I know that is easier said than done.

“Bits of our identities” are conversation starters, signifiers, hints, badges, clues to deeper things about us, personal reminders, and anchors.

“Anchors.”

“That was a very anchoring conversation.”

“This is a real anchor around my neck.”

Which is a weird way to start a post about note-taking apps, but here we are.

Like, Emacs is part of my “Unix person” origin story. I can’t name another software tool I’ve used as consistently for 33 years. I suppose the Unix paradigm itself edges Emacs out for personal longevity, but not by much. And when I think about everything I was doing with that first Ultrix account in 1991, “running Emacs” is the only thing that remains from the list. No more Netrek, I don’t use USENET in a way that would be recognizable to Past Me, and if the work I did on the Landsraad assembly hall for the DUNE MUD remains – getting cones of silence to work felt like a real triumph – I haven’t been around to visit it for a few decades.

It is a bit of an anchor in the putatively good sense of the word “anchor.” Technological comfort food. One of the first things that goes on any new machine, and one of the first server-side things I test when I’m trying out a new remote access tool. But also a bit of an anchor in the not-great sense of the word, in that I will pay the “figure out how to express this in elisp” tax for hours, well beyond practicality or reason.

Most recently, I was ignoring some excellent advice from Prot regarding people who want to use his excellent Denote for task tracking:

“If you want my opinion though, be more forceful with the separation of concerns. Decouple your knowledge base from your ephemeral to-do list: Denote (and others) can be used for the former, while you let standard Org work splendidly for the latter—that is what I do, anyway. … “Do not mix your knowledge base with your to-do items.”

… and the complexity was piling up and up, the tradeoffs were getting worse, and there was simply no joy in the experience because I had gotten myself into that bitter “make this problem yield” mindset that eventually leads to less understanding and more hacked-up, suboptimal stuff.

I just don’t have time for it right now.

So it makes sense that my fallback position was “maybe Logseq would be fine,” because it’s got the whole “supports org-mode syntax” thing going on – leave the door open for a return to Emacsland once I have more time – and perhaps because it is just odd enough to tickle another bit of my self-image. Mercifully, the second I tried to solve a problem of moderate complexity I realized how much time I’d have to invest to do anything besides pick code samples up off the sidewalk and pop them in my mouth. So I backed away slowly.

And after that it makes sense that Tiddlywiki got a look because it has been around forever and there is a sort of cheerfully prosaic attitude among its users. But the plugins started creeping in and I was trying to get it to do stuff it doesn’t really want to do without a lot of third party assistance, so the whole “it’s very simple” thing was not allowed to find much expression.

So there I was, and I’m sort of glad that I chased my tail over the past few weeks because it tired me out a little, but left me with an idea of what I wanted to do: Take notes in a connected manner, blend a little of PARA with space for a slipbox approach, and have inline todos, and I wanted it to sync across a few devices.

Obsidian does all that very well. There is always the risk of plugin creep, but in past Obsidian experiments that has been less about extending the core feature set and more about removing repetitive work. The simple mission of “write notes, link between them, keep track of tasks” doesn’t take much, with mobile and sync managed competently. It runs on every platform I’d care to run it on.

It’s a little dull. But after a few days of “just using it” and adding little affordances here and there from past vaults as I’ve remembered them, it has the benefit of just working in a non-dramatic, non-head-desking, simple way. I haven’t had to really think about it much. I haven’t inadvertently broken it or misconfigured it in such a way that I’m scrambling around a minute before a meeting trying to get back into my own notes. It’s of sufficient maturity that you can look up the answers to things and they are often in written form, which minimizes the whole “if I see one more YouTube poster frame of a slack-jawed influencer taking 30 minutes to explain something I could have copied and pasted in ten seconds I’m going to do a murder” thing.

I was inclined to say “and it says nothing about me, at all.”

But it does say a few things: “Doesn’t want to think about this problem he created for himself in any more detail,” “will settle on Markdown even though it is inferior to org,” “can stand being associated with people who think a graph of their notes is interesting and useful if it means not having to think about this any longer,” “will pay for sync,” “considers seven plugins normal and reasonable, would not admit to nine,” and – most likely and eventually – “always seems to creep back to Emacs even though it seemed like he knew better last time he did this.”

… I really did. I don’t think it’s gonna take.

There’s a lot to like about it, and if your whole thing is “I’m gonna go try this new model of thinking about things,” it’s a fine representative of the smart/connected/non-hierarchical etc. notes market. It is very outline-centric, so org-mode and Workflowy people will feel more at home.

The approach I took to my trial was to just go with its preference for daily journal pages as the starting point. I did do an overlay on that using PARA, more or less.

• Daily pages included a section for meeting notes, a task inbox, and a section I called “facts” that was just meant to be “random snippets of this and that flowing in throughout the day.”
• My project pages included project notes and project-specific tasks.
• My area pages included notes and tasks that I knew were related to a given area when I created them.

Logseq includes a built-in TODO page that gathers all your open tasks, which you really need if you’re dumping todos into daily pages. On a desktop monitor it feels manageable. On a laptop panel, especially one with a 16:9 ratio, it feels overwhelming if you have many open tasks.

I came across a number of strategies for dealing with the problem of “loose tasks flying around in your note volume” including review of the TODO page, custom queries, and the use of a plugin that offers a way to quickly scroll through available tasks and inject the ones you want to tackle today into your journal.

Another plugin provided a way to automatically roll daily journal tasks over with the creation of a new daily page, but it threw me a few curveballs after a day of trying to use it without having to think about it.

Two obvious comparisons to make are org-mode in Emacs and Obsidian.

If you like the thought of a very outline-oriented notes and tasks manager, but wish there was a more robust and purpose-built sync capability than org-mode offers, Logseq offers org-mode syntax and has a paid sync capability that seems to work pretty well. If you don’t care about a mobile use case I don’t know why you’d pay for sync when there are things like Dropbox and Syncthing. But if getting at your stuff on your phone matters and you’re sort of over trying to make Beorg or Plainorg work, Logseq might be interesting.

Obsidian is the real competitor, though, and as I read through forums and subreddits I saw some hair-splitting over which solution was “better” based on features that come out-of-the-box. Logseq does have better built-in task management stuff, but a single, very mature plugin in Obsidian closes that gap and then exceeds Logseq’s out-of-the-box experience, and you’re back to finding plugins to get parity.

I’d be remiss to leave out org-roam (Emacs) and Denote (also Emacs). If your use case is just notes with no tasks, either of these will work pretty well, too. You’re just left with the sync challenge, and neither is suitable for mobile (IMHO … livable, but not great). I said “just notes with no tasks” because org-mode’s agenda, which is generally how you’re going to aggregate todos across a bunch of connected notes, has known scale issues over time as you add more and more files as potential sources of tasks. If you don’t mind adding more custom lisp and a package or two you can overcome that. You’re still left with the mobile problem.

What about one big page?

I thought about that, too, after a little Ed trolling. I’ve done the whole “one big org file” thing in the past, but that was a simpler time.

I did model PARA into a single org file, tweaked a few org-capture templates, and made some conventions to take advantage of the :CATEGORY: property and tags inheritance. That made the agenda a lot more digestible and useful.

The single-file approach also lets Beorg and Plainorg work pretty well.

That isn’t exactly a “connected notes” thing anymore. It’s more like Workflowy, and that’s if you’re doing it in org. I can’t imagine doing it without org-mode’s assistance. Too much going on.

Migration day

I’m sitting on a call with a migration specialist from Jamf, having promised the team that I’d be available to yell or threaten if they requested it. This is our third go at effecting a migration off of self-hosted infra and into their cloud. Fingers crossed Went fine. We’ll be able to decom a load balancer along with several compute and db nodes. We’ve also observed that a few backoffice integrations work much better in their cloud instance, meaning better reporting about the state of our end user compute fleet.

We’re sort of in a change season right now: In the coming half we’ve got a new password manager, a VPN refresh and some bits and bobs to improve our SSO situation. Getting Jamf in the rear-view mirror means a significant amount of headspace opens up for the team and will vastly improve our chances of success on one major initiative I pulled forward by two quarters and have sitting on the dock, awaiting a more reliable endpoint management setup to take advantage of it. Better reporting also means we’ll be able to make good on a major change we made to laptop fleet management that has improved our hardware budgeting situation a lot, but has needed better insight to really sing.

I took this job knowing there’d be a lot of simplification work to get through, and it hasn’t disappointed. There are days I want to pull my hair out – like the first and second times we had to scrub our Jamf migration – but there are also plenty of days when things come together and I clock out knowing we took a step forward.

Most days I also ask myself what’s compelling about all this to me. IT’s generally thankless work. But “run services orgs” is pretty much what I do, even when I’ve led engineering teams who probably didn’t want to think of themselves that way. I think the simplest answer is “you can help a lot of people and make a lot of things better,” including for the people who are also asking themselves why on earth they’re in IT every couple of days.

I also happen to like unknotting Christmas tree lights.

Kill It With Fire

Through this change season, I’ve been noticing and appreciating my Readwise notes from Kill It With Fire, Marianne Bellotti’s excellent book on dealing with legacy systems. I read it thinking it would be more “technical,” but it’s as much about the leadership and cultural challenges of dealing with legacy systems. Right up there with Camille Fourier’s The Manager’s Path.

Poking around with note-taking options, I finally came back around to Logseq, which I’ll reduce to “org-mode and Workflowy gang up on Obsidian.”

I was primed to not like it, but the org-mode syntax option and some relatively strong opinions about how things should be were compelling enough that I played around with it, then tried to implement a PARA-like structure, then watched a few videos to try to internalize how it wants you to behave.

The page link syntax is interesting. You can do either a traditional wikilinks double-brackets, or you can use #hashtag notation to link to a page, which – thanks to the inbound links section of each page – serves as a tag index.

Logseq solves the blank page problem by … well, it sort of doesn’t. It’s opinionated enough about its “everything is an outline, every line is a node” thing, as well as its “start from a journal page each day” thing, that you can tell it wants you to do something besides “make a page and start typing,” but you can’t be sure what.

I ended up deciding to just go with it. Sitting in the living room, staring at the blank daily journal page, I just made a node in the outline for the first thing that came to mind and dumped out my big projects and OKRs for the coming quarter. Then I dumped a packing list for a trip this week. Then some miscellaneous tasks to get done before I fly out.

Entering tasks, then thinking ahead to how I’d find them once I moved on to a new journal page the next day, led me to the Unfinished Business plugin, which will roll yesterday’s todos into today’s journal page. I think there are other ways to gather todos but my initial read of journal page todos vs. project or area todos is that they should be more ephemeral and bound to the day.

But I guess my summary, having messed around with this for all of 18 hours, and after a morning of using it for work, is that Logseq is very, very into the non-hierarchical, “networked notes” approach. It wants structure to be a matter of emergence.

Its main competitor for my attention is Denote, which is more of a collection of Legos with opinions that are less about your workflow and more about the metadata. I like the way Logseq provides a built-in backlink block automatically, though I could implement that with a capture template for Denote. I like Logseq’s more wiki-ish page creation conventions. I like that there’s a mobile app devoted to Logseq’s point of view, as opposed to the general-purpose mobile org clients.

It’s important, for org-mode people, to not go into this thinking it’s “org-mode with a native toolkit GUI.” You can use org-mode syntax in a list. There are some familiar org-mode conventions (e.g. src blocks). If you like everything being an outline, as in org-mode, you’ll be in familiar territory. But there’s a layer of plumbing that finds its way into the markup that org-mode is not going to make sense of if you ever depart Logseq. It’s instructive to set up a few pages where you’re transcluding nodes, etc. and then open those pages in Emacs to see what it does with them. In some cases, it can’t do anything because Logseq has its own markup overlay.

Dusting off the camera

I’ll be in Vancouver, BC this week for a work event, then staying over for a few days to do some tourism. Given the time of year and likely subject matters it’s an X-T5 kind of week. I’m just taking the 23mm/f1.4 WR. I considered the X100V, but I’d like the extra speed and the image stabilization.

I have not been very excited about photography lately, but the prospect of a new place and interesting scenery got me a little more excited. Looking forward to being out on the streets at night.

Wallabag bookmarking script

I submitted a PR for my Newsboat Wallabag bookmarking script. As with the Newsboat one, you can just run it from the command line, too, if you’re interested in some pre-written plumbing for Wallabag bookmarking. Just takes a URL as ARGV[0] and lets the Wallabag service pull all the metadata.