Why did I leave OmniFocus for Obsidian? Or: Conceptual models in productivity systems and the complexity of knowledge work

Continuing the discussion from Task Management in Evernote:

Disclaimer 1: this post doesn’t mean that I think any app is better than any other app, nor that any kind of work is more or less complex, nor that complexity is a good thing. It’s just an explanation of the theory behind my practice.

Disclaimer 2: This is a bit of a mess, but I’m trying to put something out there ASAP rather than leave it for another day and never return to it.

Dangit, why do all the interesting questions get asked while I’m on a deadline? I don’t think you’re wrong, but that you’ve found a good fit between your system and your information needs. I have some ideas about that.

TL;DR: My suspicion is that there’s a sweet spot where complicated tools fit complicated work. If your work is simpler, a complicated tool is overkill. If your work is more complex, the conceptual model inherent to complicated tools may not fit, causing friction. If you aren’t frustrated by the conceptual models of the apps you use, keep using them!


Information systems & productivity systems

  • An Information System (IS) is a tool that represents the state of some part of the real world. When well-designed, they maintain that representation perfectly, they effectively inform us about the state of the domain, and they allow us to act on the domain.
  • A conceptual model is the understanding of the domain built into an information system. E.g., OmniFocus thinks that Projects have Tasks which have Due Dates. OmniFocus does not think Tasks have “do” dates.
  • Our productivity systems are information systems. They represent the state of our work: the things we want/need to do and everything that might be involved in doing those things.

Knowledge work: simple, complicated, and complex

  • All our work can be broken down into two things: the outcome we desire from their the work, and the next action required to progress on the work. If an outcome requires more than one next action to realize it, it’s a project. (Thanks, David Allen!)
  • All our outcomes can also be broken down along two axes: the number of next actions required to realize the outcome, and the reproducibility of the outcome if the same steps are followed:
    • Simple work is composed of fewer steps-per-outcome; following those steps always produces the same result.
    • Complicated work has more actions-per-outcome; following those steps generally produces the same result, but because there are more steps, it is easier for errors to occur.
    • Complex work has many actions-per-outcome; so many that interactions and dependencies between the actions become intrinsic to the work. Because of these interactions, it is rare that completing the same actions will produce the same result.
    • Tying your shoelaces is simple. Launching a rocket is complicated. Raising a child is (extremely) complex.
    • For a (reductive, abstract) knowledge work example: writing an email about something you’ve already understood/decided upon is simple. Making decisions about something for which the parameters are well-known is complicated. Generating novel solutions to unsolved/unanticipated problems is complex.

Cognitive capacity & cognitive load

  • Cognitive capacity is your ability to cognitively act on a bit of work.
    • We only have so much cognitive capacity.
    • Cognitive capacity is reduced by cognitive load. There are three types of cognitive load:
      • Intrinsic: whatever the inherent complexity of the cognitive action actually is;
      • Extrinsic: anything added to cognitive load by the system; and
      • Germane: load involving recognizing and using patterns of cognitive action.

Data quality

  • There are four measures of data quality based on conceptual modelling:
    • Completeness: the degree to which your system represents the total state of the work. If your system is incomplete, you’re missing information about your work. (e.g., you’ve captured only 50% of the projects you’re responsible for.)
    • Unambiguousness: the degree to which there’s a one-to-one relationship between your representation of the work and the work itself. If your system’s ambiguous, the information in it can be interpreted in multiple ways. (e.g., A task has a due date later than the due date for its project.)
    • Meaningfulness: the degree to which the system is easily mapped back to the real world. If your system is meaningless, then the information it contains is hard or impossible to interpret. (e.g., A task is described so vaguely that you don’t know what to do with it—think of writing “Mom” on your to-do list, when “Mom” represents “Plan mom’s birthday.”)
    • Correctness: the degree to which the system contains an accurate representation of the work (e.g., tasks that are left incomplete even though they’re finished.)

Tying it all together

In my opinion, the goal of productivity systems is to minimize intrinsic load by effectively representing your work—while adding as little extrinsic cognitive load as possible, and enabling effective use of germane cognitive load.

The more complex a piece of work, the more important it is to represent it effectively in a system. However, there’s a threshold: once the work is past a certain point of complexity, it’s harder to represent. E.g., complexity determines how easily the work can be represented in a given system.

If your system’s conceptual model does not have the ability to fit all the important or relevant things you want to represent in it, then you have to find workarounds or you will introduce data quality issues. (Some folks do quite impressive things with tags, for instance.) However, I find workarounds introduce their own data quality issues and add to cognitive load to boot.

Any issues in data quality will contribute to increases in cognitive load. Further, managing a system with a conceptual model that does not match your view of your work leads to additional extrinsic and germane cognitive load.

Similarly, displacing your representation of the work from the work itself increases cognitive load. For instance, many folks represent tasks about writing in a todo app while writing in a writing app. This means you have to keep information about the writing in mind while looking at your tasks and vice versa. This is easily resolved by having two windows next to each other, but this solution can’t always be used.

The takeaway for me

I am a researcher (PhD candidate, technically) working in three or so disparate fields, and I am a management consultant supporting governments and NGOs in complex systems change work. I generally work alone, albeit often in service of others. My work is usually in the upper-right of the graph above.

The non-app solution to that problem is to simplify the work. Find ways to define it more clearly, reducing the volatility/uncertainty/complexity/ambiguity of it. I think the enlightened among us do this. Sadly, I’m not very enlightened.

As a result, apps like OmniFocus—and combinations of apps, like OmniFocus and DEVONthink—do not let me represent all the information I want to represent where and how I want to represent it. Obsidian provides a free-form conceptual model—plain text and files—that lets me establish my own conceptual model, through text, (embedded) search, and dataview-based lists and tables.

It would still be nice to see clean lists of tasks in an app like OmniFocus for some things I do. To that end, I’m thinking about building an integration between OF and Obsi that treats certain Obsidian notes as projects, syncing task items between them. Not sure how doable it’ll be but it would be fun.


@ryanjamurphy this is fantastic. I read through it three times to make sure I captured the essence and I copied it to Craft for future reference. :slight_smile: I may quote you in my chapter on productivity! :slight_smile:

I’m not familiar with dataview-based lists in Obsidian. I genuinely like the idea of having everything in one place but I probably don’t have the skills needed to make this practical for as you said you are going to work on building an integration between OF and Obsi–that is beyond my technical skill level. You probably can’t share this due to privacy issues but it would be awesome to see how you have your system structured!

Thanks for starting this thread!

1 Like

Let me first say that I have a lot of respect for your work, your approach and all your contributions here, for Obsidian and elsewhere. However, I have a lot of issues with that model, in that it is not a whole picture. One of the big enlightening moments for me was when I realized that it does not scale, and rarely applies, to creative work. It’s just a different beast. Kourosh Dini has the best definition I’ve seen: it’s work that alters its own rules at it progresses.

You can’t write next actions for writing a novel. Or rather, they change so often that it’s not worth doing (and they are certainly never reproducible in fiction). It’s why it’s so mentally taxing and why writers are the kings and queens of task avoidance. (Task. Not tax.)

I believe, and I have found in my work, that there is an entirely different model of productivity pertaining to those projects that do not follow the model of the task but of emergence and constant clarification. Which is why I’ve become such a fan of the Zettelkasten method, which is the closest embodiment I’ve found of that reality.


Just a quick observation - OmniFocus does have “do” dates: I use “defer date” and “duration” to indicate when to start and how long a task will take.


I’ll be making that thing a plugin (if I can make it work at all), so others’ll be able to use it for sure.

See, this is where it gets fun, because my approach is trying to solve that same problem. The research and systemic design work I do is constantly evolving. So, a lot of my tasks look like this: - [ ] Continue writing the chapter on systems strategies here.

And I’ll drop that inline in the chapter, and pick it up with a special kind of search. Or, in tomorrow’s daily note, I’ll have “work on systems chapter” and then look for those inline todos to indicate where I should pick up.

The “number of actions-per-outcome” metric doesn’t mean those actions are well-defined. It might be “keep writing” x 100. But the “finish book” outcome can only be completed by a bunch of “keep writing” actions.

For sure, and that’s what I relied on while I was using OF. However: it is a workaround. Ontologically, a defer date represents how long you’re putting something off for—it doesn’t mean you necessarily intend on doing it on the date you select. I might defer a project to Monday, because I don’t need to be aware of it 'til then—but I might intend on doing the tasks in it on Monday, Wednesday, and Friday. Maybe this is nitpicky, but I think the two concepts are two different kinds of metadata.


First time I’ve seen ontological used in a non-metaphysical context, I like it. :grinning_face_with_smiling_eyes:


This is correct and one of the minor frustrations I have with OF. I use a defer date for both reasons–I don’t need to see it until x date or I intend to work on it beginning on x date. One way I have “solved” the lack of a do/start date is the Today tag. When that tag is added, it always shows up on today’s date and remains there, which means it must be done that day (which means it also has a due date for that date) or I begin to work on it, say on Monday, and work on it all week with a due date of Friday. It works but it is not elegant. Thing’s “when” date is better in this regard.

1 Like

learnt a new word today. Everyday, I learn something new visiting MPU


But the “finish book” outcome can only be completed by a bunch of “keep writing” actions.

Okay, this I do get behind: Dini has been advocating the idea of « sitting with the work » for years. However, I would argue that this still proves that the task list model is not enough to complete such a project. The intent to « keep working » only gets you so far (what do you do during the session itself?), beyond this there needs to be something else, and that’s where classical task management breaks down and the ideas of emergence take its place. You don’t write a novel with GTD. (I really wish I could. Would make my life a lot easier. You can still make a point - and I do - and the five GTD steps occur in the session, but they are so intertwined and you are changing modes so constantly that it can’t be tracked, and therefore you lose all the benefit that this kind of external cognition can bring.)

The creative work itself, I have found, does not build on a sequence of tasks but on a continuum of building clarity, which impacts a lot of continuously moving parts, and that can take years to build to get to the final product. It does require a concerted effort, but the idea of the concerted effort alone is not enough to get the thing done; another model is needed inside
the session itself. (All in order to make our lives easier of course, as always.)


I used GTD for many years, but over time I’ve become disillusioned by the method. Over the past year, I’ve been reading books by Cal Newport, specifically Deep Work and A World Without Email. I find his thinking better meshes with the world of knowledge work. He recommends doing time block planning. He has a website devoted to it: https://www.timeblockplanner.com

I basically divide my work into two types. 1) is the deep work that Newport talks about. The work that takes long blocks of concentrated time. 2) Are administrative tasks, send an email to Judy about Project X. I keep track of the Administrative tasks in Things. I do time block planning for the deep work to make sure I schedule enough time to do this type of work that is the most important. I find that the GTD system treats all Next Actions as the same and so it is easy to spend too much time on the 5 minute admin tasks and not enough time on the 2 hour deep work that really makes a difference.


I’m glad folks have systems that work for them.
I’ve tried both Obsidian and OmniFocus for task management, and have settled instead on NotePlan3. For me, it’s the Goldilocks middle-ground between the free-for-all that is Obsidian, and the button-down collar with tie that is OmniFocus.
I’ve augmented NP3 with Analog, which works well when I use it (there was recently some indecisiveness about whether to plan in morning or evening that derailed me for a while).
(Hat tip to @ChrisUpchurch for suggesting both.)

OT: Listening to Kourosh Dini’s Saturday morning music session now. Highly recommended.


Boy, I know I’m going to regret even looking at another program! But, I may depending on the answer to this question. Does Noteplan allow you to see and manage tasks by project in addition to the calendar/date view? I ask because the intro video shows this:


My projects are the folders to the left in the screenshot below. Within each are relevant notes. My Today note for yesterday is shown in the center pane. The items in the grey top area are items in notes that I tagged for yesterday. In the lower part of the center window, I’ve created some time blocks to work on those projects. They appear on the calendar timeline on the right, but aren’t actually entered into my calendars. My calendar entries are also shown (there’s one in light blue).
This video was helpful for me Stacey Roshan.


NotePlan is great. I believe it also has time blocking functions that integrate with your calendar, though I never did figure out how to work it.

You can also save searches as custom filters, and launch specific searches via the URL scheme.

NotePlan’s also got a burgeoning API for plugins coming.


I can. I apply GTD to everything I write in my personal and professional work. It sounds like you and I have different requirements for what defines a “next action,” and neither one is more valid than the other. It all boils down to what any one of us finds useful in making a draft of anything happen. Any writing that requires more work than I can accomplish in one brief sitting requires a next action. I couldn’t work any other way.

I’m intrigued then. What are your next actions for writing novels on a consistent basis?

I could of course write « keep working on chapter x » but frankly I know that and it’s not getting me any farther. What does is, sitting in front of yesterday’s pages, rereading them, adding here and removing there, getting back into the atmosphere and characters, inventorying what I’d like to move forward on, and then revving up the sentences as I start typing and build up steam. There are actions associated to all this of course, but what I’m getting at is, it makes no sense to consign the tasks as they come, because it’s about getting into flow, and flow occurs faster than next actions as you zip from one thing to the next and build up clarity.

I find when I am doing “creative” work such as a long writing project, or a photography project to build up an area of my portfolio, that techniques such as GTD, and any kind of task software are not useful. This is the sort of mental work that gets done first in the mind and then gets transferred to paper or image. Sure, there are lists of stuff involved – pack the kit with this and this and this; set alarm for 3:00 AM – but unless I keep the actions needed to do the work “registered” and focused in my mind, then the whole project falls apart.


I have none, because I don’t have that need.

I totally agree. Which is why I would never create actions for any of that, because it’s unnecessarily granular, just as I would not create a daily oral hygiene project that included an action that told me to “Apply bristles of brush to surface of teeth and move handle of brush in vigorous, repetitive motion.”

But that doesn’t mean I don’t have room for next actions in my writing life. I absolutely do. What I need next actions to do for me is to lay out in front of me all of the possible actions I could do next on anything of importance to me in my life and give each a description that provides enough context for me to decide whether I have some combination of the time, energy, interest, or need to choose that one to do next. That’s no different in my writing life than in any other area of my life. And writing competes with other parts of my life constantly, at work and in my personal life. Sometimes it wins and sometimes it loses. It loses less often when I’ve defined a next action that gives it a fair shot against everything else that has a next action.

I honestly do not understand. You apply GTD to your writing with precise next actions laid in front of you but you don’t write next actions because you don’t need them? A piece is missing for me here in how exactly this works for you in the writing itself.

Let me say that I do apply GTD in my writing life - as soon as something gets out of the pure writing mode then I’m all about next actions and projects. I even apply it to the revision process to a large extent (fix minor plot hole A, fix major thread B, remove tertiary character C and so on). And let me reiterate that I believe the 5 steps of the GTD workflow do apply to everything and everywhere. It’s just not an operating mode in my opinion when the flux of clarity is the most important thing to take care of.

John, I don’t want to take too much of your time but before I give this too hard a look, oh why am I even considering this! :-), I have another two questions.

  1. How do you view all of your projects and related tasks to stay on top of them? In other words is there a way to see a overview of all projects and tasks without going into every single project like one can in OF?

  1. How do you get emails into the system? I often bcc OF on an email to someone so that I capture a follow-up in OF.