Uncharted Waters

Mar 19 2014   12:40PM GMT

Yesterday’s Weather

Matt Heusser Matt Heusser Profile: Matt Heusser


Man with a crystal ball to predict the futureYou might call me a “knowledge worker”; I have one of those jobs that involves sitting at a computer, typing things, making phone calls, and talking to people.

The majority of my mornings start out the same. The night before I make plans at all the great things I will do – each seems reasonable. I plan the number of hours each ‘should’ take, and I’m certain to get five or six done by the end of the day, easily.

Then the alarm rings and I check my email.

A short time later, I look up from my monitor to my clock. It is 1:30PM; five hours have passed. I have not had lunch, and I am staring at a video of a cat playing with string.  (Don’t click that link; you might not come back!)

Let’s get a little more detailed about my day, and talk about a meaningful way to predict work delivered.

 A Day In The Life

As I said, each day I plan how long things should take, and expect to knock out three of four in my day. For me, they are usually either piece-rate projects that should take “a couple of hours”, but not always. Right now, my tiny company, Excelon Development, is sponsoring its first conference — which means we have to get a tablecloth, promotional materials, the new website up, and a bunch of other things.  We are organizing training in Lansing, Michigan, a micro conference in New York City, I’ve got to get plane tickets to Europe and figure out a visa for Scotland.

Each morning, I make my todo list. My wife walks in with the baby “what are you planning on doing for the next 15 minutes?” and as much as I’d like to answer “WORKING“, I know she needs me to say “whatever you need.” A client calls with a problem that needs attention right now, an invoice needs writing, I have a couple of meetings scheduled … and the next thing I know, it is 1:30PM, I haven’t had lunch, and there, on the screen, is, something like this:

cute cat

Or email. Or twitter. Or Skype. It doesn’t matter.

The amount I can get done in a perfect engineering day (“in theory”) is different than what I actually get done. If I commit to perfect engineering plans, my company will quickly break all its promises. After that, reputation is the next thing to go. Yet clients do want me to make commitments.

Having done a realistic task analysis of my day, how can I commit to anything?

I use Yesterday’s Weather.

Yesterday’s Weather

The simplest, and surprisingly accurate, way to predict tomorrow’s weather is to say that it will be like today.

To apply that to work, I’ll expect this month’s work output to match last month. Right now, for me, that means about twenty two reasonably large things per month, and twenty-two things that “should be small” — that phone call or tracked-down payment or check that needs to be deposited, account that needs to be balanced, or sales brochure.

That works out to one large thing, and one small thing, per business day. (This is pre-new-baby. Post-new-baby, the number of large things is … less.)

Notice that I don’t plan on getting the three things done a day I “should” plan on, which would get me to sixty-six things a month. Nor did I plan on a “little push.” Instead, I plan on about what I have been doing before. Since an ‘average’ is going to be wrong half the time (as it is in the exact middle), I try to plan on less, and structure projects to give myself more time than the original plan.

Long-time readers, and especially serious software folks, will look at this and scream about “construct validity“; that these things are not the same, and I am fooling myself.

Here’s why I am not worried.

On Not Being Fooled

Years ago I worked for a large corporation that counted the number of projects done per quarter. The result, as you might suspect, was a tendency to create a lot of small projects, or to “label” things as projects that might take minutes. Management was trying to get the metrics right, while the technical people, I am disappointed to say, were incented to game the system. (That’s not a judgement. The system rewarded lots of projects, so they were incented to make lots of projects. What management wanted was value, but they lacked a way to measure it.)

In this case, the person who would benefit from the measurements is the same person taking them. My motivation is to figure out my capacity, not play games best left to the pages of Dilbert.

My motivation is to not fool myself!

How do I do that? Reducing the variation between work items, understanding how black swans happen, studying the flow of the system, and a little statistics go a long way.

But to start, yes, I count things. I expect that the crying baby will come in tomorrow; I expect that I will get bogged down in email. When things go better, I try to figure out why, to figure out what to do more of.

I’ve hit 800 words, and I’ve barely scratched the surface.If you want to hear more, there’s plenty more to come.

12  Comments on this Post

There was an error processing your information. Please try again later.
Thanks. We'll let you know when a new response is added.
Send me notifications when other members comment.
  • Michelle Greenlee
    I recently started to accept my list of things to do each day simply cannot be completed. I'm lucky to get several big things out of the way each day. Weekends are even worse and I don't even have a new baby in the house.
    670 pointsBadges:
  • Matt Heusser
    I'm sure it is a familiar problem, Michelle! :-)

    I'm suggesting: Look at the things you did last month, actually did, to done-done. Subtract a bit, and consider that your capacity. You can take on additional work as long as it is not (as) deadline-driven. Make sense?
    4,960 pointsBadges:
  • Gabby06
    Great article! Describes how I have been feeling lately( except for the crying baby!) Good advice and no, you do not owe me a royalty ( the cat will not be part of my book or website for long!)  Gabby 
    20 pointsBadges:
  • Ben Rubenstein
    Great stuff Matt - I definitely plan to re-read this again on Sunday night, which is when I tend to make the grandest plans (and set my alarm for way too early Monday morning). And again in 3 months when my own little bundle of joy/distraction arrives...

    I've noticed that my 'to do' checklist, while helpful in some ways, can be very misleading - I can have a very productive day but only really check off a couple of things, or check off a lot only to realize that my progress wasn't in the most important places. 

    11,260 pointsBadges:
  • troymagennis
    Nice article Matt.

    I agree with yesterdays weather for forecasting, but just like the weather we get seasonal trends and also bad storms from time to time. Rarely though do we get major changes though, an extra jacket maybe, but rarely anything life threatening. Systems that have survived have dampened out the extinction level changes. And if you fly to another destination, all bets are off - unless you have a record of that locations yesterdays weather (same performance of the team last Thanksgiving break?).

    My only question or comment: I think what we should be estimating with Yesterdays Weather is "what do we know about tomorrow that was different from yesterday." In your case, is there a likelihood of a second or third child in the next few years? (long term permanent trend) That could impact things, or you have to be onsite at a client for a period of time (short term temporary impact). These changes "adjust" yesterdays weather in my opinion, up or down. So, I find the right question to ask even in a #NoEstimates world is "What changes in the system will happen" "will these changes improve our throughput, or decrease it" "what can or should we do about these to alter the throughput trajectory." I apply these adjustments when forecasting and track actuals to improve the teams ability to predict risk impacts like this.

    These factors play a far bigger role than the variation of work size in many cases (not all, perfectly stable teams might be able to make good decisions with estimates, because the system factors are nullified). 

    Thanks for the article,
    20 pointsBadges:
  • Matt Heusser
    Great point, Troy. Perhaps a little much for an 800-word article, but it's the kind of thing I'd like to explore in a part two. Thanks!
    4,960 pointsBadges:
  • troymagennis
    There is an 800 word limit! I'm calling Amnesty International, your rights are being violated. 

    I can't say hello (and my comment above shows) Hello in 800 words.

    Looking forward to part 2, 3 and 4.
    20 pointsBadges:
  • jlottosen
    "a little statistics go a long way" - test progress and defect trends too. Thank you for giving it a name :) 
    congrats on the baby too!
    10 pointsBadges:
  • Matt Heusser
    I want to be /really/ careful with test progress and defect trends, because someone is going to jump in here and scream about "CONSTRUCT VALIDITY", and they'll have a point.

    My recent work, over the past two years, is in applying some of these ideas to test in a way that stands up to scrutiny and I believe I have a few. More to come. :-)

    4,960 pointsBadges:
  • Veretax
    The biggest problem I have with planning, is we plan all this work for an eight hour day, and do we ever think about time lost in the following:

    * waiting for meetings
    * context shifting between projects
    * down time to defocus before continuing or starting new work
    * bathroom breaks
    * trips to help a colleague with some problem you have experience in fixing

    I hear Matt's suggestion about padding the time a little bit, but how do you get a good handle on how much to pad something, so your mind has time to breath, and reduces your stress level just enough to avoid burn out?
    8,675 pointsBadges:
  • Matt Heusser
    It's not padding, Tim, as much as reducing the variation between 'things', counting the 'things' you actually got done last month and planning to get that many 'things' done next month.

    I didn't go into depth here, but by 'padding', I mean using standard deviation to figure out about how far the variation is (on average) from the middle, then planning on getting done the amount of things you get done if all of them turn out to be a little bigger than you expect. (Size of task = average size+standard deviation.)

    That gives you a /dial/ to adjust risk - crank it up for aggressive 'estimates' and down for 'commitments'
    4,960 pointsBadges:
  • PascalDufour
    Great Matt. To get things done as an optimist is not easy. 
    As a father I can tell that concentrated work and a crying baby is not possible. 

    I tend to use a goal for a week. The idea of Yesterday’s Weather I will try. 

    20 pointsBadges:

Forgot Password

No problem! Submit your e-mail address below. We'll send you an e-mail containing your password.

Your password has been sent to:

Share this item with your network: