A Lengthy Reboot

Finding Things

Mar 15, 2020

Digital things of value to me are in many places (Emails, Google Docs, PC documents, iPhone). At work, we use Wiki, SharePoint, Emails, documents in network drives, tickets, and recently OneNote as an information store.

Finding things is a challenge.

There are three ways I find things (in order of preference):

1. Go to a specific location
2. Full-text search
3. Ask someone

Often, I will use more than one of the above to get to the information I am looking for. For example, I'll go to a specific Google Drive folder where I keep my scanned paper receipts, then do a full-text search ¹ with the item or store name to find it. Or, at work, I will ask my colleague where to look for something, and I will learn it is in a certain folder in SharePoint, once I navigate to the folder, it is easy to browse through the documents and find it. Option 3 isn’t available at home.

Going to a specific location (my first tendency) requires hierarchical storage, like books in a library (room, aisle, shelf). Find things is a process of elimination.

• iPhone > Notes app > Folders (browse) > Note (browse)
• PC > Documents > Pictures > 2019 > (browse) > (scroll through large thumbnails)

To find things faster, the obvious solution is to have fewer options at every step of the hierarchy and that means more consolidation. At home, I plan to reorganize and move things into just three places:

1. Simplenote²
2. Google Drive
3. Documents folder in the PC

It would be ideal if it was just 1,2 and I didn’t need 3 but I don’t feel comfortable putting everything into Google (or equivalent), yet. Likewise, at work, we are also moving all digital things into OneNote. Looking forward to that.

One major problem with hierarchical storage: when the location of things isn’t obvious, where does it go? (do you organize photos by year/month/date folder hierarchy or by subject: kids, places, food). It will be harder to find no matter how you choose. Either it will be tedious, or you will go down the wrong path and don’t find it at all, because where you decided to search wasn’t the same place you decided to put it.

Full-text search is very powerful, but it tends to be slow and returns irrelevant results. You also can’t search for non-text things³.

So hierarchical navigation is my favourite, but it has its disadvantages, whereas full-text search has a wider reach, but it has problems too.

Tags

What is missing in my methods is labels or tags.

The only place where I have them is Pinboard. As of last week, I had over a thousand website links in Pinboard and 300+ tags for them, but whenever I went looking for links, I just browsed the list of links page-by-page or did a full-text search. Filtering by tags, despite being a prominent feature and easy to use, was the last resort.

The main advantage of tags over hierarchical storage is an object can have many tags (one-to-many). A photo can only exist in one folder (hierarchy) but it can have many tags (location, rating, names). I don’t like complexity, and tags are complex and arbitrary. That’s why I had so many useless tags (interview, park, review, video) in Pinboard. Tags weren’t important to me, so I didn’t name them well, because I had useless tags, I didn’t use them to find things.

I need to put more care into the naming of tags and refine them periodically. But I didn’t bother till recently when I read an article about the Zettlekasten method of keeping notes. Tags play a key role in that method.

Sasha Fast recommends naming tags by objects, not topic. Topic tags have a weaker connection, results in wider search, and don’t age as well as object-based tags.

e.g. of topic tags
idea, funny, quote

Tags named after objects referenced in the digital thing (note, photo) will result in better search relevance and are durable (years later, it will still mean the same), but they tend to have narrow use (tags are used only once), and you end up with a large number of tags.

e.g. of object tags
kids, caroline, NHL

So, a good tag must strike a balance between a topic and object to find the right scope of relevance.

This brings me to the second realization about useful tagging systems, all the tags should fit into one screen, be easy to browse through, and shouldn’t need a filter or search to find. If it is made into a tag cloud, the tags would still be readable, not requiring any squinting. I have decided on the arbitrary upper limit of 50 tags. A periodic review of tags will keep it relevant and keep the number at or below 50.

The third thing about tags is to avoid applying too many tags to an object. I am still early in my tagging process, but I feel this will keep tag count under control and simplify decision making.

I used to think of tags as keywords to add that I can full-text search for, and that made me add many tags to an item. Now I think of tags as a shortlist of predefined labels that I will use to filter things. Filter , not search.

Guided by what I learnt; I have reduced my Pinboard tags from 300+ to under 80 with the help of a homemade desktop app. I intend to konmari it down to less than 50 and stay with that upper limit regardless of how many links I have.

Another way I am experimenting with tags is in the Simplenote app. It supports tag completion, tags filter. It lets you export all notes to text files (no vendor lock-in). The app is available on iPhone, iPad, and Windows 10. It is open-source, free, and maintained by the maker of Wordpress. Simplenote is my tool of choice for the Zettelkasten method for note keeping. More on that another day.

I should probably start tagging my photos too because the only option I have now is hierarchical navigation (year/month/date) and photos make up 90% of my items. Windows 10 has a built-in search for tags in photos (search with tags:{text}) so the tooling is already there. I just need to start tagging new photos, and find some way of doing it for the 30,000+ past ones!

At work, the use of #tags is problematic with OneNote 2016 because you can’t search for # (it will remove the # and find the keyword returning false positives). The built-in tag system is a pain to use (drop-down menus, dialog boxes). One workaround is to prefix tags with some alphabet instead of # (e.g. zToDo) but this idea doesn’t have traction. We might just go with # and hope a newer OneNote⁴ works better.

I have just begun to take tags seriously, it will take time and fine-tuning, but I hope it will give me a useful second option between hierarchical navigation, and full-text search when I try to find things.