Tsundoku, but for Software

Alina Popa peeking over a tall stack of old, unread books in front of a full bookshelf

There’s a particular kind of optimism that lives on bookshelves. Every unread book represents a slightly better version of ourselves, someone with quieter evenings, more patience, and a much healthier relationship with reading before bed. The Japanese even have a word for buying books and letting them pile up unread, tsundoku, which I find oddly comforting. If an entire language needed a word for it, I’m clearly not the only one.

The same optimism shows up everywhere else. The online course sitting at 14% completion, waiting for “a calmer month.” The kitchen gadget that was going to change how we cook and now mostly changes how full the cupboard is. The drawer of cables we keep because one of them might, someday, belong to something important. We rarely buy these things because we need them today. We buy them for the person we imagine we’ll be once we finally have time.

The bookshelf becomes a subscription list

It took me an embarrassingly long time to notice that I do exactly the same thing at work. Tsundoku, it turns out, works perfectly well for software. The bookshelf just becomes a list of subscriptions.

In marketing and analytics, the collection usually looks something like this: a couple of analytics platforms, a reporting tool, a few tag debugging extensions, an AI assistant, a second AI assistant for comparison, a third one because everyone on LinkedIn seemed excited about it, an automation platform, and a note-taking app that was supposed to become a second brain and is currently more of a second attic. Each one arrived with a perfectly good reason. Individually, they all make sense. Collectively, they start to resemble the cable drawer.

My tech stack, mapped honestly

Analytics platform #1used daily
Analytics platform #2for when #1 disagrees
Reporting tool
Tag debugging extensions ×4one of them works
My tech stack(a bookshelf)
AI assistantused daily
AI assistant #2for comparison
AI assistant #3LinkedIn seemed excited
Automation platformneeds maintenance
Second brainrenewal email only
Every branch had a perfectly good reason at the time.

Looking honestly at what we already pay for can get a little awkward. Some overlap so much that they’re quietly doing the same job in two different interfaces. Some we only remember when the renewal email arrives. And now and then we buy a new solution for a problem one of our existing tools could already solve, if anyone had read past the second page of its documentation.

The real cost starts after the demo

The subscription fee is the easy part. The real cost starts after the demo, where the data is always clean and the integrations connect on the first try. Real life adds configuration, permissions, figuring out why the numbers don’t match the other dashboard, and a new set of logic, naming conventions and settings to keep in your head.

Where the time goes after you subscribe

  • Watching the demo (clean data, everything connects)
  • Configuration and permissions
  • Figuring out why the numbers don’t match the other dashboard
  • Learning its naming conventions
  • Actually using it
Proportions based on vibes, not a study. Your mileage may vary, mostly upward.

Then there are the small absurdities: subscribing to a productivity tool and spending an entire afternoon configuring it, or building an automation so elegant that it eventually needs its own maintenance schedule. At that point the work hasn’t really disappeared. It has moved somewhere with nicer icons.

Complexity and curiosity aren’t the problem

I want to be fair here, because I genuinely love this stuff. A complex tech stack isn’t automatically a problem. A measurement setup with a web container, a server container, a CRM integration, and a handful of APIs isn’t bloated just because it takes a while to describe. Using only a small slice of a specialized tool isn’t necessarily wasteful either. If that slice solves a problem that really matters, it might be the best money in the budget. Nobody uses every feature in GA4, and I’d be slightly worried about anyone who claimed to.

Curiosity isn’t the villain either. Trying something new simply because it looks interesting is how most of us learned anything useful in this field, and I have no plans to stop. The trouble starts when experiments quietly become permanent, when “let me see what this does” turns into “this is part of how we work now” somewhere between the free trial and the annual plan, without anyone actually deciding. Experimenting is cheap. Adopting is a commitment.

How an experiment becomes infrastructure

  1. “Let me see what this does.”
  2. Free trial
  3. “I’ll set it up properly later.”
  4. Connected to two other tools
  5. Annual plan (20% off!)
  6. “This is part of how we work now.”
Number of steps where someone actually decided: 0.

AI turned the volume up

AI has turned the volume up on all of this. There’s a new model, agent, framework, or automation platform practically every week, each one hinting that if you’re not using it yet, you’re already behind. Meanwhile, many of us are still working out how to get real value from the tools we adopted last month.

What makes this tricky is that collecting things to learn feels a lot like learning. Saving the tutorial, starring the repo and signing up for the beta all give us a small sense of progress, and once again we’re doing it for that future, better-organized version of ourselves. Real learning is slower and less glamorous. It happens when we use one tool long enough to hit its limits. There’s a lovely irony in spending hours discovering tools that promise to save us time, and some weeks it’s entirely possible to be busier optimizing how we work than doing the work itself.

Digital minimalism, without the pencil

This is where digital minimalism gets interesting for me, as long as it doesn’t turn into deleting everything and writing reports with a pencil. The number of tools matters much less than knowing why each one is there. That means noticing the difference between collecting possibilities and getting value from them. A tool I know deeply is usually worth more than three I know vaguely, and sometimes the most productive thing to do with an exciting new tool is to read about it and not subscribe quite yet.

I’m writing this as someone who is very much part of the problem, so I won’t pretend to have a neat system. But the next time something shiny appears, I suspect a better first question than “what can this do?” is “which of my current tools already does this, and have I ever checked?” The answer is often slightly embarrassing.

Before you hit “subscribe”

1. Does a tool I already pay for do this?

  • YesRead past page two of its documentation.
  • Not sureCheck. Then come back to this list.
  • NoGo to question 2.

2. Does it solve a problem I have this month?

  • NoRead about it, bookmark it, and don’t subscribe quite yet.
  • YesGo to question 3.

3. Who owns it once the free trial ends?

  • NobodyThat’s how the attic fills up.
  • Me, by nameFine. Add it to the spreadsheet.
Which spreadsheet tool? See the last paragraph.

The funny part is that I already know which tool would help most with all of this: a simple spreadsheet listing every subscription, what it’s for, and whether anyone still uses it. I’d make one today. I just need to decide which tool to make it in.

Stay curious.

Found this useful? Share it.