Fix you some bad estimation habits

Preview:

DESCRIPTION

Slides as given for the Feb. 12, 2014 talk at Bay Area Software Testers. (btw, I failed to give credit for the "Stand Back!" t-shirt image, it was from the XKCD T-shirt here: http://store-xkcd-com.myshopify.com/products/try-science) Also forgot reference to the paper on Fibonacci numbers in planning poker affecting estimates: http://simula.no/publications/Simula.simula.1282/simula_pdf_file

Citation preview

fix youestimatingbad habitsTED M. YOUNG

RENAISSANCE CODER & AGILE/LEAN HERETIC

AT GUIDEWIRE SOFTWARE

estimating bugsAND OTHER "OVERHEAD"

ignoring all previousor other projectsREFERENCE CLASS FORECASTING (OUTSIDE VIEW)

LARGE ORGS HAVE AN ADVANTAGE HERE

The prevalent tendency to underweight, or ignore, distributional information is perhaps the major error of intuitive prediction.

TVERSKY & KAHNEMAN, 1977

ignoring variabilityof velocityVELOCITY CAN EXHIBIT A ± 2× SPREAD OF VARIATION

Bad habit number 8

not using statistics

STATISTICS

From: Troy Magennis 2012

using single valuesNEGLECTING DISTRIBUTIONAL INFORMATION

WHAT WE THINK OUR ESTIMATE DISTRIBUTION IS

Image from "Estimates in Software Development", Dubakov 2013

flaw of averagesWHY EVERYTHING IS BEHIND SCHEDULE

ESTIMATE DISTRIBUTION IS MORE LIKE LOG-NORMAL OR WEIBULL

ATTRIBUTE SUBSTITUTION?Image from "Estimates in Software Development", Dubakov 2013

treating every type of estimate the same wayEPICS, THEMES, STORIES, TASKS: POINTS FOR ALL!

Bad habit number 3

ignoring value ofthing being estimatedSPEND LOTS OF TIME ON THINGS OF LITTLE VALUE

CUZ IT'S EASIER!

Bad habit number

think about your estimate not theirsYOU UNDERESTIMATE YOUR OWN COMPLETION TIME

BUT NOT OTHERS'!

Bad habit number

estimatingthings you'llnever work on

Bad habit number 5

only estimatingtouch timeDISCUSSION? COORDINATION? BAH, WHO NEEDS IT?

Bad habit number 3

estimating timeinstead ofcomplexity or effortPOINTS/SIZES GET EQUATED TO TIME

IN BOTH DIRECTIONS!

Bad habit number 9

completion dateorhours of work?THEY'RE NOT THE SAME THING

using non-linearvalues for estimatingFIBONACCI AND NON-LINEAR SEQUENCES DECREASE ESTIMATES

USE T-SHIRT SIZES INSTEAD

using planning pokerATTEMPT TO REMOVE ANCHORING BIAS

REPLACES IT WITH SOCIAL ANXIETY

estimating what we think the story isINSTEAD OF WHAT IT ACTUALLY IS

WHAT WILL BE DIFFERENT IF THIS STORY IS IMPLEMENTED?

premature estimationHAVE YOU EXPLORED THE OPTIONS BEFORE COMMITTING?

treating all estimatesas commitmentsNOT APPROPRIATE

UNTIL WE GET BETTER AT ESTIMATING

practicing estimationwithout deliberatenessPRACTICE ≠ DELIBERATE PRACTICE

REF. DEFINITION OF INSANITY

not experimentingIF ESTIMATING DOESN'T DO EVERYTHING YOU NEED

HOW ARE YOU TRYING TO IMPROVE?

overvaluingestimationSTOP FIGHTING OVER PRECISION

YOUR ACCURACY DOESN'T MATTER EITHER

deliberate deceptionTHE MACHIAVELLI FACTOR OR OVERLY SUBSERVIENT

BOWING TO POLITICAL PRESSURE

assuming youneed estimatesMAYBE YOU DON'T?

HOW WOULD YOU FIND OUT?

#NOESTIMATES

Bad habit number 1

Final Questions?TWITTER - @JITTERTED EMAIL - TEDYOUNG@GMAIL.COM

HTTP://ABOUT.ME/TEDMYOUNG

Resources 50 Quick Ideas to Improve Your User Stories (In-progress Leanpub e-book)

◦ http://leanpub.com/50quickideas

Throughput (Blog entry by Zsolt Fabók)◦ http://zsoltfabok.com/blog/2014/01/throughput/

Intuitive Prediction (1977 paper by Kahneman & Tversky about the "planning fallacy")◦ http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA047747

Judgment under uncertainty (1974 paper by Tversky & Kahneman)◦ http://www.sciencemag.org/content/185/4157/1124.short

Delusions of Success: How Optimism Undermines Executives' Decisions (Lovallo & Kahneman)◦ http://j.mp/delusions-of-success

Modeling, Simulation & data mining (Slides from Troy Magennis & Larry Maccherone)◦ http://j.mp/simulation-magennis

Estimates in Software Development (Dubakov, May 2013)◦ http://www.targetprocess.com/articles/estimates-software-development.html