Lifehack: Is GTD generally too difficult for people to use?

[Folke]

I think what you're saying is that one might be more fluid than that, and change the kind of tagging that's done on the fly...?

That's possible - I think. It would take some really good software, however!

I am not sure if I am understanding your understanding [sic], but I think, based on the rest of your post, that we probably are in full agreement. In other words, no, I do not see it as viable to keep changing the definitions on the fly, spontaneously for each task. It needs to be a carefully thought out "system", as you said, and you need to stick with it until you change it, and when you do change it you have to review your tasks and make sure they are all classified correctly according to the new system. For example, if you used to had a context called Errands, but found that this was not detailed enough, so you need to split it into North and South, then this will affect all tasks previously classified as Errands.

But what I did mean, in my previous post, is that this redefinition and subsequent review of task classifications need neither be dramatic nor difficult to recognize the need for. For example, if you discover one day that you are extremely sensitive to sound, and your environment is often noisy, then it will probably occur to you quite spontaneously that you might want to consider implementing a Silence tag, and then if after some careful deliberation you do in fact decide to go ahead with that new tag, then the work of defining the tag(s) in the software and adding them to your tasks is not really all that overwhelming.

I adjust my task classification system (tags, contexts, use of energy and time fields or whatever my current app may have available) several times a year. I hate unnecessary classification (takes work and space). Conversely, I hate not being able to see (filter/sort etc) for what I need to see. So I modify my system whenever I see a way to get the accuracy I want with as little work as possible.

I totally agree that most software do not support classification and filtering (and relevant defaults for all this) nearly as much as they could (or, IMO, should). For example, a NOT filter (Hide filter) is extremely useful and simple (for ruling out the impossible tasks without necessarily having to randomly select just one single context of those that are equally possible). Another example is implicit tags (or hierarchical tags) which would allow you, for example, to automatically associate several people tags with a common company or department tag etc, and thereby with one single tagging operation allow you the choice of filtering either for specific persons or for all people in that company. Tagging/filtering is something that could be taken to a much higher level of convenience and accuracy for the user without much difficulty for the developer.

 

[fwade]

Different ways of tagging

I totally agree that most software do not support classification and filtering (and relevant defaults for all this) nearly as much as they could (or, IMO, should). For example, a NOT filter (Hide filter) is extremely useful and simple (for ruling out the impossible tasks without necessarily having to randomly select just one single context of those that are equally possible). Another example is implicit tags (or hierarchical tags) which would allow you, for example, to automatically associate several people tags with a common company or department tag etc, and thereby with one single tagging operation allow you the choice of filtering either for specific persons or for all people in that company. Tagging/filtering is something that could be taken to a much higher level of convenience and accuracy for the user without much difficulty for the developer.

Wow - that's an awesome comment. I can see that you have thought about this for some time. A NOT filter would be great - I'd love to have that available. It would be easy to view one's tasks on a calendar with a certain tag, or even flip between a calendar view and a list view of tasks with dates on them.

I also wish that I could go back in time and see what happened to particular tasks with a given tag - a historical record. For example, how many of the tasks in my schedule were completed on time? Which ones do I faithfully execute?

I am toying with the idea of writing a special report on tasks (I define them differently and call them "time demands") and tagging is a major concept I want to include. If you have ever read the pdf, "The Book of Promises" you might see where my inspiration is coming from.

Is the topic of tags as they relate to tasks being studied or discussed anywhere in the world outside of this thread? (Hope so...)

 

[Oogiem]

But out of genuine curiosity, what do you actually mean by tag? If you mean the kind of "random" words that you just happen to associate when entering the task, then I totally agree with you, as I already said.

But if you are referring to systematic task characterization, I'd be really curious to know what it is that we actually disagree about; I might have a chance to learn something important here.

Tags relate to the contents of a file or a task. Contexts relate to the external environment.

For example, I use tags in Lightroom because civilians do not have access to ways of using pictures to characterize other pictures. I use word tags to specify the content of my picture files. I have a well defined set of tags with hierarchical definitions that is similar to how the Library of Congress requires historical images to be tagged. It's a defined vocabulary that allows for increasing refinement. So for example a picture of one of our Black Welsh lambs would get tagged with lamb (which carries along with it, sheep and animal) Black Welsh (which carries livestock, breed). A picture of one of our guard dogs Winnie is tagged with her name Winnie which carries along with it the hierarchy of Winifred, dogs, working animals, animal). But none of this relates at all to anything external about that piece of information. It's all related to the content.

I use contexts to specify the external factors that affect a given item. So contexts include things like tools, time, location but are not related to the content of the item in question.

So for me tags and contexts are radically different concepts.

 

[Oogiem]

Hi Oogiem - so far you have mentioned spatial and temporal contexts - do you use others that you can share with us with some examples maybe?

I also use tool contexts.

So I have separate contexts for major software packages I use regularly. It's must easier to do everything I need to do in Lightroom all at once even if it relates to a number of different projects or is a number of different actions. Similarly I have contexts for LibreOffice, Scrivener, Calendar, DEVONThink and so on.

I have a context for my die cut machine. I use it for scrapbooking but also for making vinyl letters and other items we use in the business. So I stack up all the cutting tasks and do them at once time. It's faster that way because I have to get the cutter out, get it attached to my computer and then locate all the materials (vinyl, paper etc). I often will cut things I think I might use in scrapbooks and store them as it's faster to do that than it is to get it all set up each time.

Likewise I have a context for things that require the flatbed scanner. I'd rather do a bunch of those scans at once than do a few intermixed with other computer tasks.

I have contexts for specific web sites that I use regularly. All the actions related to managing the installation of our web site, all the account stuff and installs of Wordpress, PHP and other tools is given the context of that one web site. Similarly I have contexts for the USDA web site that I have to use to report shipment of official tags now that I am an agent for a UK tag manufacturer.

I also use energy contexts but sparingly. I usually activate those contexts during lambing when most of my time is doing work as it appears taking care of newborn lambs. However in between time I have high and low energy contexts that put stuff in so nothing totally falls through the cracks during lambing.

I also create, use and delete contexts on a whim. If something isn't feeling right for me, or a list gets too long, I look at how I can separate it. I work better with contexts that have no more than 1 screen full of items at a time, that's about 25 items. Much beyond that and I don't work well with it. So when a list gets to say 35 items or so I split it up.

 

[Folke]

Tags relate to the contents of a file or a task. Contexts relate to the external environment.

I use contexts to specify the external factors that affect a given item. So contexts include things like tools, time, location but are not related to the content of the item in question.

So for me tags and contexts are radically different concepts.

I understand perfectly. Thanks for explaining.

One thing we have in common, then, is that neither of us labels our tasks by content, for example type of work (bookkeeping, cleaning etc). We both primarily label our tasks by location, tool, time etc. And I think you mentioned people (assistance) in an earlier post? Do you also label by weather (your job perhaps being weather dependent)?

One difference between us might be that I also try to label by energy etc (internal context), not just external. Basically I see all these as "context requirements" - what kind of context I need to be in (or get myself into) in order to be able to do the task.

A big difference between us is that I use some words much more sloppily and interchangeably than you do. You consistently use the word context for these things, whereas I tend to mix all the words that various app developers have chosen for their "context assignment" features (label, category, tag, context, text, comment, location ...).

A further difference might be that I prefer - if my app permits me - to use overlapping contexts (for filtering purposes) and a limited subset of contexts (not overlapping) for list grouping etc.

But "content tags", definitely no, not for me either

 

[cwoodgold]

Oh, I do suffer a lot from not having good enough tagging and filtering options. I even wrote a long post about it here: http://www.davidco.com/forum/showthread.php?15024-Desired-Feature-Elimination-Filters-AND-OR-NOT

OK, here's an idea. You may have already thought of this and you may not like it.

When you tag an item, always put a tag for each category of tags. For example, if you sometimes tag things with people e.g. "John", then always put a person tag, perhaps usually "nobody". If you sometimes tag "high energy" or "low energy", then tag everything with an energy tag, perhaps using "medium energy" as a default. Etc. Then when you want to use NOT logic, you can for example list all the tags for "nobody" and for all the people except "John" (provided you have a list somewhere of all your people tags).

Another idea that was raised on this thread was thinking up new contexts on the fly. I think this is fine provided at the same time you put something in place to ensure that at some point you'll remember to look at that context list. I think I do things like that from time to time. The point is just not to leave open loops.

 

[fwade]

Multiple contexts

I also use tool contexts.
So I have separate contexts for major software packages I use regularly. It's must easier to do everything I need to do in Lightroom all at once even if it relates to a number of different projects or is a number of different actions. Similarly I have contexts for LibreOffice, Scrivener, Calendar, DEVONThink and so on.

Likewise I have a context for things that require the flatbed scanner. I'd rather do a bunch of those scans at once than do a few intermixed with other computer tasks. .

OK - I'm blown away. Thanks for sharing- this is very interesting to me and I wish I had the software capability to experiment in this way. Maybe I do and just don't know it... Hmmm....

This post is going to stay with me for a long time - I didn't know this approach was even an option... "dynamic contexts."

All in all, it appears that by investing the time up front in smart tagging, it makes things easier down the road when the time comes to execute. What a lesson.

Tx!

 

[Folke]

OK, here's an idea. You may have already thought of this and you may not like it.

Thanks for trying to help. I sincerely appreciate that. But no, I am sorry, that would not do the trick for me, and I'll explain why.

I want it to be simple and quick to use, with a minimum of "tagging" (or we'd perhaps better call it "overlapping context assignment", bearing in mind the common use of "tags" to convey other meanings than contextual requirements, for example content, feelings etc). If I have to sit and manually fill in "default" context/tag values for each such "category of overlapping contexts" (e.g. required energy or required person(s)) I cannot afford to be very inventive when it comes to pinpointing situational requirements that really do affect my ability to do tasks. I would have to keep it to the barest possible minimum, which is what I do (out of necessity). To be able to use "overlapping contexts" to their fullest useful extent I would need either defaults to be automatically filled in, or, much better and cleaner, be able to use the absence of such overlapping context tags to indicate "normal".

Second, what you describe would require an OR filter (show "nobody" OR Emma OR Alice OR ...) which is essentially the exact inverse of a NOT filter. Usually apps have neither. They usually have only a "show this only" filter, often usable consecutively as an AND filter, which is useful mainly if you have huge numbers of tasks, e.g. be able to see only those tasks that require me to be both @Out AND with Anna at the same time.

Ironically, although tagging and filtering features is one of my top wishes overall, I am now in the process of probably moving from Nirvana to Doit, which is, if possible, even worse than Nirvana when it comes to managing contexts. Its Tags feature is almost completely useless, so I am forced to restrain myself to the basic "GTD as on paper" type of non-overlapping Contexts (which Doit has). I have defined a Context called @Person which I use for phone calls, conversations, meetings, etc - anything that requires another person's time (and his/her effort of course) while I (we) get the task done. I'll wait and see for a while if I can manage without specific people tags for those people that I have many dealings with.

And this brings us back to one of the original topics of this thread. What I do get out of Doit instead is visible colored priorities, one of the big conflict issues in the larger GTD community (app developers, app users and paper users alike). Whether GTD does or does not actually use priorities seems to be an area of major confusion.

I use priorities (in the app) in the sense of "danger of delaying" - independent of the actual task selection (starring the task). I star things that I decide to do, no matter why I selected it. I often star low priority tasks for immediate action if I find this appropriate due to contextual considerations, e.g. doing another errand while doing errands. But this does not make it "high priority" in my app - its simply makes it a starred low priority task. And conversely, the fact that a task is marked as high priority does not mean I thereby automatically choose to do it now - I might have good reasons not to do it. But what high priority does mean is that I will see it more often and more clearly in my list, and will consider doing it every time I see it, and will not reject doing it unless I have a strong contextual reason (e.g. required person or tool or energy not available) or other high priority tasks that are even more important or even more contextually suitable. This means, "statistically", that tasks will tend get done sooner when marked as high priority. And I base the selection on all four factors - context, energy, time and priority. To me, this is perfect GTD.

 

[fwade]

One thing we have in common, then, is that neither of us labels our tasks by content, for example type of work (bookkeeping, cleaning etc). We both primarily label our tasks by location, tool, time etc. And I think you mentioned people (assistance) in an earlier post? Do you also label by weather (your job perhaps being weather dependent)?

A big difference between us is that I use some words much more sloppily and interchangeably than you do. You consistently use the word context for these things, whereas I tend to mix all the words that various app developers have chosen for their "context assignment" features (label, category, tag, context, text, comment, location ...).

I also tend to plop these terms together and make them interchangeable. There's a cost to that lack of rigor - I can see that now.

We need a clearly defined syntax so that we can distinguish one from another. And we need it for tasks.

Anyone know where to find that book? It could save me a few months... at a minimum...

Francis

 

[cwoodgold]

OK - I'm blown away. Thanks for sharing- this is very interesting to me and I wish I had the software capability to experiment in this way. Maybe I do and just don't know it... Hmmm....

You can just use a separate piece of paper for each context, e.g. "flatbed scanner".

 

[fwade]

You can just use a separate piece of paper for each context, e.g. "flatbed scanner".

Paper -- aaaaaaaaaaah! - Running away screaming.... :mrgreen:

 

[vbampton]

OK - I'm blown away. Thanks for sharing- this is very interesting to me and I wish I had the software capability to experiment in this way. Maybe I do and just don't know it... Hmmm....

Sign up for a Connect trial so you can listen to Oogie's Slice of Life interview. You'd love it!

 

[Folke]

Question about Omnifocus

... while I do have physical contexts (particular pastures or buildings) I also have contexts that relate to what help I have available and to the time I can complete them (phone business hours vs just phone)
...

BTW the GTD manager I use does not do tagging at all and I've never found that a problem.

Does Omnifocus allow you to put tasks in more than one context?

You mention that you have defined context for locations, tools etc, and also for what help you need (= required people, I assume) and what energy you need (although you use this sparingly). So if you a have a very demanding (high energy) task that you need certain help/people to do, and that the job needs to be none in pasture1, can you let the task carry all three contexts, and it will show up in whichever of these three contexts you look, or do you have to choose just one single context - probably the most significant/scarce/difficult/infrequent of all the contexts that apply (which is what I need to do in Doit, and had to do in Toodledo at an earlier stage, whereas Nirvana allows unlimited overlapping contexts).

 

[Oogiem]

Does Omnifocus allow you to put tasks in more than one context?

No, I've only very rarely needed it. I use contexts as the limiting factor. One reason I so rarely use energy ones. Usually better to see all tasks in a context and then use my intuition to handle the priority, time and energy avail parts of deciding what to work on.

In your example the context with help is the limiting factor. If the item requires help to do it doesn't matter if it's in the far west pasture or the house or the barn, if I don't have help I can't do it at all and I'd find it upsetting to see it in those contexts. So 'd put it in the Outside with help context and worry about what pasture /place it was in when I was looking at that list. That's one reason I keep my lists to about 25 items, so Ic an glance at pretty much the entire list in a context at once time.

 

[Folke]

I use contexts as the limiting factor.

Yes, that's pretty much how I do it now in Doit, too - and also how I used to do it on paper in the last century. I had been hoping, though, that computers would help me do all this a lot more conveniently and accurately :-(

Luckily, Doit has a second "multiple-context" field (which they call Tag), which allows me to use contexts such as energy, individual people, timing factors etc without increasing the total number of "main contexts". Unfortunately, though, this tag-context feature is so poorly designed that these contexts are not visible in the task lists - you can only use them for "show this only" type filtering. Otherwise, I probably would have used this feature to manage all my contexts. In Nirvanahq, all contexts are independent/overlappable, which is fundamentally better, IMO, but Nirvanahq has just as limited filtering capabilities as Doit does, so I could not really benefit much from the flexibility in its context assignment.

Currently in Doit one of my main contexts is @Person. Just as you point out, people are usually the most critical factor, so I typically put tasks down as @Person in caes where many contexts apply. I have not yet implemented neither any individual person tag-contexts nor any "combo contexts" such as Out with Person or Out with Anna etc. I used to have a few dozen people contexts when using Nirvanahq and I am very curious to see how I will survive this "forced simplification". Individual people contexts are very useful when you speak with a frequent contact and want to see all you need him/her for on one screen (across Next, Waiting, everything). (For infrequent contacts I used to use generic "contacts" such as Client, Supplier, Authority etc.)

(If I sound negative about Doit it is only because it does have its cons, as do all apps - but it does have a couple of pros, too, which were important to me, and which is why I chose it. I don't have a Mac - otherwise I would definitely have tried Omnifocus)

 

[AJS]

@ folke,

I am in the same situation wrt Doit.im. Some of the features I find quite annoying, such as editing tasks usually means an extra click than many other apps. But then tasks and projects are archived properly so it's more of a complete project management solution which means I don't need to use a plethora of different apps to manage a project. It's an acquired taste though, and for simple GTD it would probably be quite off-putting. There are so many options that it can get quite confusing. I find though that if you limit it to those you find work for you, it can provide quite a powerful solution. I've heard it described as a decent alternative to Omnifocus, which I've never used as I don't have a Mac. I do wish they would add bulk editing to tags though. Without it, it's one of the features I don't use.

 

[AJS]

Just to add

Addressing the central topic, one of the reasons GTD can seem so complicated is that David Allen has been very specific on the minutiae of the productivity process. This is where other "simpler" systems tend to obfuscate, but that doesn't mean the problems go away. You still need to know how to file reference material. It's still nice to have some advice how to define projects and action items even if you don't need it.

While there are other methods out there that have their own strengths, I really appreciate that GTD pretty much covers every level from small actions to life plans and there is advice available on how you might implement it. Which other productivity guru has provided so much specific information at such a low price entry level?

 

[Cpu_Modern]

Yes, that's pretty much how I do it now in Doit, too - and also how I used to do it on paper in the last century. I had been hoping, though, that computers would help me do all this a lot more conveniently and accurately :-(

They do. The problem at the moment is that the mainstream software packages are still too limited or too limited on purpose. OmniFocus is the one contender and it certainly is worth to keep an eye on it.

If you really want to bend the computer to your will, you can, but that involves scripting.

I don't know if OmniFocus has a decent AppScript port, but if it has, you are just a mere 50$ or so away from getting what you want. Pay a script kiddy to implement the missing features you need.

Alternatively this could be done in a similar environment like Excel or SmallTalk. Peace of cake for your next door geek.

OmniFocus itself started as such a scripting package for OmniOutliner. It is possible to do.

The one software that does is already is Emacs Org Mode. This has the drawback of being a text-mode based thing, so you end up learning all the keyboard shortcuts to navigate that thing. (Tell me specifically what you want to do in Org Mode and I can tell you how to do it.)

 

[mcogilvie]

Retro me, Satanas

They do. The problem at the moment is that the mainstream software packages are still too limited or too limited on purpose. OmniFocus is the one contender and it certainly is worth to keep an eye on it.

If you really want to bend the computer to your will, you can, but that involves scripting.

I don't know if OmniFocus has a decent AppScript port, but if it has, you are just a mere 50$ or so away from getting what you want. Pay a script kiddy to implement the missing features you need.

The one software that does is already is Emacs Org Mode. This has the drawback of being a text-mode based thing, so you end up learning all the keyboard shortcuts to navigate that thing. (Tell me specifically what you want to do in Org Mode and I can tell you how to do it.)

Apple Script is surely the work of the devil, some sort of unholy Frankenstein monster made from parts of other programming languages. And Emacs- a blast from the past. Unfortunately, my fingers twitch vi, not emacs.

Seriously, the problem with scripts for modern GUI programs is that they take time to write and debug, and they are fragile and break easily. I've used scripts a lot using shell, awk, et cetera, but that's a very different environment to work in. Paying someone to write a script if you can't do it yourself just sounds like a heap of pain somewhere down the line.

 

[Cpu_Modern]

Well, vi is the devils's editor...

Are you so seriously advanced in vi usage that viper-mode is not good enough for you? If it doesn't hurt too much, you could use org-mode via viper-mode...

Anyway, I can understand your arguments. Paying someone to do the scripting can be a good exprience, qeustion is how you manage that project. Obviously your computer skills are tops, so chances are your people skillz are bad. I am kidding I am kidding.

Still I think a lot of it comes down to feelings. You just know that GUI scripting is brittle in comparison to a compiles app, so it feels worse than it actually is. For a geek, that is.