Returning to Omnifocus after trying auto schedules like Skedpal and Why

ivanjay205

Registered
I do periodically ask questions and always love the plethora of people willing to jump in and provide their thoughts. Often I see some interesting conversation start as a result. There is no question here and I felt I would share my experiences with the group just for conversation and to see others opinions....

For those that might remember I have a few posts awhile back about exploring AI auto schedules like motion and Skedpal in an attempt to better manage things a bit. It was a great exercise, one that probably cost me a lot of time on setting up systems to change them, but one that really made me appreciate the roots of GTD a lot more.

To provide some context....I have been practicing GTD to the best of my ability for a very long time. I have a family with a wife, kids, and pets, a house and all the responsibilities that come with them all! And I own a business with lots of clients, 40 team members, conferences , meetings, strategies, financial decisions, etc. Just a little responsibility there too :) I found with GTD the problem I was having was my committment was so large that I would get brain fog when looking at my next actions list. Rather than pairing them down and moving lesser priorities to the someday only to have to mentally juggle what is the most important and bring them back I turned to looking for modern day computing to help me out. Seeing lots of facebook ads, and admittedly I LOVE tinkering with new technology.... Time to give it a go.

I turned to Motion first. At very first interaction (and I mean day1-3 type of thing) Motion was COOL. It was amazing. My schedule CONSTANTLY changes in the week and motion was seamlessly looking at my next action list, my calendar, and moving things in and out via priorities. VERY VERY cool. I was in love and said this is it! However, we got through week 1 and it was time for my weekly review..... Well, this is where things fell apart quickly. There was no structure, no real way to organize at the time (this since had an update adding some features in this area) and I felt loss.... I quickly realized that everything we are afraid of with computer AI was coming true. I had no clue what was on my radar, what I had committed to, and was relying on this computer to tell me when to do what. I had no idea how long it was deferring things out for and when they would come to me. I got a bit scared, but decided to play with the priorities AND at the same time they released an update supporting projects.

Fast forward, and while slightly better the feeling grew on me. On top of that.... Lots and lots of bugs and very hit or miss support. I was out. For a non GTD-er purely with a task list that is on the lighter side I could see Motion being great. But for me, too much loss of control.

I did lots of searching and lots of asking on these forums and I found Skedpal. Now, Skedpal is fantastic. I still say that even though I am leaving it. It works really really well, the bugs are pretty minor. And if you need support the owner is insanely active on his slack channel where I often nearly have live conversations with him. Skedpal has a lot more intuition around how it prioritizes. It does support GTD and the owner is clear on that as he is very familiar with the methodology. You can create areas of focus, projects, and next actions. You can see an outline view where you can see everything at a large level. It schedules based on time and priority and you can establish a very large robust set of priorities. I found 90% of the time it brought to the table what I really needed to work on at the right time. I thought, now this is it!

However, a few issues came up. First off context switching. Using GTD for a long time I take fore granted how effective staying in a context is. Skedpal would give me an errand to run, a call, and something on my computer back to back, followed up by another errand, another call, another computer in a different order. Not at all how I would arrange my day. It became clear to me that a computer can, obviously, only follow a set algorithm to stack priorities. It cannot work like our human brain can to sort that all out. Supposedly there is an update coming in early 2023 to help this situation a lot. When it does release I will take a peak out of interest.

Another major issue for me is the ancillary benefits of GTD. Someday list, waiting for list, etc. These are completely doable in skedpal. But they take some overhead and creative setup to make them work. Once I get it down, they do work well. But it took some practice.

While these were challenges the development of Skedpal is impressive and the owner listens to what people are saying and brings these ideas into the product. However, this weekend the major deal breaker showed its face.... Two events collided and turned my week and weekend upside down. And here is where it just all changed.

Skedpal understood my priorities when I am cruising along. Business is status quo, all wheels are turning. My life at home was the same. I was being presented with the tasks I wanted to do on the days I want to do them. Skedpal was nagging at me I am overcommitted but par for the course. While the order of the tasks didnt make 100% sense I could easily look at the day and what it wanted me to do and do them in the order that was more efficient. Great....

Rewind back to this past Wednesday. My wife and my oldest son catch Covid (all are doing well). All of a sudden I am adding into my CEO responsibility my wife's family responsibilities in getting the kids (that are not sick) where they need to go, taking care of the laundry while she cannot, errands, etc. We are planning a backyard renovation and that was high on our priority this week. Skedpal had that dead on accurate. Well once she had COVID she cannot go shopping for the deck colors and all of the priorities shifted.

Continue to this weekend and another curveball hits. I have my day planned (today) and something comes up with my wife's car. We have been needing to replace it but the right car isnt available. Dealer calls me today, he got one in. Can I drop things to get into the dealership. Well that takes half a day as we all know. So dropping errands, priorities shift again, etc.

And this showed me the single flaw that is insurmountable in a computer generated scheduler. The human brain can process non-linear, non-predictive, cause driven changes in priority.... Not only can it process it, it can process it FAST. I would have had to spend probably 1-2 hours modifying my priorities in skedpal to backburner the backyard project, prioritize the carpool stuff, rearrange my calendar to get all these activities in the system, and bump up this car situation. Without thinking my brain parked several things, jumped into other things. Because my GTD setup is effective in Skedpal I am not worried as I know it is all there. But, when I hit update schedule it is not going to rebound into triage mode tomorrow and realize what needs to happen. It is going to look at my time and all of my prior set priorities and send me shopping again for the backyard renovation, etc.

So long story short.... (although this is far from short).... I am back to getting my next actions back into OmniFocus. And posting this article so in a year or so when I start tinkering again I read it and hopefully learn to leave it alone! Pure GTD plus OmniFocus just gives me the ability to control what I do and how I do it. I must get a lot better about sidelining stuff to prevent overwhelm. But it is clear that the computer AI stuff is very cool and at this point can work when life is not throwing curve balls. But that is the point in which it breaks.... Fast....

Looking forward to responses, comments, and thoughts. If you managed to make it this far through my rambling!
 

cfoley

Registered
This was a very interesting read. Thanks for posting it. Your experience with automation echoes mine, although I have not tried any AI approaches.

One thing that I expect is difficult for an AI is reflecting on the higher horizons to make choices in the moment. It sounds like they delegate this to you by making you enter weightings and rules. This seems easy to write software to do this but as you discovered, it won't help when there is a crisis.

A better approach for the developers might be to watch the sorts of decisions you make in the moment and build a predictive model from that. It could be tuned over time by comparing what it recommends you do with what you actually do. A huge challenge here would be capturing enough data about your real actions and enough meta data about the projects and next actions to help make choices. Shifting this to a data entry burden seems like it would make the approach very unattractive to many.

A final thought: For me the largest benefit of GTD is being able to focus on one thing with confidence. I don't know if I could do that unless I had a good sense of what my commitments were. Trusting an AI to choose actions in the moment wouldn't give me that sense.
 

ivanjay205

Registered
This was a very interesting read. Thanks for posting it. Your experience with automation echoes mine, although I have not tried any AI approaches.

One thing that I expect is difficult for an AI is reflecting on the higher horizons to make choices in the moment. It sounds like they delegate this to you by making you enter weightings and rules. This seems easy to write software to do this but as you discovered, it won't help when there is a crisis.

A better approach for the developers might be to watch the sorts of decisions you make in the moment and build a predictive model from that. It could be tuned over time by comparing what it recommends you do with what you actually do. A huge challenge here would be capturing enough data about your real actions and enough meta data about the projects and next actions to help make choices. Shifting this to a data entry burden seems like it would make the approach very unattractive to many.

A final thought: For me the largest benefit of GTD is being able to focus on one thing with confidence. I don't know if I could do that unless I had a good sense of what my commitments were. Trusting an AI to choose actions in the moment wouldn't give me that sense.
Glad you enjoyed. After writing I felt like I just rambled but it was a bit therapeutic to get it out there. Even made me throw a journaling/diary idea on my someday/maybe list lol.

While I think the approach you describe of watching decisions and trying to haver a predictive model sounds great the problem is since dealing with surprise chaos I did not "fix" or reorganize my system until after. Kind of did a mini weekly review/reset to get it all back on track. I think what I learned is that kind of as you said.... the computer can follow all the rules you throw at it. Motion did not do a good job, too simplistic in the rule category. Skedpal allowed enough configuration where the rules did mimic my real world majority of the time. Now, it didnt understand that not all high priorities had to be done in 1 day. In the real world I might take a high priority item today and do it Wednesday, sliding a few lower priority things that have been sitting in first while still being on target with other expectations (whoever is the recipient of the outcome of my task). But, it was very very linear.

As to your last point I can tell you that with Motion I am 100% with you. I did not have a good sense of my commitments. In Skedpal I did, it was a bit more overhead than OmniFocus so I found myself taking a lot of shortcuts on projects vs next actions. but I generally had it there and had a very concrete sense on my entire workload. That was not an issue.

But already rebuilding my Omnifocus (and now I get to live in two places until I get it all moved back, yuck) and feeling better about it.
 
Top