My Latest Book

Product Details

Also available on,, and other Amazons and bookshops worldwide! 

To Think About . . .
Change leads to disappointment if it is not sustained. Transformation is sustained change, and it is achieved through practice. BKS Iyengar
My Other Books

Product Details

Product Details

Product Details

Product Details

Click to order other recommended books.

Find Us on Facebook Badge

Search This Site
Latest Comments

Discussion Forum > Best machine learning/artificial intelligence scheduler

I'm looking for a digital scheduling application that makes use of artificial intelligence. I have used machine learning at work, so I am very familiar with it and its ideal use cases, which could include scheduling. I imagine the current AI products have quite primitive algorithms, but which would be the best one for me to try?

About my schedule
- have certain daily or recurring tasks, eg I study one career-related topic 25 minutes/day, another one 5 minutes/day. I exercise QOD (every other day) for 12 minutes. I don't know the best time of the day for any of these
- would like to group tasks together, like errands, phone calls, home computer, and so on
- have probably between 100-200 next action items and many Projects
- would like the schedule to be able to take into account psychological readiness (Mark Forster came up with this, I think). So if the task it schedules is not right, it would learn (without knowing why I am not ready for the task potentially) and reschedule it to a more appropriate time
- I have a few rough deadlines. I want to study for and take a standardized test in the next 3 months, eg, but it could be longer than that
- I have literally thousands of e-mails to go through. some of these will create more next action items. I use thunderbird as my mail client
- I use google calendar, so nothing should be scheduled when a hard appt is already scheduled
- most of the tasks/my work is knowledge-based, without hard deadlines
- should be able to handle sequential tasks. so when one is completed, it could load the next one onto the list
- would like it to handle everything, both personal and work items (as in GTD)

Because I have so many next action items, I do not like choosing what to do (eg, what would make the most sense for me to do now?), so I have had difficulty with systems that require me to review all 200 items and selecting one, as it takes a long time to determine priority. And even reviewing a shorter list I tend to get overwhelmed. So I would like to have it assigned/scheduled.

Also while I have a general idea of the importance of different projects, I have not had success with methods focused on priority (eg, Tracy's ABC method), because it will frequently be the case that a high priority task (not urgent) may take 20 hours, and if I exclusively work on it (single handling approach), mundane things like e-mails and going through physical mail piles up, and I have unpaid bills and late fees.

I do not have a good sense of what % of time to spend on various tasks/areas, but a computer should be able to figure this out (if not now, certainly only because no one has applied what is known in machine learning to scheduling).

I have tried a plethora of methods in the past. The Linenberger approach didn't work for me as it is too focused on deadlines, and I have very few. I used GTD several times but never was able to keep up with it (and the difficulty is choosing from all the "next action" items).
February 14, 2017 at 21:14 | Unregistered Commenterzeloc
BTW, the ones I came across from google are todoist smart schedule and skedpal, but I have not tried either and don't know if there are others. Ideally it would work on linux.
February 14, 2017 at 21:17 | Unregistered Commenterzeloc
I wouldn't try automating this at all. You'd spend more time entering and adjusting than actually doing. (I know this from experience with a much smaller list of things to do.)

It sounds like you have way too much on your plate. You'll be happier with your progress if you focus on just a few projects at a time. Put everything related to the other projects in a safe place. Don't even organize it, just get them confined in a safe place.
February 15, 2017 at 2:44 | Registered CommenterCricket
I need some way to organize everything, as I am losing track. When I neglect one area for too long, it builds up and I begin to get mental distractions because I don't have an effective way to capture everything - it is in many different places now. David Allen describes this problem well in GTD.

I did some more research and in addition to the above 2, I am considering Autofocus.

Some questions on AF. 1. Which one do I try? I found 2 different descriptions as followed, or is it on another page?

2. In different descriptions of autofocus, it is described as both a complete system (excluding calendar appointments), and as something to be scheduled on a scheduler. Eg, if I know I want to study physics for 25 minutes/day and study English vocabulary for 5 minutes/day and AF tasks for 6 hours, do these 3 go on a planning app, or do I put physics and english vocabulary on my auto-focus list? I am thinking about this in the context of habits. I am now using a spreadsheet and working on 3 habits, with the intent to add more when they are strongly engrained. I am wondering if I should continue doing that, or can I simplify by writing down all habits (without limiting it to 3) on my AF list, they will recur when they are checked off on the next page, and so over time they will all be incorporated and I do not have to worry about "forgetting" my list of habits. Even if they are not done daily at first, they should become daily with time. Is this a recommended approach? I found maybe one reference to using AF for habits and some who used external solutions.

For those of you using autofocus, how long have you been using it exclusively, or did you go back and forth with various methods, and then did you settle on AF or something else?
February 15, 2017 at 6:59 | Unregistered Commenterzeloc
Get a binder and use it as a capture system. (Binder, so you can tear pages out of one book, or photocopy them, and put them in the capture system.)

For now, if it's easy to put a task beside others in the same project or context, do so. Otherwise,, just get it down. Out of your head and onto paper.

The description on Mark's site is more authoritative. He has control over it.

What to schedule vs what to leave on the list varies with person, and, for me, type of day. If I have a lot of appointments, I'll schedule my other tasks. Otherwise, I'll do something else between appointments. On days that are looser, I won't.

Stop fussing over the right way to do it. The right way is the one that works for you today. Try one and see if it works. The only way to learn that is by experimenting. Experimenting means you might learn that the way you tried doesn't work. (Put in a safety net so you get the truly critical stuff done, and then test a system and learn from those tests.)

I usually put my daily habits on a single page within my big list, and stay on that page for a while. The risk of that is spending the entire day on that page, working on my habits, and not getting to tasks on other pages. (It's great if I'm trying to build habits, not great if I'm trying to build so many habits that there's no time for other things.)

I go back and forth between systems. I keep thinking AF1 with a sticky-note for habits and another for urgent tasks would be best, but never actually get around to trying it.
February 15, 2017 at 17:59 | Registered CommenterCricket
Thanks for the input. I just spent a lot more time reading about all the various methods that MF has come up with it, and many threads on the forum, as well as other articles. I didn't realize there was so much content (4 versions of AF) and so many people here are using different methods, I expected everyone would converge on one. I think I was active here before FV and I never got into AF. I am going to try FV on paper and keep track of habits separately as you suggested, as well as the daily/recurring. Since it uses Colley's rule, it should have some level of optimality. At some point, I can decide if I want to go electronic in order to categorize similar tasks. If this doesn't work I can consider skedpal or another solution.
February 15, 2017 at 19:06 | Unregistered Commenterzeloc
I'd start with the question, "Would bad things happen if I didn't do it this year / month / week / half-week / day? How bad?" If over a month is fine, clear it off your list for now. That will shorten your list drastically. For habits, "if I don't develop this habit?" For little-and-often projects, "if I do this less often than planned?"

For how bad, catastrophize. Really push your imagination, and see how bad it could get. Fail your degree; lose your job, spouse, home; high blood pressure; nasty depression spiral; ignore an email conversation that the other party lost interest in months ago or continued it in another way.

("Will good things happen if..." doesn't work when you're overloaded. Every task leads to the possibility of great things.)

(Also catastrophize the other way. If I do this, might bad things happen. Less time for more important and urgent things. More stress. Energy-sucking conversation.)

No matter what you do, though, you will have to make some hard decisions. You might decide to give each project equal time, and not do well on any of them. Most of us find choosing a few and being happy with our work on them works best. (That assumes we can get past the fear of missing out on the other projects, and don't reactivate them. That's a bad habit I have. When I feel like things are under control, or I"m bored, or procrastinating, I'll re-activate a project, with predictable results.)

Best of luck!
February 15, 2017 at 22:58 | Registered CommenterCricket
Zeloc, your situation and habits sound similar to mine: many projects, many tasks, a tendency to generate backlogs, and always trying to find a way to manage it all better. And also a tendency to get absorbed in a few important but time-consuming tasks and neglect everything else in the meantime!

With these habits and tendencies, it is very easy to overwhelm Autofocus (all versions), SuperFocus, DIT, DWM, FV, and FVP (I have used them all extensively!) The lists grow very long, and tend to get slower and slower, ultimately creating resistance to using the system itself, only to find relief by changing to a new system.

But that relief is an illusion, because it's mostly the result of starting from scratch with a new, short list, and not the result of the latest-greatest algorithm. The algorithms can be very helpful, but they all have limits as to how much volume they can handle. Mark detailed the problems in several posts, and when you tend to generate a high volume of tasks and projects, all of these problems go to an extreme level.

The only systems here that I have found to be sustainable are the no-list systems. They don't provide an overall comprehensive personal management system -- but neither do any of the other systems here. But the wonderful thing about the no-list systems is that they prompt you to develop new supporting habits and systems, while keeping you engaged in your most critical, time-sensitive, and important work, and keeping you focused by limiting your work-in-progress.

I wrote a summary of the way it works for me here:

Mark's avowed favorite no-list system is The Next Hour Of Your Life:

Lots of great articles and insights on no-list here:

It's easy to try different variants and see what works for you.

My own approach doesn't really have any rules -- I write things down on a whiteboard or scraps of paper, and erase/discard frequently. That's about it.

The key for me is that writing down my no-list is a lot more like a "thinking tool" than a "to-do list". It just helps me think. Catch-all lists (like Autofocus) can also help me think, but once they reach a certain size, they tend to get in the way of clear thinking and just bog me down.

Daily routines and habits also tend to emerge in the course of using no-list, as Mark has written extensively. My own workday routine almost always follows something like this:
1. Attend meetings and appointments as needed. (But try to avoid them if you can!)
2. Clear out emails from yesterday and any tasks/reminders due today
3. Focus on projects, thinking, whatever seems important.

The need for good supporting systems also emerges. For example, if you do a lot of projects, you will find yourself needing to prioritize the creation of good project systems.
February 16, 2017 at 7:01 | Registered CommenterSeraphim
Thanks for the response. I recognize your name as one of the regulars when I was here years ago.

Am just on the 2nd day of using now FVP (read about FV vs. FVP this AM), but I will keep in mind the possibility that the system can get overwhelmed. For now I am going to include any projects that come to mind on my master list and see what happens. I can't see myself using a no-list method at the moment, because I really want to capture everything, but will be open to it.
February 16, 2017 at 20:12 | Unregistered Commenterzeloc
If you're going the master list route, you don't have to list each and every sub-task on that list. "Do next part of project X" is fine keeps the master list much shorter.

"Spend 30 minutes on the room that needs it most that day, in the corner of that room that needs it most, ideally 3 x / week" works better than listing each and every room and corner and then trying to prioritize and schedule. (That's for routine maintenance. Cleaning for a guest would get its own line just to remind me. That also gives it two chances each pass: once as room that needs it, and once for party prep.)

I found FV the most dangerous of all the systems. The first things on my master list were old projects, that I'd put off because they were much less important than the rest. There they were, at the root of every chain, along with the naive belief that the system would automatically make me pick important things. I ended up re-activating way too many of those projects, at the expense of some critical work.

FV didn't make me do that, but it allowed it. In hindsight, when I reached those lines, I should have said, "Currently not important. Rewrite at end of active list for future consideration, or, even better, get them off the active list and onto a tickler or, even better yet, the someday/maybe list." (With AF, I skipped the rule about having to do something on each page or deleting everything, so I didn't feel the pressure to do the old projects.)

Check out the spinning plates method for deciding how many projects you can actually carry.

It's very reactive, and has you pick up or put down plates as your status changes. Even though I had too many plates going, my time went into the important ones, and the others were eventually put down without breaking.
February 16, 2017 at 21:56 | Registered CommenterCricket
Hi zeloc,

I can really relate to your situation, I'm the same type where I have a bunch of projects and lists as well of things just to remember for later, and I can't afford combing through the list to find my next task. I see you've already found a new method you're experimenting with so this might not be relevant anymore, but I have a method I've been using for 7+ years now that's based on how a product manager prioritizes his backlog, and it also incorporates automation to make choosing the next task efficient.

I set impact/effort/urgency values for each task to determine it's priority and it's trajectory over time. Once I do this the system more or less takes care of itself, and the most 'valuable' tasks will bubble to the top. It can of course be overwhelmed, but I have a few safeguards built in to avoid these situations (and working on other safeguards constantly).

I never need to schedule all my activities into a calendar since whenever I have time I just take an item from the top of the list (around the top three items), but I can also set due dates and then the item will surely jump to the top before it's due.

I used an Excel sheet with a lot of VBA macros for years personally with much success, and since I didn't see such a working system in any other tool I made it into an iOS app to share it with others.

You can find more info at, or feel free to message me if you're interested, I'm happy to describe it more and maybe even show you the excel sheet.

Hope you find this useful, and good luck with whichever method you choose to go with!
February 16, 2017 at 23:45 | Unregistered Commenternzen
nzen --

FlowTree looks really nice!

This reminds me of some experiments I did with "Weighted Cost of Delay" as a sorting technique (a concept borrowed from Reinertsen's Principles of Product Development Flow, which is central to many Lean product development books and concepts).

Cost of Delay looks at the value per week of having some solution -- works best if it's quantified with actual dollar amounts or best guesses. E.g., my solution will help us deliver 10% more product every week, which is worth $N per week. So every week my solution is delayed costs the company $N.

"Weighted" cost of delay also considers the duration of effort required to implement the solution. Let's say we have Solution A that will take 10 weeks to deliver and will earn the company $10,000 per week. And we have Solution B that will take 20 weeks to deliver and will earn the company $50,000 per week. Which one should we do first? Or should we do both at the same time?

Cost of delay is a great way of prioritizing these because it can actually answer that question. :-)

You can figure this out quickly by following a few simple rules:
1. Don’t do multiple projects simultaneously. It drastically increases the total cost of delay.
2. Assess weighted cost of delay for each project: (cost per week) / (duration)
3. Work on the highest weighted-CoD items first.

Solution A: CoD / Duration = ($10,000 per week) / (10 weeks) = 1000
Solution B: CoD / Duration = ($50,000 per week) / (20 weeks) = 2500

Solution B is higher, so do it first.

Here's a more detailed walk-through:

Let's say we do A then B:
Solution A is delivered after 10 weeks. Cost of delay for Solution A is 10*$10K = $100,000
Solution B is delivered after 30 weeks. Cost of delay for Solution B is 30*$50K = $1,500,000
Total cost of delay = $1,600,000

Let's say we do B then A:
Solution B is delivered after 20 weeks. Cost of delay for Solution B is 20*$50K = $1,000,000
Solution A is delivered after 30 weeks. Cost of delay for Solution A is 30*$10K = $300,000
Total cost of delay = $1,300,000

Let's say we do A and B simultaneously, with our time split 50/50 between the two projects during the duration of each (assuming no delay due to task switching):
Solution A is delivered after 20 weeks (2*10 weeks), CoD = 20*$10K = $200,000
Solution B is delivered after 30 weeks (20 weeks during which we got 10 weeks of work done on Solution B while we split time with Solution A, followed by another 10 weeks of focused work on Solution B) = 30*$50K = $1,500,000
Total cost of delay = $1,700,000
If you include delays due to task switching, this simultaneous approach is even more expensive.

Thus the fastest path to the highest value is first do B, then do A.

More here:

I tried to simplify this for use with tasks and projects (actually, user stories and "features"), using "impact" and "size" as proxies for "cost" and "duration". But it's too hard to quantify these things for small tasks and projects, so I used T-shirt sizes (which is what your app does!), and then translated those into Fibonacci numbers (1, 2, 3, 5, 8, 13, etc.) and then did the division (impact / size) to get a rough priority. (Reinertsen doesn't approve this non-quantified approach, but it worked for me!)

That division was hard to do in my head, and wasn't practical. But then I realized that the Fib series are really just an exponential series of G^N where G=1.618… (the golden ratio).

So if my T-shirt sizes are XS, S, M, L, XL, this translated to 1, 2, 3, 5, 8, which translated to G^0, G^1, G^2, G^3, G^4. Now the division is really easy! You just subtract the exponents! And then do the items with the highest value first.

Priority = (impact) - (size of work)

This made the process super easy, and didn't even require absolute assessments, just RELATIVE assessments.

Example, take 10 features, then (1) sort them into groups by relative impact, and number each group 1, 2, 3, 4, 5... Then (2) sort them into groups by relative effort, and number each group 1, 2, 3, 4, 5… Then *subtract* the effort number from the impact number to get the priority number.

This is a very quick & dirty way to sort, but it's extremely effective, almost always giving a sort order that's significantly better than raw intuition and always better than sorting by "importance".

Anyway, this is a very long way to say, it seems like you are doing something similar here with your T-shirt sizing. You also include urgency. Maybe it could work like this: (impact)*(urgency)/(size). Or if you are using Fib exponents, it could be (impact + urgency) - (size).

Does your app work something like that?
February 17, 2017 at 1:27 | Registered CommenterSeraphim
I do not have an iphone so cannot try the app. But thx for the idea. I have used similar multi-factor rankings in the past.

Thx for the post about the projects. I really don't want to limit myself because I like being able to write everything down as I think of it; I am hoping that FVP will allow the priorities to appropriately sort themselves out. In any case, I will be giving FVP a good attempt without ANY modifications for the time being.
February 17, 2017 at 17:27 | Registered CommenterZeloc

"Then *subtract* the effort number from the impact number to get the priority number.

"This is a very quick & dirty way to sort, but it's extremely effective, almost always giving a sort order that's significantly better than raw intuition and always better than sorting by 'importance.'"

That sounds great! It took me two reads to grasp it, but I am quite intrigued. I would never want to do that with every item on a catch-all list, but at a project level it solves a lot of dilemmas.

I've finally discovered a good way to store future projects ("Someday/Maybe") that don't belong on the current list, a way of queueing them up while still clearly showing each project thread, so I don't have to rethink what to do next (I'm calling it an "analog kanban" and maybe I'll write it up later). I think I could use your calculation on these future projects to get them queued properly and reduce the anxiety over those decisions.

BTW, the Someday/Maybe/Analog kanban has solved my problem of Teh Overwhelmed List, because the clutter was all those future items that I didn't want to lose track of by dismissing or deleting. Neither of those works for my thinking style, but storing them on a *properly sorted* future list works wonders to get them off my mind while keeping them marching forward. And it turns out that "properly sorted" means an analog kanban!
February 17, 2017 at 19:15 | Unregistered CommenterBernie

Awesome comments there, your depth of knowledge is really impressive!

You are correct, what FlowTree does is very close to what you describe, with a few minor differences:

1. a base priority is determined by division (impact/effort), not subtraction, since it's done by the computer, so it's not a hassle. But I'll run some tests with your suggestion to subtract, I'm interested to see what difference it produces.

2. I tailored the effect of urgency to be personal task specific (different from projects in business), and added the capability to actually decrease overall priority if the urgency is small. So an urgency set at 'XS' will cause the task to slowly sink to the bottom of the list. An urgency of 'S' will give 0 weight to the urgency (static priority). With the latter, it is possible to create static lists that I don't want to forget but will never need to do in a 'task' way (e.g. interesting books to read, etc). The former handles the decreasing relevancy of certain tasks (e.g. an article on the current state of XY won't be relevant in a year, so it should sink to the bottom in case I don't get to it by then).

3. You can also set a due date for any task, which will cause it to start rising with greater velocity as the deadline approaches. Once the deadline has passed, the task returns to it's original trajectory.This ensures that I notice any tasks that are due soon, acting as a reminder without a calendar. I find this much more flexible than to 'hardwire' certain tasks into a calendar schedule.

4. Further tweaks are also possible and planned to the formula - I want to add modifiers that act based on the users actions regarding the task (e.g. if it's at the top but avoided by the user for some time, it's importance should decrease slightly)
February 17, 2017 at 20:15 | Unregistered Commenternzen

"1. a base priority is determined by division (impact/effort), not subtraction, since it's done by the computer, so it's not a hassle. But I'll run some tests with your suggestion to subtract, I'm interested to see what difference it produces."

If I understand correctly, Seraphim is suggesting subtracting the exponents, not the values themselves. You would have to take the log of your impact and effort quantities before subtracting, which would give you the exponent of exactly what you'd have got by dividing them. Seraphim is simply operating in "exponent land" to make the math easier to do mentally.

"3. You can also set a due date for any task, which will cause it to start rising with greater velocity as the deadline approaches. Once the deadline has passed, the task returns to it's original trajectory.This ensures that I notice any tasks that are due soon, acting as a reminder without a calendar. I find this much more flexible than to 'hardwire' certain tasks into a calendar schedule."

Required tasks with a hard deadline have an infinite cost if the deadline is missed (the hardest definition of "required") but a zero gain/cost if done early. E.g., if you have to write a report due in two weeks, which will take you one week of effort (with appropriate buffer for uncertainty), and you will be fired if you fail, then the impact is zero all this week and becomes infinite at the beginning of next week (assuming that being fired is infinitely bad). If you were to do it this week (early), you would merely displace to next week other projects which have a non-zero impact, which chalks them up as a loss for this week. This is only the "last responsible moment" doctrine restated.

To rank various deadline-oriented tasks and be more realistic about their consequences, you can give them various very large impacts rather than infinite.
February 17, 2017 at 20:56 | Unregistered CommenterBernie

Thanks for the comments!

1. You're right, sorry for missing that.

3. Your point is definitely true in case of 'hardest definition' deadlines, thanks for pointing that out. These should probably be handled as special cases to allow for such situations and avoid catastrophe. :-)
In my experience though, most deadlines we use in our plans aren't so hard. For example, paying my hydro bill by next Tuesday. I can do it any time really up until then, and even if I happen to spill over next Tuesday's deadline, it's not the end of the world, I'll just have to pay a surcharge. I still want to avoid it though, if at all possible, so the reminder is important.

Oh, and it's also good to factor in the effort value here in a new sense - the bigger effort a deadline-task takes, the sooner it's priority should start moving upwards as the deadline approaches.
February 17, 2017 at 21:21 | Unregistered Commenternzen

I am still wrapping my head around this approach, especially the time-variance of these quantities. Seraphim's URL is very helpful:

It is the root of a series of lessons. I'm finding it makes the most sense to follow them in order listed in the pulldown menu "Cost of Delay" up top, rather than following the links inside the text.
February 17, 2017 at 22:05 | Unregistered CommenterBernie

This page ( ) finishes with a discussion of items that have a deadline and some quantified examples. Very interesting, addresses the points you and I have been making.
February 17, 2017 at 22:18 | Unregistered CommenterBernie
I haven't been paying much attention to this thread because my mind has been on other things recently (like staying alive), but the mentions of Black Swans in Seraphim's posts caught my eye. Recently I've been thinking a lot about how to make time management more antifragile and wondering whether we (ok, I mean I) have got hold of the wrong end of the stick all these years.

I've come to quite a few conclusions as a result, which I hope to write up soon.
February 18, 2017 at 12:59 | Registered CommenterMark Forster
There's a chapter in Nassim Nicholas Taleb's book Antifragile in which he wonders how projects like the Great Exhibition's Crystal Palace (1851) or the Empire State Building (1931) were built without any time overruns.

One of his conclusions: "They didn't have computers".
February 18, 2017 at 13:06 | Registered CommenterMark Forster

This topic is fascinating, the links you (and Seraphim) shared are great. I can't wait to experiment a bit with trying to port the scenarios listed there to the 'personal task' universe and see if I can find any changes or further urgency profiles in that world.


I am new to the concept of Antifragile but it seems like an intriguing topic. I'm not sure I understand what you mean by making time management more antifragile, but it brought something to my mind again that I've been pondering recently, namely the merits and drawbacks of goal setting (some possible drawbacks being tunnel vision, the demotivating effect of not reaching one's goals and the difficulty of doing so stemming from our inherent inability to provide realistic estimates as humans). I've long been a fan of serendipity and allowing enough freedom in our plans to remain open to opportunities while maintaining our vision. From reading your comment and the book's synopsis, I gather 'antifragile' is somewhere close to this train of that a correct assumption?
February 19, 2017 at 0:22 | Unregistered Commenternzen

<< I've long been a fan of serendipity and allowing enough freedom in our plans to remain open to opportunities while maintaining our vision. From reading your comment and the book's synopsis, I gather 'antifragile' is somewhere close to this train of that a correct assumption? >>

On those lines, yes. Though I'm still thinking it through myself.
February 19, 2017 at 12:18 | Registered CommenterMark Forster
Bernie - I'd be very interested to see your "analog kanban" idea. I've been thinking about different ways of managing someday/maybe kinds of stuff, and I feel like I need some fresh ideas.

Mark - I will be very interested to read more about your thoughts on antifragility and its application to time management!
February 20, 2017 at 22:08 | Registered CommenterSeraphim

I've described the Analog Kanban in a new thread: .
February 22, 2017 at 12:25 | Unregistered CommenterBernie
ToodleDo has an importance calculation.

I've been using it for about a week. If one sort method (due date, priority, calculated importance) doesn't inspire me, I use a different method.

At first, I had to play with priorities to get things in a good-enough order, but within a few days that was down to one or two changes a day.

I've been making great progress on the higher importance projects, but in a few days I'll probably start to worry about what might be lurking at lower levels. At that time, it will be easy enough to sort by due date.

It also has a start date field. I haven't used it, but I'm considering the pros and cons of using that to track the age of my backlogged items rather than due date.


Posted: May 20, 2008 Score: 1 Quote Reference
Importance is defined by a relatively simple equation. We had been keeping it as a trade-secret, but since it can be reversed engineered pretty easily, we might as well disclose it here:

Importance = 2+P+S+D

S=is it starred (0 or 1)
D=0 if due-date is non-existent or further than 14 days out, 1 if due-date is between 7 and 14 days out, 2 if due-date is between 2 and 7 days out, 3 if due-date is tomorrow, 5 if due-date is today, and 6 if overdue
March 1, 2017 at 0:24 | Registered CommenterCricket

In the importance calculation, as someone who's well acquainted with ToodleDo (which I'm not), could you explain a few things in the importance calculation:

- What is the point of the "2" at the beginning?

- How many levels of priority are there?

- What are the criteria for starring, and how is it different from priority?

Many thanks!
March 1, 2017 at 9:37 | Registered CommenterMark Forster
The "2" prevents negative results.

Priority has five levels: -1, 0, 1, 2, 3, called negative, low, medium, high, top.

Importance ranges from 0 through 13.

Starring (as with everything else) is up to the user. You can also filter so that only starred items show. (You can filter or sort any way you like. Three levels of sorting on the web version, but only one on Android.)

It's not a perfect system. Nothing is! After a week or so, I know for certain that I'll always have to adjust values every few days to keep the top tasks relevant. Fortunately, it fails safely, in that it's more likely to say a task is more important than less. A longer list for me to pick from, but it's complete.

It's easier to adjust a few fields than rewrite or add or erase dots, and it does a reasonable job with recurring tasks.

I use the free version, and don't use subtasks. I entered contexts and folders, but haven't actually sorted by them. Tags are useful for delegating. Very easy to print the honey-do list.

It's still in the honeymoon stage, and I'm on a roll with my scariest project, but so far happy enough. We'll see what it's like after the Lenten challenge.

Even better, if I stop using it, it's very easy to print the list and go back to paper. (Not so easy the other way around.) Without that escape route, I would not have tried it.
March 1, 2017 at 16:28 | Registered CommenterCricket
Thinking more. Fails safely might be the way I use it. If in doubt, rank it higher, so I review it again. It's better to look at something and say Not Now than to forget it exists.
March 1, 2017 at 16:29 | Registered CommenterCricket

Thanks for the explanation.

I can see that it gives a reasonable heuristic, which is probably more practically useful than a more sophisticated calculation would be.
March 1, 2017 at 19:24 | Registered CommenterMark Forster
Seraphim, nzen, and Mark,

Regarding the "Priority = Impact / Effort" concept, I'm thinking of a way to use FVP for Seraphim's sorting method:

"Example, take 10 features, then (1) sort them into groups by relative impact, and number each group 1, 2, 3, 4, 5... Then (2) sort them into groups by relative effort, and number each group 1, 2, 3, 4, 5… "

FVP's specialty is sorting your list, through repeatedly asking The Question. If The Question were "What has more Impact than X?" then you would end up with a dotted chain of high-impact items. This is similar to Seraphim's first sort, although it does not identify a contiguous batch of top-rated items, since FVP will not discover some of those until certain items are crossed off and the next partial scan takes place.

The trouble with this dotted chain is it will change quickly as urgencies shift, and that will make the rest of this concept difficult. So consider the suggestion at that BlackSwanFarming site:

Impact = Value * Urgency,
which leads to Priority = Value * Urgency / Effort

Well, Value won't be changing much for most items. Neither will Effort. It is Urgency that keeps shifting and causing us trouble.

So, suppose we do an FVP scan with a Value question: "What has more Value than X?" You have to adjust your thinking to consider the sort of value that does not express urgency. Imagine that both X and Y were to get instantly done by snapping your fingers right now, and just gauge which of them would create more value. Let's write a "V" next to these items, instead of dotting them, and let's call this chain of "V" items the "Value Chain."

This Value Chain gives us a piece of Seraphim's first sort, except since we have ignored urgency, it will not go stale very quickly. The V's should be able to sit for a long time, and we won't need to make another full V scan for a while.

Next, let's build a completely separate Urgency Chain. Make one more full FVP scan, from the top, asking "What is more Urgent than X?" and dot the resulting items. Ignore the "V" markings. Some dots will fall on "V" items, and some won't. Having to make a second scan may be a drag, but remember that the Value scan will rarely need to be redone. Most of the time, this Urgency scan is our main scan.

Now, imagine that all unmarked items are invisible, and you can only see the items in the Value and Urgency Chains. Do one last FVP scan on only those marked items, using an Effort Question: "What requires LESS effort than X?" Mark the resulting items with an "E." You will only be adding E's to items that already have a V or a dot or both. Hopefully this scan will not feel too onerous, since we'll be skipping over most of the list, including any V's that already have E's from a prior scan.

Now, we can find our highest priority items by starting at the bottom and looking for combinations of V, E, and dot:

1. The first dots from the bottom are the screaming-urgent ones. Their Urgency figure is so high that it dwarfs any V or E consideration. I think they will be obvious at a glance. So there is a sort of Stage I in which we do the bottom-most dots that scream out, regardless of V and E. Each completed dot calls for another partial FVP scan below the completed item, so we'll keep doing that using the Urgency question and dotting the results, and working on those screaming-urgent dots.

2. Once the high pressure is off, and there is a feeling of freedom to wonder what to pick next, then any remaining items with V, E, and a dot (all three markings) are the highest priority. As before, each completed item will trigger a partial FVP scan of the lower part of the list. I guess this would require a Value scan and a separate Urgency scan of the same lower part. Then we'll scan those new V's and dots with the Less-Effort question to mark any new E's. So a bit of extra scanning here, unfortunately.

3. The E-dots and V-dots come next,

4. Then the EV's and just-dots,

5. We'll be lucky to get this far down, but if so, then just-V's and just-E's finish it out.

Items 3-5 also require partial rescans after completing each item. The type of marking done determines which question(s) need to be rescanned (right?). E.g., after finishing an E-dot, we would rescan for Urgency but not Value, and then scan the new Urgent (dotted) items with the Less-Effort Question.

I may try this—I'm very curious but a little intimidated. I liked the test I tried of Seraphim's impact/effort method on a spreadsheet, but of course it went stale quickly, and it was only practical for projects, not an unfiltered catch-all list. As Seraphim had said, it showed me a fresh ordering that was a little off from my intuition, but I immediately recognized it as a better plan. If this multi-scan method does the same, I might not mind the extra work.
March 12, 2017 at 21:48 | Unregistered CommenterBernie
Similarly, FVP can sort us an Eisenhower Matrix using the two questions "What is more Important than X?" and "What is more Urgent than X?"

The Important markings would not change much, and hopefully the Urgent markings would disappear quickly—they're urgent!
March 12, 2017 at 21:55 | Unregistered CommenterBernie
... and I made on error in #5:

"5. We'll be lucky to get this far down, but if so, then just-V's and just-E's finish it out."

There are no just-E's, because we only marked E's when scanning the V's and dots.
March 12, 2017 at 22:02 | Unregistered CommenterBernie
Very interesting, Bernie.

I was never able to get this kind of thing to work at a task level, or even a user-story level. Whenever things change (things added or deleted or updated), you have to start all over again. It works OK for feature-level. It seems to work OK for

Maybe it would work at the task level if it's automated (like with nzen's Flowtree app). Whenever you get new tasks, the app could switch into Planning mode and ask enough comparison questions to get a good read on a reasonable ordering, and then switch into Do mode. That might work. But doing it manually is hard on the brain. :-) It's hard doing this three-pass method, but even harder needing to redo things whenever you add new items to the list.

Also, for tasks and small projects, I think it may be trying to manage at a finer level of detail than the statistical noise. The value of the sorting is less than the time it takes to perform the sort.

For example, you wrote:
<< The first dots from the bottom are the screaming-urgent ones. Their Urgency figure is so high that it dwarfs any V or E consideration. I think they will be obvious at a glance. >>

I'm guessing these would be almost as obvious with no algorithm at all! So why not just do them first, without any algorithmic sorting or scanning at all! Get them out of the way, and then proceed with the algorithmic scanning. Might get results faster, with less mental overhead.

Currently, for my DIT list, I've been experimenting with "do the smallest task first". The dynamic is a little different, since I need to get the whole list done -- it's not a matter of filtering to see what's highest value. It's just a matter of getting it all done as quickly as I can.

Actually, maybe this DIT example is just a special case of CoD sorting. It all needs to be done, so you could say it's all equal value. And it all needs to be done TODAY, so you could say it's all equal urgency. So the only factor remaining is size. So the right way to get value fastest is to do smallest tasks first.

"Smallest first" gives many other benefits. Number of items in WIP is reduced more quickly. Momentum (getting things completed) starts quickly and builds quickly.
March 13, 2017 at 2:23 | Unregistered CommenterSeraphim

"I'm guessing these would be almost as obvious with no algorithm at all!"

I agree. These items don't usually get onto my list, but I covered them for completeness.

As for the constant updating, that is what FVP's process specializes in: keeping a growing list sorted. Or should we say, sorta sorted? Because it accomplishes this by giving us that partial sort, leaving mid-level items stranded in between the dotted peaks. With each completion, we rescan the tail of the list and the sorta-sort is up to date, new tasks and all. None of the prior dots need to be changed, still valid. That's why I see it having some potential.

I didn't try it today, because I had a bunch of stuff that I just needed to do today. I often feel lost for priorities, though, when nothing is too urgent. Pure "standing out" always sounds good, subconscious mind and all, but it too often takes me into the weeds. So next time I may resort to "multi-dimensional FVP" in desperation. Perhaps it's "VUE FVP" (Value, Urgency, Effort--and my favorite mind mapper)?

I'll report on it if I do.
March 13, 2017 at 5:35 | Unregistered CommenterBernie

<< Pure "standing out" always sounds good, subconscious mind and all, but it too often takes me into the weeds. >>

This is a skill which needs practice like any other.
March 13, 2017 at 11:17 | Registered CommenterMark Forster
"The V's should be able to sit for a long time, and we won't need to make another full V scan for a while."

That feels like where I found myself going with Halves with Multiple Time Frames, but you described it much better.

Looking at impact instead of importance is interesting. If it has no impact, then why do I think it's important? (This is a very good question to ask of my very long reading list. What impact will reading or not reading it have?)

Instead of FVP and Colley's Rule, maybe give each line a score for each dimension. Very roughly, so it's fast. Say you use a score out of 5. Give the numbers quickly. Anything under 3 doesn't even need a number until the higher ones get done. When it's time to do the 4's, do a quick pass of the 3's and 4's, just to see if any can / should be demoted / promoted.
March 13, 2017 at 20:25 | Registered CommenterCricket
I would add from my experiments the more a list is structured or organised the more it is difficult to act and do things. A single list on paper (master list) is for me perfect for doing things. GTD is a perfect example about this. My aim is not to structure stuff. I am paid for acting ie buniness making. that's said I structure some project ie the most crucial one.

I also noticed something there is a real psychological effect about things with me. Often when I am really engaged on something I dont need any list. I just do things (which are often written on my list) Then later I will clean the list and work on it.

I re discovered this way of working on a master list when I spoke about jane wesman method and on this forum. We usualy take to much time for organizing stuff or making system we would like beeing perfect. But life is complicated and surprising. I indeed think we ought to believe and trust the way we do things and stop taking false reason for not doing them.

The core system is colecting and acting first on what is worth to do and make plan or planification for very little things which are often what we are reluctant to do.

I have compleately change my point of view since a while. I work with habits, check list, and only but one list which is often full of crossings and notes and sometime a single paper list were I focus detailled actions about one thing to do at a present time which will be trash after. It is just for me a reminder.
March 15, 2017 at 6:13 | Unregistered CommenterJupiter