Random Time Management
Wednesday, January 22, 2014 at 14:12 As promised in my last post, here’s the method I am using at the moment with great success. You need a random-number generator to work it. The one I’m using is at http://www.random.org/integers/
I am using paper and pen, but I’m sure it can be adapted for electronic use. I just haven’t yet attempted to do so.
I’m using a loose-leaf binder with lined pages of 32 lines, but the method will work perfectly well with a bound notebook and pages of any number of lines.
First I list all my tasks in the notebook - one per line.
I then set my randomizer to produce integers in the range 1 and 32 inclusive. The upper number is the same as the number of lines on a page. This is just a convenient number which produces reasonable results, but you can use a lower or higher number if you wish. [Afternote: I now strongly recommend setting it for the number of lines on the page and not any other number]
Starting from the beginning of the list I use the randomizer to produce a number and move down the page that number of lines. I then do some work on the task on that line. Please note that I don’t have to finish that task, just do some work on it.
Once I have worked on the task, I cross it off. If I have not finished it or if it is a recurring task, then I re-enter it at the end of the list.
I then use the randomizer again and count to the next task (going to the next page if necessary).
When the number the randomizer produces would take me beyond the end of the list, I circle back to the beginning of the list, ignoring empty lines on the last page.
I continue circulating through the list in this way.
When I’m counting forward, I INCLUDE in the count the lines which have been crossed off. If I land on a line on which the task has been crossed out, I move to the next line in which there is an active task. I call this movement a “slide”.
For example imagine I have the following tasks:
Email
In Tray
Invoices
Date of next meeting
Write report
Cash check
Tidy desk
Performance reviews
I throw a five, so I count down the list, remembering to include the crossed out lines. I land on the “Write Report” line. I then “slide” to the next active task which is “Performance Reviews”. Slides work slightly different from counting. If a slide takes you to the end of the page, you circle back to the beginning of the SAME page. So if “Performance Reviews” in the example had already been done, you’d have circled back to “Email” at the beginning of the page.
Email
In Tray
Invoices
Date of next meeting
Write report
Cash check
Tidy desk
Performance reviews
Counting crossed-out spaces and sliding are very important, because they have the effect of increasing the chances of the older tasks on the list being selected. Note that if you don’t include lines with crossed-out tasks in the count, then every task will have an exactly equal chance and there will be no preference for older tasks.
A few points to note:
1) Random numbers behave randomly. They don’t behave in the way we expect them to behave. If they did, they wouldn’t be random. You will find that you are constantly surprised by them.
2) The system as described has a built-in bias towards clearing the older tasks off the list. This means that nothing will stay on the list for very long. How long that is depends on the length of the list and the amount of time you can devote to working on it. If you want things to move on really quickly then keep the list short.
3) The random-number generator is quite indifferent to your priorities, wishes and time-pressure, so if something needs doing now - do it!
4) Any attempts to increase the probability of certain tasks being selected will result in the chances of all the other tasks being reduced. So I advise against it.
Reader Comments (135)
I will have to figure out how to incorporate this concept and make it my own; I seldom have more than 10-15 tasks (projects) active at any one time... Perhaps there's stuff rattling around in my head that needs to be captured. Wait! There's a new task!
<<Also using cards with one task on each could work. Just shuffle the card. Do task on the top card, move it to the bottom, do second card...occasionally shuffle again.
January 21, 2014 at 20:02 | Unregistered CommenterDaneb>>
And thought: Damn, I've got the perfect tool (arrived a couple of days ago) for this: http://frictionless.bigcartel.com/category/index-cards
One card per project. Hm...
First, I got into a bit of trouble Sunday because I had a looming deadline Monday which meant I had to break out of the random system just to work on the deadline project. That might have been because I only started the system on Saturday. I'm thinking that normally if the system is truly random with a bias towards older items, bumping up against deadlines won't happen. If it does happen, it might be a sign that I just have too much on my plate, which is a good thing to know. Of course, as in all Mark's systems, genuine emergencies must take priority - as always, if something needs to be done, do it.
Second, I'm wondering about dismissal. There really is no dismissal in this system, because there is never a time when I don't feel like doing a task. In AF1, if I got to a page with a few tasks and no tasks stood out to be acted upon, I would dismiss the whole page. In this system, if it's on the list it's going to get done unless there is some external reason for not doing it. I don't know if that means anything or not. I've been throwing everything at the list, and I love it. In fact, it's kind of addictive. I can't wait to see what comes up next, because there are a few favourites I want to get to. In a way, its unconsciously training me to populate the list with fun stuff I want to do. This new method is causing me to completely rethink the way I manage my tasks.
<< I had to break out of the random system just to work on the deadline project.>>
Whenever an exception to the rule pops up - and I rejig the rules to fit the exception - the system loses it's power. And then I lose trust in the system, and soon after, stop following the rules. I know that sounds childish, but it has happened time and time again. So either I follow the rules, and deal with the consequences and rewards, or find something else that works.
A suggestion for those that like doing things analog, I can recommend using a twenty sided die (they are available from most hobby stores)
You stated:
"4) Any attempts to increase the probability of certain tasks being selected will result in the chances of all the other tasks being reduced. So I advise against it."
I've been doing exactly what you advise against for years and it has saved my bacon each time. I'm working from my day's closed list that I've prioritized from a weekly list. When I'm stalemated, the game helps me to get back on track. Even though I'm temporarily stuck doesn't mean that I want to do any random task. By priming the deck, I'm still able to stay on track. I don't see why you'd advise against favoring tasks from your WILL DO closed list. This way I'm likely to draw close to the number of cards I need to complete the tasks. The ones that take longer get more cards than the ones that don't and I prime the deck to choose my CI sooner than later. The reason I'm doing this is to get unstuck. Once I'm unstuck, I can return to the list. I honestly don't see how this is a bad thing especially because I know it works perfectly for the times I need to get unstuck. OTOH, you might advise against it because your list is used for more than one day.
I know that you're going to say that I'm going to do the entire closed list anyway. That's my aim but I want to clear the most important things first in case something happens where I can't actually finish it. Also, I like to get the PITA stuff out of the way first. In fact, it's probably those PITA tasks that are stalemating me. I just need a few nudges that this game helps with.
I like to have one list (home & work). I wonder if it is suitable to use the 'slide' to go to the next task that can be done in the current context (if one randomly lands on an task that cannot be done)?
< When the number the randomizer produces would take me beyond the end of the list, I circle back to the beginning of the list, ignoring empty lines on the last page.>
Do you a) circle back to the first available task on the list or b) continue the count on from the top of the list? a) would seem preferable as it is simpler and seems to lend itself to the bias of addressing earlier tasks.
I continue the count on just as if the first page were appended to the bottom of the last page.
<< any thoughts on naming the system? >>
I suppose in the spirit of the system I ought to use the "Random Article" button, rather than the Featured Article of the Day. So here goes....
I proudly announce the name of the system to be:
The Remelsbach System.
Because otherwise all tasks have an equal probability. Counting the crossed-out tasks biases the system toward the tasks at the beginning of the list.
How about, "The Final Version ...2" :)
Now in the example:
Email
In Tray
Invoices
Date of next meeting
Write report
Cash check
Tidy desk
Performance reviews
If I count the active tasks alone, there is 25% (1 active out of 4 active) chance of choosing ‘In Tray’. If I count all tasks, there is 12.5% (1 active out of 8 total) chance of choosing ‘In Tray’. If 25%>12.5%, isn’t there higher chances of selecting an old task by ignoring the crossed out items?
In the next example after some work we have:
Email
In Tray
Invoices
Date of next meeting
Write report
Cash check
Tidy desk
Performance reviews
If I count the active tasks alone, there is now 33% (1 active out of 3 active) chance of choosing ‘In Tray’. If I count all tasks, there is still 12.5% (1 active out of 8 total) chance of choosing ‘In Tray’. If 33%>12.5%, isn’t there still higher chances of selecting an older task by ignoring the crossed out items?
I understand that the list grows with new tasks, unfinished items, and recurring items. But the larger the list the larger the denominator. But, it will always be 1 active task/ N_active > 1 active task/(N_active +N_inactive).
since: N_active < (N_active +N_inactive),
then: 1 active task/ N_active >1 active task/(N_active +N_inactive).
I’m not understanding the bias. And I feel as if Im missing something significant. Sorry for being a little slow here :) Maybe its how Im reading the rules? I’m eagerly looking to be corrected, because I feel as If Im going crazy.
At the moment, after completing or working on a task (which is a Page on the CP task list), I am resorting to inserting a strikethrough page immediately below the task I worked on, then moving the task to the bottom of the list. The strikethrough page then acts as a placeholder for the task I just worked on. If the Randomizer then hits that page, I slide to the next page.
Anybody else here using CP and that has a better implementation idea?
The chances of any one number coming up are of course exactly the same for each number.
What is different though is how many numbers relate to any one task.
At the moment my first active page has 9 active tasks on it. There are 32 numbers (lines) on the page, all of which will result in one of the 9 tasks being selected..
My last active page has 23 active tasks on it and 29 numbers (lines). So you can see that there is considerably less chance of any specific task on the page being selected.
If only the active lines were included in the count, then on the first page there would be 9 numbers relating to 9 tasks, and on the last page 23 numbers relating to 23 tasks. So there would be exactly the same odds for every task on all the pages.
This is important because with equal odds for each task, it can take an unbelievably long time for every task to be actioned. You can try the experiment of seeing how many throws it takes to come up with every number between 1-32 (which is what I have my randomizer set on). I can guarantee you will be surprised.
<The completed tasks remain on the list. If the randomizer hits one of them, you "slide" to the next unfinished item. This increases the probability of landing on the unfinished task that immediately follows one or more completed tasks. This leads to the items remaining on older pages having a high chance of being selected.>
This made sense to me a bit. The slide ensures that ‘landing on a crossed off item slides you to the next active task’. Makes sense in terms of proximity. But if I’m moving FORWARD in the notebook to the next active task, doesnt that mean that items remaining on NEWER pages have a high chance of being selected?
I would think the only way to ensure older tasks are biased is to slide backwards (to tasks before or older than the crossed off item).
Mark:
<At the moment my first active page has 9 active tasks on it. There are 32 numbers (lines) on the page, all of which will result in one of the 9 tasks being selected..
My last active page has 23 active tasks on it and 29 numbers (lines). So you can see that there is considerably less chance of any specific task on the page being selected.>
My math must be incorrect but:
9 active tasks out of 32 lines: 9/32*100= 28%
and
23 active tasks out of 29 lines: 23/29*100=79%
So, isnt there higher chance of a task on the last active page being selected since more of the lines are open to being selected?
Suppose that your entire notebook had only the two pages you describe (the first page and the last active page). So that in total there are now 61 lines to land on (32 lines from page 1 + 29 lines on page 2= 61 lines total). And there are now a total of 41 active tasks (9 active tasks from page 1 + 23 active tasks from page 2= 41 active tasks).
My math gives:
41/61*100=67%
meaning 67% of the notebook is filled with open tasks. and the remainder, 33% will yield slides.
The only time slides create a bias is when there are more inactive tasks than open tasks. But that how will that happen if the list is always growing with new tasks?
Regardless, of the 67% of active tasks, still theres more weight on the second page (newer page) with open tasks.
<If only the active lines were included in the count, then on the first page there would be 9 numbers relating to 9 tasks, and on the last page 23 numbers relating to 23 tasks. So there would be exactly the same odds for every task on all the pages.>
But your chances of selecting any 9 tasks out of the 9 options on the first page is:
1/9*100= 11%
and your chances of selecting any 23 tasks out of the 23 options on the last active page is:
1/23*100=4%
The odds arent the same on each page. one has 11% odds and the other has 4% odds.
I want to clarify, that the randomizer’s lower and upper limits always remains the same. That is, it always selects a random number between 1 and 32 (in your notebook). I get that. Just like rolling a die with 32 faces. But from the math above, Im not understanding how theres a bias towards older tasks. It seems as if everything that is told is the opposit for me. Which means either my understanding of the rules are messed up or my math is incorrect.
I realize this is annoying. But the system seems interesting and I would like to understand it rather than be using it and ‘believe’ its working. I appreciate any clarifications. LOL
<< Which means either my understanding of the rules are messed up or my math is incorrect. >>
Or both!
<< My math must be incorrect but:
<< 9 active tasks out of 32 lines: 9/32*100= 28%
<< and
<< 23 active tasks out of 29 lines: 23/29*100=79%
<< So, isnt there higher chance of a task on the last active page being selected since more of the lines are open to being selected? >>
Yes, you are right - your math is incorrect. What you have worked out is the chances of a task being selected if there were no "slide" rule. And the same applies to all your other calculations.
But since there is a "slide" rule your calculations are irrelevant.
The correct answers are that there is a 100% chance of selecting at least one task on the first page whenever the randomizer lands on that page - this is because the "slide" rule always selects a task on the same page that the number lands on.
On the last page there is a 100% chance of at least one task being selected if the randomizer lands on one of the 29 used lines.
Since I have 9 tasks on the first page and 23 tasks on the last page, the average chance of a *specific* task on the first page being selected is 100/9 = 11.1% and the average chance of a *specific* task on the final page being selected is 100/23 = 4.3%
Note that I talk about the "average" chance, as the individual tasks have different chances according to the length of the slide leading to them.
If you can't follow the logic in my previous post then I suggest you try an experiment.
List the numbers 1-32 and see how many throws of your randomizer it takes before you delete the last number from the list using these two different methods:
1) If the randomizer lands on a deleted number you "slide" to the next undeleted number and cross that off, circling back to the beginning of the list when necessary (i.e. the number after 32 is 1).
2) If the randomizer lands on a deleted number you throw again until you land on an undeleted number.
HINT:
You can print off a list of random numbers from RANDOM.ORG which will save you time and make it easier to count the number of throws.
<< The slide ensures that ‘landing on a crossed off item slides you to the next active task’. Makes sense in terms of proximity. But if I’m moving FORWARD in the notebook to the next active task, doesnt that mean that items remaining on NEWER pages have a high chance of being selected? >>
Here's how the "slide" rule works. If you land on a completed task, you slide forward to the next unfinished task. However, if you slide all the way to the bottom of a page, you continue sliding from the top of THAT SAME PAGE, rather than sliding to the top of the next page.
So, as Mark said, if you land on page 1, you are ensured of selecting a task on page 1, whether you land there directly with the randomizer, or by sliding.
Mark has already explained this, but maybe it will help to get it from a slightly different perspective as well.
I'm only now noticing that you and mark mentioned that slides can circle you to the beginning of the same page like 4 times.lol For someone who didn't read the instructions carefully (myself) it is very confusing to find the bias. Now I get it.
P.s. I believe its the way the post is written. It seems like you have instructions, an example and then notes. But that key instruction is hidden between two examples. So it could be overlooked like I did. I'm talking about this part : " Slides work slightly different from counting. If a slide takes you to the end of the page, you circle back to the beginning of the SAME page. So if “Performance Reviews” in the example had already been done, you’d have circled back to “Email” at the beginning of the page."
Anyways, now that I get it, seems pretty cool. I'll be trying it (correctly) tonight.
Do you write proposals, contracts or specs as part of your job? You are quite articulate.
But I think that's the first time anyone ever called me articulate! Usually it's "long winded" and "can't you up-level that?" and "can't you write that shorter?" LOL!
When you write a thorough post, I appreciate the details. You make things easy to understand. None of your words are superfluous.
I, on the other hand, am not graced with your talents. My ramblings sound addled....
This makes me wonder : it is theoretically possible to completely strikethrough a page in the middle of the notebook while there are still unstriked items on older pages. (A trivial example on how it is possible is to land on the first item of a page and roll ones all the way to the bottom) Did it happen to anyone who tried the system so far? It did not happen to me in three days.
(Just out of curiosity)
I'm asking as most electronic mediums do not have pages as in a notebook. That has been a problem for me trying to do AF and sf through an app. Any suggestions?
Has that happened to anyone with the official system? A block of crossed-out tasks at the end of the page that you can't seem to jump over, so you're stuck on the same page for long enough that the system loses the random feel?
Until just now I thought that sliding backwards instead of forwards would work, and keep a bit of pressure on the older tasks like the official system does, without the need for pages or the often-missed slide to top of page rule, but it's even more prone to those long clean stretches.
Once again, Mark's system works best as written.
(Wishing the universe would stop playing whack a mole with me for long enough that I can find my list. Even with the small rubber mallet, it's getting annoying.)
<< Has that happened to anyone with the official system? A block of crossed-out tasks at the end of the page that you can't seem to jump over, so you're stuck on the same page for long enough that the system loses the random feel? >>
I think it's unlikely that this would happen for the simple reason that if you had a long block towards the end of the page there would almost certainly be only a few tasks left on the rest of the page. Your attention then gets grabbed by whether you can completely clear the page that visit or not.
<< This makes me wonder : it is theoretically possible to completely strikethrough a page in the middle of the notebook while there are still unstriked items on older pages. >>
Yes, it's happened to me. It's quite possible, though it doesn't happen very often.
<< I'm wondering if having pages gives any advantages over one long list (one lone single page). Does the sliding possibility on each page have a stronger bias over say sliding at the end of one long list? >>
I don't think one long page is necessarily worse than dividing the list into pages, but there is one problem you would have to overcome:
Because there would be only very rare slides from the last page to the first page (due to the constant adding of new and re-entered tasks), the first task on the list would nearly always have to be hit exactly by the randomizer. If you did the exercise I suggested earlier, you will know that it can take a very long time for one specific number to be hit (it took me 122 throws to clear all 32 numbers). Any task that finds itself at the beginning of the list will therefore find itself at a severe disadvantage.
I don't think that is an unsurmountable problem, but you do need to be aware of it.
<< there is one problem you would have to overcome >>
I think you could address that issue by sliding backwards instead of forwards. But that introduces new problems... Since you don't have pages, you can't "clear" pages. This would result in large blocks of completed tasks. Your random number range might not be big enough to cross these large blocks. Example: you decide to generate numbers from 1 to 32. But you have a large block of completed tasks, say 35 of them. Thus you can't cross this chasm - you'd always be sliding backwards to the task immediately preceding this block. You'd basically be forced to work backwards through the tasks preceding this block. Maybe that's OK. But I'd guess it would introduce resistance to the system because the random nature of task selection is essentially removed, till you clear out all those tasks.
Another problem introduced by a pageless version of the system: You'd need to choose an arbitrary range for your random number generator. Maybe that would be OK. But what would be the best range? You'd want it high enough the avoid the issue described in the previous paragraph, but not so high that counting tasks becomes tedious.
Because of these kinds of problems, I suspect going pageless would totally change the dynamics of the system, and you'd need to rewrite the rules completely to achieve a similar outcome.
For a notebook, its neat. However, I'm concerned with the pageless issues described above. Chris's and Seraphim seem to have a solution.
Going backwards on a slide as Seraphim described still has that chasm issue.
Switching directions based on even/odd throws as Chris suggested seems as if it would even things out a bit. Though, Im guessing this adds a layer of overhead, remembering to switch directions
Are there any other solutions that can even the bias between page (notebook/analog) vs pageless (one long single list/electronic/digital) versions Mark? I'm asking as I know it can be annoying when people bring up the notion of tweaking your system and you said that it probably can be adapted for electronic versions.
How would you adapt the system for electronic use?
> How would you adapt the system for electronic use?
I'm still in early days of using the random method, but here are some ideas I've used in the past, using a simple text editor (actually, I'm now just using the default Notes app on the Mac/iPhone/iPad):
- Add a blank line every N number of items to mark the start of a new page
- When I tried with longer pages, I would just “open a new page” by pasting N empty bullets as placeholders for the list items
Currently I'm doing one long list because the number of open tasks I have is quite small (twenty-ish)… I few things I did:
# What random number to use?
I started to randomise with around 1/3 or 1/4 of my starting number of tasks. Since I had 20 tasks in my initial brain-dump, I'm just rolling a dice (1-6) to select a task (the physical action of rolling a real dice is turning out to be more fun than I thought :D)
I remember Mark mentioning at some point that he has approximately 90 tasks open and since he was randomising with 32, I figured that would be a good enough ballpark ratio for getting started.
# Hard to hit first item?
Because there wouldn't be any pages, I though of leaving 3 scratch lines before the first item (half of the dice) to make some sort of cushion for the wrap around… but luck has this things and the first draw I got was 1 so I started immediately on the first one :)
One key though is to try to avoid seeing the entire randomized list at once, because it is tempting to let your eyes wander to the tasks in cells A2, A3, A4 etc, and then a familiar feeling of procrastination kicks in. Usually I make cell A1 (the first task) very large so I can only see the first task and HAVE to do it.
This sounds like it would be very easy to implement in a spreadsheet. On Sheet 1 just list each task on its own line in column A. On Sheet 2 have a macro which generates a random integer between 1 and LastRow. You could even have conditional formatting for all of Sheet 1 Column A such that if the random number shown on Sheet 2 matches the line number then the line has a yellow background. This will automatically highlight the selected task on Sheet 1.
Once the task is done simply delete the row, the remaining tasks will always be sequentially numbered and selected directly by the randomiser. Sliding won't crop up.
Thanks for the diversion. I downloaded the 30-day-free-trial of CP NoteBook.app and it's pretty neat. I especially like the different grids - e.g. Engineering Pad. But I realized that the formatting keeps changing, depending on the window size on the screen. Is there any way to keep it consistent on screen and also have PDF output that matches the on-screen look (WYSIWYG)?
<< Once the task is done simply delete the row, the remaining tasks will always be sequentially numbered and selected directly by the randomiser. Sliding won't crop up. >>
That will result in the odds being equal for every task on every throw. There's nothing wrong with that as long as you realise what it entails.
The sliding means that the system clears all tasks within a fairly short time (the exact length depending on the length of the list). Without sliding there is no such guarantee. A task which has been on the list for a week will have no more chance of being selected than a task which you put on the list five minutes ago.
It is easy to copy in the spreadsheet. Hidden column B contains the number 1 for each task entry. Adding a new entry automatically populates column B with the number 1. Each use of the randomiser macro increments column B. The selection algorithm factors in the number in column B, so an entry with the number 2, 3, 4, ...has increasingly more chance of being selected than an entry with number 1.
This weighting will ensure that the longer an task hangs around for, the more the selections will be biased towards them.