monitoring multiple graphs; rocks and pebbles

1

My mom is distressingly busy, she says. I take that to mean that her
urgent tasks are crowding out her important tasks.

Well, one of Beeminder’s taglines – thanks to Andreas Stuhlmüller –
is “make the important urgent” so here’s my
everything-looks-like-a-nail advice:

Beemind all the important things in your life!

That’s overwhelming, objects my mom, keeping track of all those graphs.

I don’t think it is! Besides the fact that you can go to
bmndr.com/yourusername and see at a glance what colors all your graphs
are, we just made it so your number of safety buffer days is shown in
the subject line of the bot reminder emails and SMS reminders.

(If you can think of any other way to make it easy to stay on top of
the different things you’re beeminding, let us know!)

Here was the rest of my advice to my mom:

I think it really helps to beemind all of the important things in
one’s life that you’re not happy with: weight, workouts, time spent on
hobbies (like writing) that get short shrift (ha, nice pun!),
cleaning, jobhunting, etc…

Bethany beeminds cleaning – bmndr.com/b/housework – to force herself
to do a bare minimum of 15 minutes a day, but in your case you already
keep your house immaculate so you’re probably more like the
stereotypical college student who has to write a paper but first
decides they can’t concentrate in such a cluttered room so they better
first clean everything and do the laundry and etc etc.

Recall the rocks and pebbles parable [1]: some things will fall
through the cracks no matter what. Make sure that what falls through
the cracks is not major stuff that actually matters to your life. As
you know, just maintaining a household can easily be a full-time job
if you let it be. But 20 years from now you don’t want to look back at
your accomplishments for the last 2 decades and be like, well, I kept
the household running, and, that was it. When you were a full-time mom
that was another story, obviously, since raising kids is definitely a
full time thing (hence the need for full-time daycare if you want to
also have a job).

Anyway, beeminding is so gentle now with the road dial that I think
you should create yellow brick roads for all the things you might want
to beemind and then leave the roads flat so there’s zero pressure
until you’re sure you want to commit.

[1] http://messymatters.com/akrasia/#PEB

–
http://dreev.es – search://"Daniel Reeves"
Follow the Yellow Brick Road – http://beeminder.com

2

(If you can think of any other way to make it easy to stay on top of
the different things you’re beeminding, let us know!)

The simplest and easiest system is lexicographic priorities - work on
X until it’s safe, then work on Y, then work on Z.
A UI that allows you to sort your goals, and only displays at most one
of the in-danger goals, might be feasible.

In some situations, lexicographic priorities may be appropriate.
However, if you’re overloaded (eyes bigger than stomach), following
strict lexicographic priorities means you just never put effort into
some of your goals.
In some situations, that is unacceptable.

If you wanted guidance as to which to work on, you might use a
nonlinear utility function.
Then the system could use the gradient of the utility function to find
which of your priorities is currently most advantageous to pursue.
A moderately plausible family of nonlinear utility functions are the
constant elasticity of substitution functions.
Constant elasticity of substitution functions can be built to have a
particular “goal bundle” - e.g. 60% of my effort ought to go to X, 20%
to Y, 20% to Z.

Johnicholas

3

On Sun, Oct 2, 2011 at 4:01 PM, Daniel Reeves dreeves@beeminder.com wrote:

I think it really helps to beemind all of the important things in
one’s life that you’re not happy with: weight, workouts, time spent on
hobbies (like writing) that get short shrift (ha, nice pun!),
cleaning, jobhunting, etc…

Well said. Commitment devices are so helpful in achieving goals that we
would otherwise neglect.

That’s overwhelming, objects my mom, keeping track of all those graphs.

While it does require a small amount of bookkeeping to make sure you’re
living up to your commitments, the deeper problem is that behaving in
accordance with a commitment is intrinsically unpleasant to your short-term
self, at least until it becomes a habit. (If it weren’t unpleasant, you
wouldn’t be using a commitment device to make yourself do it.) And one can
imagine this accumulated unpleasantness becoming overwhelming.

For that reason, I try to limit the use of self-binding to the one or two
goals that directly address the problem that I think most limits me from
getting the most out of life – that is, my constraint, bottleneck, or
limiting reagent. Then I cut myself some slack in the other areas of my
life, either by not committing to anything else, or by committing only to
the minimum necessary to prevent one of those other areas from becoming the
new bottleneck. For instance, maybe it’s OK if I only clean once a week, as
long as I’m staying on track with my weight loss.

4

I’d probably start with a very simple approach, just provide a nice graphic
summarizing all goals, e.g. safe days, last activity date – and let the
user decide what to do next.

On Sun, Oct 2, 2011 at 6:19 PM, Johnicholas johnicholas.hines@gmail.comwrote:

(If you can think of any other way to make it easy to stay on top of
the different things you’re beeminding, let us know!)

The simplest and easiest system is lexicographic priorities - work on
X until it’s safe, then work on Y, then work on Z.
A UI that allows you to sort your goals, and only displays at most one
of the in-danger goals, might be feasible.

In some situations, lexicographic priorities may be appropriate.
However, if you’re overloaded (eyes bigger than stomach), following
strict lexicographic priorities means you just never put effort into
some of your goals.
In some situations, that is unacceptable.

If you wanted guidance as to which to work on, you might use a
nonlinear utility function.
Then the system could use the gradient of the utility function to find
which of your priorities is currently most advantageous to pursue.
A moderately plausible family of nonlinear utility functions are the
constant elasticity of substitution functions.
Constant elasticity of substitution functions can be built to have a
particular “goal bundle” - e.g. 60% of my effort ought to go to X, 20%
to Y, 20% to Z.

Johnicholas

5

I think the current gallery you see when you sign in or click
someone’s avatar – for example my gallery is bmndr.com/d – is not
too far from that.
You see the color of all your graphs, and the mouseovers tell you the
number of safety buffer days.

On Sun, Oct 2, 2011 at 18:58, Julian Schvartzman
julian.schvartzman@gmail.com wrote:

I’d probably start with a very simple approach, just provide a nice graphic
summarizing all goals, e.g. safe days, last activity date – and let the
user decide what to do next.

On Sun, Oct 2, 2011 at 6:19 PM, Johnicholas johnicholas.hines@gmail.com
wrote:

(If you can think of any other way to make it easy to stay on top of
the different things you’re beeminding, let us know!)

The simplest and easiest system is lexicographic priorities - work on
X until it’s safe, then work on Y, then work on Z.
A UI that allows you to sort your goals, and only displays at most one
of the in-danger goals, might be feasible.

In some situations, lexicographic priorities may be appropriate.
However, if you’re overloaded (eyes bigger than stomach), following
strict lexicographic priorities means you just never put effort into
some of your goals.
In some situations, that is unacceptable.

If you wanted guidance as to which to work on, you might use a
nonlinear utility function.
Then the system could use the gradient of the utility function to find
which of your priorities is currently most advantageous to pursue.
A moderately plausible family of nonlinear utility functions are the
constant elasticity of substitution functions.
Constant elasticity of substitution functions can be built to have a
particular “goal bundle” - e.g. 60% of my effort ought to go to X, 20%
to Y, 20% to Z.

Johnicholas

–
http://dreev.es – search://“Daniel Reeves”
Follow the Yellow Brick Road – http://beeminder.com

6

Thanks Josh! I like the simplicity of this approach a lot. I think
having up to 12 goals is not so bad – mine are at d – dashboard – beeminder – but
I think you’re right that you want to make sure there are only a
couple where you’re really skating the edged (where the yellow brick
road is forcing you to toe the line). Dial down the rest. You may have
contracts out on them but if you’ve got a good safety buffer you don’t
have to think about them.

Johnicholas, I’m not sold on your idea of a nonlinear utility
function. I fear it would feel too arbitrary picking the parameters.
Or maybe the real issue is with the CES family [1] you have to decide
a degree of complementarity of a set of goals and I’m really not sure
where to start with that. Josh’s rough heuristic – have only a few
that are pushing you, make the rest easy – seems like it’s enough. I
might not be appreciating the power of this yet though.

Love this discussion; thanks everyone!

PS: To distill my original email:

Danny’s Guide to Beeminding:

  1. identify an instance of akrasia (want-can-will test)
  2. create a flat yellow brick road (it’s pressure-free!)
  3. reply to the bot (email or sms)
  4. dial up the steepness when you see what’s realistic

to which I’ll add

  1. if you’re skating the edge of too many roads, dial all but a couple down

[1] Constant elasticity of substitution - Wikipedia

On Sun, Oct 2, 2011 at 18:04, Josh Jordan josh@joshjordan.name wrote:

On Sun, Oct 2, 2011 at 4:01 PM, Daniel Reeves dreeves@beeminder.com wrote:

I think it really helps to beemind all of the important things in
one’s life that you’re not happy with: weight, workouts, time spent on
hobbies (like writing) that get short shrift (ha, nice pun!),
cleaning, jobhunting, etc…

Well said. Commitment devices are so helpful in achieving goals that we
would otherwise neglect.

That’s overwhelming, objects my mom, keeping track of all those graphs.

While it does require a small amount of bookkeeping to make sure you’re
living up to your commitments, the deeper problem is that behaving in
accordance with a commitment is intrinsically unpleasant to your short-term
self, at least until it becomes a habit. (If it weren’t unpleasant, you
wouldn’t be using a commitment device to make yourself do it.) And one can
imagine this accumulated unpleasantness becoming overwhelming.
For that reason, I try to limit the use of self-binding to the one or two
goals that directly address the problem that I think most limits me from
getting the most out of life – that is, my constraint, bottleneck, or
limiting reagent. Then I cut myself some slack in the other areas of my
life, either by not committing to anything else, or by committing only to
the minimum necessary to prevent one of those other areas from becoming the
new bottleneck. For instance, maybe it’s OK if I only clean once a week, as
long as I’m staying on track with my weight loss.

–
http://dreev.es – search://“Daniel Reeves”
Follow the Yellow Brick Road – http://beeminder.com

7

I agree with Julian

Sent from my iPhone

On Oct 2, 2011, at 8:02 PM, Daniel Reeves dreeves@beeminder.com wrote:

I think the current gallery you see when you sign in or click
someone’s avatar – for example my gallery is d – dashboard – beeminder – is not
too far from that.
You see the color of all your graphs, and the mouseovers tell you the
number of safety buffer days.

On Sun, Oct 2, 2011 at 18:58, Julian Schvartzman
julian.schvartzman@gmail.com wrote:

I’d probably start with a very simple approach, just provide a nice graphic
summarizing all goals, e.g. safe days, last activity date – and let the
user decide what to do next.

On Sun, Oct 2, 2011 at 6:19 PM, Johnicholas johnicholas.hines@gmail.com
wrote:

(If you can think of any other way to make it easy to stay on top of
the different things you’re beeminding, let us know!)

The simplest and easiest system is lexicographic priorities - work on
X until it’s safe, then work on Y, then work on Z.
A UI that allows you to sort your goals, and only displays at most one
of the in-danger goals, might be feasible.

In some situations, lexicographic priorities may be appropriate.
However, if you’re overloaded (eyes bigger than stomach), following
strict lexicographic priorities means you just never put effort into
some of your goals.
In some situations, that is unacceptable.

If you wanted guidance as to which to work on, you might use a
nonlinear utility function.
Then the system could use the gradient of the utility function to find
which of your priorities is currently most advantageous to pursue.
A moderately plausible family of nonlinear utility functions are the
constant elasticity of substitution functions.
Constant elasticity of substitution functions can be built to have a
particular “goal bundle” - e.g. 60% of my effort ought to go to X, 20%
to Y, 20% to Z.

Johnicholas

–
http://dreev.es – search://“Daniel Reeves”
Follow the Yellow Brick Road – http://beeminder.com

8

I agree with Danny’s mom (Laurie). Multiple goals is crazy making and formula for failure for most, I think. And keeping it simple (no visible utiliy functions is impt.

Sent from my iPhone

On Oct 2, 2011, at 7:04 PM, Josh Jordan josh@joshjordan.name wrote:

On Sun, Oct 2, 2011 at 4:01 PM, Daniel Reeves dreeves@beeminder.com wrote:
I think it really helps to beemind all of the important things in
one’s life that you’re not happy with: weight, workouts, time spent on
hobbies (like writing) that get short shrift (ha, nice pun!),
cleaning, jobhunting, etc…

Well said. Commitment devices are so helpful in achieving goals that we would otherwise neglect.

That’s overwhelming, objects my mom, keeping track of all those graphs.

While it does require a small amount of bookkeeping to make sure you’re living up to your commitments, the deeper problem is that behaving in accordance with a commitment is intrinsically unpleasant to your short-term self, at least until it becomes a habit. (If it weren’t unpleasant, you wouldn’t be using a commitment device to make yourself do it.) And one can imagine this accumulated unpleasantness becoming overwhelming.

For that reason, I try to limit the use of self-binding to the one or two goals that directly address the problem that I think most limits me from getting the most out of life – that is, my constraint, bottleneck, or limiting reagent. Then I cut myself some slack in the other areas of my life, either by not committing to anything else, or by committing only to the minimum necessary to prevent one of those other areas from becoming the new bottleneck. For instance, maybe it’s OK if I only clean once a week, as long as I’m staying on track with my weight loss.