An estimate is a guess. A judgment about how much time it will take to do a software task.
Ask questions, read about scope, compare to past work, ask when they want it, what are the risks, provide worst case guess.
There are going to be more costs and it's your fault.
Why do you want the estimate? What will you do with it?
Customer collaboration over contract negotiation.
Most important thing is honesty. I don't know.
Accurate vs precise. Between now and 20 years from now. Accurate. How much confidence do we have? A lot.
When you look underneath, what will you find?
Destructive inspection. Then they discover what is needed. They brought in a temporary kitchen. Then you can decide on cost.
If too wide, can you sharpen it up?
What is the confidence level? Worst case, best case, average case.
If project is over we can cut scope. We're cutting assurance of quality by cutting test. We're cutting workers quality of life. Best ideas when waking up, walking, when not stressed. We're cutting the stuff we haven't gotten working yet.
Average case is estimated by breakdown. 82 features. Then break it down more. Making more guesses. Then buffer it by adding more time. What about the things we don't yet know? What about wait time on other blockers. Elapsed time or work time? We do not have a way to estimate. It's all pretend. Null + 4 is null. Null * 8 is null. How can we get more precision? We can break down more. We can just do the work and be precise. Meanwhile we've burned time in estimating.
All that estimating just bought us comfort. It is an illusion. Lie with statistics.We have guessed about the unknown unknowns.
We helped someone make a decision based on a lie. What decisions? If this does not serve to help make a good decision we need to make different decisions.
What is the alternative?
Agile is a whole different way to make decisions. Are we good at doing this? Is it worthwhile? Valuable? Stop when it's no longer valuable. We need a way to decide what to start and when to stop.
Ask questions, read about scope, compare to past work, ask when they want it, what are the risks, provide worst case guess.
There are going to be more costs and it's your fault.
Why do you want the estimate? What will you do with it?
Customer collaboration over contract negotiation.
Most important thing is honesty. I don't know.
Accurate vs precise. Between now and 20 years from now. Accurate. How much confidence do we have? A lot.
When you look underneath, what will you find?
Destructive inspection. Then they discover what is needed. They brought in a temporary kitchen. Then you can decide on cost.
If too wide, can you sharpen it up?
What is the confidence level? Worst case, best case, average case.
If project is over we can cut scope. We're cutting assurance of quality by cutting test. We're cutting workers quality of life. Best ideas when waking up, walking, when not stressed. We're cutting the stuff we haven't gotten working yet.
Average case is estimated by breakdown. 82 features. Then break it down more. Making more guesses. Then buffer it by adding more time. What about the things we don't yet know? What about wait time on other blockers. Elapsed time or work time? We do not have a way to estimate. It's all pretend. Null + 4 is null. Null * 8 is null. How can we get more precision? We can break down more. We can just do the work and be precise. Meanwhile we've burned time in estimating.
All that estimating just bought us comfort. It is an illusion. Lie with statistics.We have guessed about the unknown unknowns.
We helped someone make a decision based on a lie. What decisions? If this does not serve to help make a good decision we need to make different decisions.
What is the alternative?
Agile is a whole different way to make decisions. Are we good at doing this? Is it worthwhile? Valuable? Stop when it's no longer valuable. We need a way to decide what to start and when to stop.
No comments:
Post a Comment