I am stuff-averse. I don’t like piles of anything and I’ve learned to hate paper. From sticky notes to all forms of pamphlets, documentation, and all the little paper things that come with modern lifestyles. It seems the promise of a paperless world has not materialized as well as I had hoped. Piles like this build up over time, and I just don’t know where to put these loose ends of life and work.
There was a time when I thought Things would rescue me from the madness. It didn’t. I was very excited when Evernote promised a life simplified by digitizing everything including my tablet scribbles. That wasn’t such a good strategy. Would more recent and modernized platforms such as Trello provide the place where knowledge goes to live? Perhaps, but not for me and not for these types of loose paper things that range from a hastily jotted phone number to a complex instruction guide. If not my productivity, my psyche is being decimated from [almost literally] a thousand paper cuts.
Transient Paper Artifacts
For me, the tools and experiments I used as I tried to coral what I often refer to as “transient paper artifacts” (TPAs) addressed a variety of helpful and productive solutions, but nothing that compelled me to sustain adoption for capturing the loose couplings of paper artifacts that seemed to languish everywhere. I desperately want to physically eliminate them and their annoying presence.
I think the systems I have experimented with over the years failed to hold my interest for two key reasons. Bear with me though, I’m still learning after a forty-year struggle with the transient paper artifact challenge.
-
Classification - The solution has to make a decisive dent by reducing or eliminating the tagging effort. A picture is not worth a thousand words if you cannot easily find it. And finding it requires classification at the very least. Ideally, it should add keyword tags and convert all text into something searchable that makes findability effortless.
-
Integration - It has to work in a context where many integrated services must collaborate. My ideal solution is not a product; rather, it’s a solution baked into [grander] product that I use for many information-related aspects of work and life. Whatever this is, it needs to be seamless, silent, and effortless.
What if I could take a photo of this solar charger instruction sheet and without hesitation, drop the physical instance of it in the trash can next to my desk?
That takes some faith and confidence that the photo is both reliably stored and findable over the long term. We’ve all tried this, of course, from the inception of digital photography. It was 1999 when I first attempted it by the way.
AI?
I was recently toying around with a Coda form that could capture a photo and use AI to convert the pixels to text and organize it in a Coda table. It has to do this without any human effort except point-and-shoot.
I wasn’t too surprised how easily this approach came together. Coda Packs are ideal for AI integration and there are ways to expose the underlying URLs for images that you capture via a form and into a table. There’s no need to waste ink describing the agility Coda provides for a data collection of transient paper artifacts that contains both image and all texts in that image. Let the automated tagging and other grouping and sorting begin. 
A Business Card
Curve Ball - Multiple Pages
As I experimented, I soon encountered a case where I needed to encapsulate multiple snapshots into a single artifact. Coda, of course, is well prepared for this at the forms and storage levels. My Pack needed a little encouragement, but I soon had a universal multi-image handler that captures all the text for a collection of photos and assimilates it into a single table record. Cool, right?
A Small Booklet
I tested it on a 21 page Raspberry PI booklet and not so cool; I need to work on scale issues. But it’s wonderful for up to ten or so pages.
Findability
I was hoping this requirement was a slam-dunk with Coda, but it’s not. I want a full-text search index and discoverability feature; the inverted kind with blistering performance, fuzzy logic, and wildcards. I have some ideas, but that’s for another day. The table-search feature is working well enough for me to put this experiment in the win column as long as it stands the test of time and scale.
Will It Stick?
I dunno. I’m still working through this experiment and trying to gauge the viability, but so far, I’m pretty much satisfied that all the little transient paper thingies can be gone but hardly forgotten.
Thoughts?
