Today, customers say what they want (good). Then engineering estimates it (OK), then that estimate because the basis for cost calculation and scheduling, usually bound into a contract (inflexible = bad). He argued for the customers communicated vision, permissible cost & time, then the deveopers should convert that vision into reality -- something which excite the users as much as possible -- given the budget constraints.
So, instead of estimate the project at a 6 months and 1 Million CHF, allocate 4 months and 500'000. Build what you can. Examine the results. If it's good, release it, if it needs work, do more until your happy. The customer stays in control, but over the total costs and when release occurs.
This was really only a small fraction of his talk. The presentation is now online -- and for those of you who will be in Toronto, he will give the final version in the 'chanson française' track.

0 comments:
Post a Comment