Home | Login | Recent Changes | Search | All Pages | Help

SessionTwo025

C25. Exploring Tradeoffs: Quality versus Speed
BobKing, SteveSmith, JerryWeinberg (SessionTwo025)

Learning Objectives

• Experience how teams make tradeoffs
• Contrast different approaches to making tradeoffs
• Formulate strategies for changing traditional approaches
• Experience the interplay of logic and emotion in making tradeoff decisions

• The different teams all converged well on optimum "points" for the time-quality trade-off. None chose poorly on delivered quality.
• As in projects, there was some "luck of the draw".
• Explicit value/penalty costs both for quality and for time made trade-offs effective. (We discussed getting customers to supply value metrics for schedule points.)
• Starting "dirty" (as in quick & dirty into testing) was costly in overall score! (What would you pay to improve your testing costs now, sir?)

Everyone had a good time! --BobLee 2002.11.08

• ask what it is worth to ship on this date, then what the drop off is.
• the kind of calculation used in the simulation can be used to see if it is worthwhile to be fast.
• if you are told you must and you don't believe that you can, ask what will happen if you don't.
• don't say it can't be done, say you don't know how. (There is a lot more about this in No BOF)
• whether you are an employee or an external resource, you need to be able to say No or to be willing to walk away.
• if the schedule is too tight, do an explicit plan, then ask what they want to drop. If they are not willing to drop anything, keep reporting the results.
• if you do ship because this is the ship date and the product is not ready, survey the users after a few weeks and report their feedback.

SherryHeinze 2002.11.11

• Ask "What is that worth?"
• As long as there is planning, management can make tradeoffs
• Early part of process is to get info rather than fixing
• Simulate automated test by upfront cost of 5 chips
• Be careful with automated tests because they continuously test the same thing
• It's long been known in hardware testing that less defects are esier to fix. This knowledge doesn't seem to have reached software people.
• Try to get the best before you start
• "What are you going to do, hit me?"
• Acknowledge people's desires
  • It would really be great...
  • I don't know how to do that
• Tell me what you don't want me to do on this explicit plan
• Often management doesn't have the information to make a decision. Bring them the information

Team Orig; Ship; #Tests; Score
Team6 18,2; 19,1; 05; 800K
Team5 12,8; 17,3; 14; 240K
Team4 15,5; 19,1; 15; 480K
Team3 17,3; 20,0; 08; 900K
Team2 17,3; 20,0; 11; 800K
Team1 15,5; 20,0; 17; 600K

SteveSmith 2002.11.18