD, but it also helps in keeping me from creating ever-deeper folder hierarchies. In this particular adaptation there's no real benefit, except that I dislike going through long lists (friction!).

Metadata for archiving
I used SpotMeta to create some metadata attributes for archiving:
archiveDate (date): date of archiving
archiveCategory (list, single choice): this is the metadata version of the 3 folders I had before
The archiveCategory list contains "Archive", "Reference" and "Software".
I also set up 3 smart folders for these categories, for a quick browse.

Archive from anywhere
An Automator plugin "Archive to..." sends items to the Archive (script).
It contains 3 actions: Get Finder selection > Run AppleScript > Move selection to Archive.
The script will get/set the attributes on the selected items.
To use: select files/folders, select plugin from context menu, and choose archiving category from dialog window. That's all.

Rule to sort items as they enter the Archive
When items get into the Archive, there's a Hazel rule to sort them by name into the A-Z subfolders. Item type is not considered, Hazel treats folders same as files. You do need a condition to keep Hazel from sorting the A-Z folders into themselves.
Looks like this:
if name starts with 'a'
and name is not 'a'
then sort file into subfolder with pattern 'A'
and so on, until 'Z'.

Maintain Archive: rule on A-Z folders
Items stay in the Archive for one year, and then get looked at: keep or trash?
When they are used during that year their archive date gets reset to the date they were used:
if archiveDate is before Date Last Opened
then run AppleScript: reset archiveDate to Date Last Opened

With this rule items get to Review only when they have not been used for a year.
Keeping recently used items out of Review means you get to trash a larger fraction of what does show up in there; a satisfying outcome.

The same rule on the top folder looks after items like "2006 widgets catalog" (these don't get sorted - this is preferable to having to remember to rename items like that: less friction). To keep the alphabet folders out of Review an additional condition is needed:
Color Label is not <take your pick>
And don't forget to actually label them.

Review Archive
I first thought of moving items for review, just as I do in Mori (where my GTD-ish lists live) but eventually decided to go for a smart folder. Drawback: you cannot browse folders in a smart folder. You also cannot open things; this resets archiveDate, and the item will disappear from the smart folder. Move them instead.
To collect old items the smart folder uses a raw query (look here for info on how to do date-based queries):
org_spotmeta_sd_archiveDate < $time.today(-365)
(view metadata with this workflow).

To review
Open smart folder 'Archived over a year ago'.
Thrash anything no longer needed ('be ruthless' is the advice generally given).
Use the archiving plugin to send items back to the Archive.

Your views?
Ways to improve on this?

Ways to collect stuff without actually going into the Archive?
GTD author David Allen says: "One simple alpha system files everything by topic, project, person, or company, so it can be in only three or four places when you forgot where you put it." This, of course, refers to an analog system; doing it digitally means adding metadata. Note that he does not file by name_of_document: what name would that be? Right now, I feel sorting on name is sufficient.

Do you archive in an entirely different way (if at all...)?

I welcome your thoughts.

Thanks for posting this. It's always interesting to see how people integrate Hazel into their larger flows. I will be adding features to aid in doing more GTD stuff (mostly date stuff). Also, with any luck, in the next version, you won't have to specify separate rules for each letter of the alphabet. Instead you would be able to have a single rule match a pattern where the first character is a letter and to substitute that letter (or whatever part matched) in the pattern for renaming or sorting.

Also, after the next version is out, I will probably post some articles on neat ways of using SpotMeta with Hazel.

Would it be doable to create "choose from list" within Hazel, to allow selection from all rules it knows about?
Just a thought...

Do you mean for creating new rules? There is a duplicate rule function plus you can copy rules by dragging rules between folders (hold down the option key to make sure it's a copy instead of a move).

For attaching an existing rule to a new folder. Dragging rules is OK, as long as you can see both folders. When the list gets much longer (like when having an alphabet...) you get lost quickly, and spend much time checking.
So I thought of a dropdown list that shows all rules present; selecting one or more would attach them to the currently selected folder(s).

I got lost once already, and started over (exporting working ruleset is good :oops:

)