I used to spend more time than I would like on Twitter. To change that, I intentionally made my password some random string that I don't write down, and log out after each browsing session. This way, I have to go through the painful password recovery process each time I want to log in. Eventually I don't log in to Twitter anymore.

Recently I picked up piano again after a 1-year hiatus, and it has replaced a lot of rote gaming as something I do for break. The impetus has been moving the digital piano from the living room to my next to my computer, directly next to where I work. This way, the piano is by me for about 8-12 hours a day. This actually didn't work for a few months and I got as far as unplugging the piano, intending to move it back, but eventually worked out nicely.

What surprises me a lot is how far you can take this strategy. It's almost always further than you expect it to be able to, both in the types of habits you can apply it to, and the complexity of obstacles you can set up to make it harder to do unproductive things. My Facebook account requires me to log in to a second email account, which is needed to log in to a third email account, to be able to log in. My phone has a byzantine process that I have to go through to unlock access to news sites. And in a classic fashion, I don't have candy in the house (and often throw them away if I get a batch somehow). This kind of thing often requires a one-time investment to eradicate habits that eat up just as much time every day.

And there's something about utilizing opportunities when I have willpower to control my behaviors when I am low on willpower that is quite neat and appealing to me. Although I've had successes as well with effortful productivity tricks (say GTD), after some significant recent successes I think I should just default to this strategy whenever possible. The process can get to be incredibly hacky (like having 3 emails accounts), but the results are almost always impeccable.

— Categorized under: #workflow, #productivity

For those not familiar with the app, Anki is an Open-Source notecard studying app. It's mainly used by students and language learners, but seems to be popular just among folks who likes to memorize things in a reliable way.

Looks like this:

My main use case for it has been studying Japanese, but more recently I found myself tossing random useful things in there too – like definition for some word I haven't encountered before, hard-to-remember names, keyboard shortcuts for Windows, that kind of thing.

From reading the manual, I found that it's surprisingly deep for something consumer-facing. As a developer with a bunch of things I wanted to automate, I had thought that I might need to write some APIs to interact with Anki, but now I think manual importing of spreadsheets is enough.

That does still leave the issue of creating the spreadsheets though. I've recently been using the “Language Learning with Netflix” Chrome extension, which unfortunately doesn't generate the romaji reading for Japanese words that were highlighted. I ended up writing a quick Python script on GitHub for it, which was pretty fun in it's own right.

Using Netflix with Anki has dual purposes. For one, it's given me a good reason to try out some different shows (Kakegurui was a recent one). For two, I'm actually kind of pumped to study these words now and see how well I can do after having them memorized.

Anyways, will report back on the experiment later. Hopefully if someone else is looking to study something, this post can point out some useful tools out there.

—

Categorized under: #workflow

Given that I decided to migrate from Microsoft To-Do to Notion recently, I spent some time this weekend looking into writing my own Microsoft To-Do exporter (official exporter is broken) and testing out Notion's capabilities.

The Microsoft To-Do part went better than expected, all things considered. The public API hides some folder structure information about that I needed, but with some additional web scraping I'm about 85% sure that I can get all my data exported.

The Notion Importer part was a bit more questionable. To my disappointment I found that task descriptions cannot be exported or imported. This is not quite up to par with what I'm looking for.

So I shopped around a bit more for alternatives, and one option that I looked at briefly early in January seems to be a promising candidate – RememberTheMilk (RTM). It looks like this:

It's kind of cute!

RTM's been around since 2004 (!!) and seems to meet all of my minimum requirements: 1. Good keyboard shortcut integration. 2. Android widget app. 3. Expressive public API. 4. Don't lose data. 5. Good search feature. 6. Excellent export feature. 7. Support for offline use. 8. Few crashes. 9. Cross-platform.

(Well, “few crashes” remains to be seen).

As for my nice to haves, referencing the list again: 1. Multi-layer nested pages. 2. Automatic sync with Instapaper. 3. Automatic sync with Kindle. 4. Automatic sync with Audible. 5. Good app longevity. 6. Suitable for long-form quotes and notes. 7. Good developer community. 8. Don't have to write my own service for it. 9. Integration with Anki. 10. Ability to sync with Twitter. 11. Cross-linking functionalities.

I saw RTM has IFTTT and Zapier integration, so it passes all the integration stuff. Longevity is definitely not a concern, and there are a surprising number of dev tools out there for various languages. So ignoring the low pri “cross-linking functionalities” feature, the only thing it's missing is multi-layer nested pages & suitability for long-form notes.

RTM does not allow for folders of lists, and given that this hasn't been done over 13 years of asking from users, I'm inclined to think that it's more of a philosophical choice as opposed to a feature gap. TEM does have a handy “smart list” feature that lets you generate lists based on queries on date and keywords. There's also a tagging feature that can work like lists. I'm a bit worried about how the smart list feature will scale with several megabytes worth of tasks, but assuming that it scales, I can work with not having folders of lists.

That leaves the final piece of the puzzle – suitability for long-form notes.

RTM only allows plain text in task descriptions. This isn't a downgrade from Wunderlist per-se, but does mean that RTM will be less suitable for screenshot and sketch-based notes. This is a pretty common limitation of task-tracking apps, with good reasons. Pictures are incomprehensibly large – the entirely of War and Peace in English is is around 3 megabytes, about as big as any random photo you might see on a Twitter post. The storage requirements would have been tremendous, as would server requirements and coding complexity. It makes sense that most task-tracking apps wouldn't support it.

I think I'm not quite ready to give up pictures in my notes yet, but don't mind it if my tasks don't have them. If anything, the limitation should allow me to easily migrate my tasks to a different app in the future. And the fact that pictures are so heavyweight means that modern servers wouldn't give a hiccup transferring my notes as long as the code is robust.

As for note-taking, I think I'll stick with OneNote for now, and see if I can sync the text portions (with Optical Character Recognition as a bonus if possible) to a separate RTM list.

So it seems that I'll likely be migrating to RememberTheMilk in a few weeks. Fingers crossed for a painless API transaction that'll let me import my ~5 years worth of tasks and thoughts!

— Categorized under: #workflow

For the past few years, Wunderlist has been my most used app and the main reason I still use my Macbook (for better keyboard shortcut integrations).

I use the app not only as a todo list, but also as storage for my notes. However, the app has recently been discontinued, and I'm not happy with the successor (Microsoft ToDo) for a few reasons.

1. Crashes at least once a day.
2. API is still in beta.
3. Developer community is a bit barren.
4. Export feature doesn't do anything.
5. Loses some of my data (~1 per 100 items).

On the plus side, it does have a really good Android android widget, which is one of my top priorities.

Specific complaints about To-Do aside, Wunderlist/To-Do model always had some pain points for my use cases. The two-layer list format (one layer of folder and one layer of list) isn't really suitable for organizing notes whilest including citations from articles, books, scientific papers, etc. The separation between my thinking todos and the media that inspired them makes it hard to track back my reasoning.

Finally, my 5-year-old Macbook Air (bless it) has been having battery, storage, and processing issues. Some of these are not really solvable without buying a really expensive new one, and it would be much better to just switch to Windows.

Moving away from ToDo (as soon as my Export issue gets resolved) seems like a good opportunity to solve all of the above problems at once. Since I have a bunch of use cases the app, I decided to try brainstorming the features I'm interested in and rank them to help my search.

Minimum requirements: 1. Good keyboard shortcut integration. 2. Android widget app. 3. Expressive public API. 4. Don't lose data. 5. Good search feature. 6. Excellent export feature. 7. Support for offline use. 8. Few crashes. 9. Cross-platform.

Nice-to-haves, in order of priority: 1. Multi-layer nested pages. 2. Automatic sync with Instapaper. 3. Automatic sync with Kindle. 4. Automatic sync with Audible. 5. Good app longevity. 6. Suitable for long-form quotes and notes. 7. Good developer community. 8. Don't have to write my own service for it. 9. Integration with Anki. 10. Ability to sync with Twitter. 11. Cross-linking functionalities.

It's ok to get features through Zapier/IFTTT. In fact, I think most of the automatic features will probably come through Zapier or IFTTT. I can't think of a single product that would meet all of the requirements without using those.

What's interesting here is that all the integrations – Instapaper, Kindle, Audible, Twitter, Anki, cross-platform – will probably come from specifically two conditions being met: good developer community and expressive public API. I haven't really given this much thought before, but the lesson here seems to be that as long as you have an expressive public API and a product that developers like enough, developers will make the product more useful for everyone. There is a real network effect going on here. In a sufficiently open platform that developers like, the number of developer contributions becomes a leading indicator of consumer usage.

Anyways, my current forerunner seems to be Notion, which meets all of the minimum requirements except “Android widget app” and “expressive public API”. For the prior, it seems like I might be able to get away with using a different app like Todoist as my widget, and do a once-a-week copy of the contents to Notion. For the latter, there are some popular repositories that has reverse-engineered Notion's API (!) and so opened the floodgates for integration. The obvious caveat being that tools can in theory break anytime.

The nice-to-have list will be a bit fiddly. There's a service called Readwise which can sync and collect Kindle & Instapaper highlights, while Snippet seems to be a service that has better automatic syncing of Kindle highlights. These support IFTTT/Zapier, but Notion don't. I'll have to look around for ways to get syncing onboard.

Other than some issues in the automatic sync items, which are admittedly most of the nice-to-have features, Notion seems to be the ideal candidate here. I am looking into Roam as well, though.

Finally, Evernote is an interesting alternative. I consider it abandonware at this point, so it's Really Weak in the longevity department. I'd like the new system to last for a while, so hopefully I don't have to look into Evernote too soon.

Anyways, these are my notes on the todo/notes app wishlist so far. Would be curious to hear recommendations, and expect to post back here when the search is done.

— Categorized under: #workflow