“Missing Requirements”
10 hours ago
Wednesday, August 12, 2009
End of Round One
Why this question:
To talk about usability of metrics.
Idea Behind:
Theoretically all aspects of any given process cannot be measured. Complete controllability would mean that all possible states could be achieved from all others states which of course is also practically impossible.
So what and how can be measured? Measuring can be qualitative or quantitative.
Example:
The speedometer of the car is needle pointing at number 60 – so you know that the speed of the car is around 60 km/h. You do not know exactly due to the number of small factors. You check the GPS and see that it shows 59,1 km/h.
You are very hungry, typically describable as “ I could eat a horse”. You do not know how hungry you are but you know it is very hungry and typically you know you’d have to eat 3 Big Macs to be satiated.
In both cases you can take control of the situation – you can brake if you see a sharp corner coming up and you could go and eat something.
The point is, that even if you cannot quantitatively measure the process you can still take rational measures to control the process.
We cannot exactly know how much we have tested or how many bugs are in the software – but based on the knowledge what we have done and previous experience we can make reasonable decisions on what to do next.
Below are the answers i received:
Tester 1:
b. - Process is controllable if it is measurable and controllable. Measurable means that we can gather useful information from particular situation (how was it done, what happened) and controllable means that we can reproduce that situation with certain action and it was not generated accidentally.
Even with simple process we can not assure that all variables are considered and because of that we can not state that our process is completely measurable or controllable.
Tester 2:
a) … it is measurable
If there is nothing that can be measured in the process then there is no way to control it.
Tester 3:
b) ... it is measurable and controllable
For the process to be controllable you'll have to be able to control it with an input signal. You will also have to be able to measure its parameters to determine if the process has achieved the desired state.
The control & measuring processes do not have to be absolute. In fact - they cannot be. There will always be a measurement error and therefore the current state of the process can only be estimated from a series of measurings.
The required accuracy for measuring depends on the process. Tolerance for measurement error is much lower in nuclear power plants than say... consumer thermometers. However both of these processes are controllable.
Tester 4:
My choice is b).
In real world only very limited processes can be completely measurable and controllable. Measuring is not enough to achieve control over a process.
Tester 5:
A-Process is controllable in range where it's measurable. It's not necessary that process should be completely measurable. B and D are not a correct because 'controllable' cannot be in same time a precondition and a result. (Similar example: Shape is _square_ if it is _square_ and have 4 right angles.)
Tester 6:
Correct answer is B. Process must be measurable to get the info
about it and controllable to be able to change the input based from measurement info.
The correct answer can not be D because no process can be fully measurable or controllable because of faults, time constraints or lack of accuracy.
Tester 7:
Process is completely controllable if it is completely measurable and completely controllable. If the process is fully measurable and controllable, the control system can use signals to send process to a desired state from any of the proccess's states, including initial state.
To talk about usability of metrics.
Idea Behind:
Theoretically all aspects of any given process cannot be measured. Complete controllability would mean that all possible states could be achieved from all others states which of course is also practically impossible.
So what and how can be measured? Measuring can be qualitative or quantitative.
Example:
The speedometer of the car is needle pointing at number 60 – so you know that the speed of the car is around 60 km/h. You do not know exactly due to the number of small factors. You check the GPS and see that it shows 59,1 km/h.
You are very hungry, typically describable as “ I could eat a horse”. You do not know how hungry you are but you know it is very hungry and typically you know you’d have to eat 3 Big Macs to be satiated.
In both cases you can take control of the situation – you can brake if you see a sharp corner coming up and you could go and eat something.
The point is, that even if you cannot quantitatively measure the process you can still take rational measures to control the process.
We cannot exactly know how much we have tested or how many bugs are in the software – but based on the knowledge what we have done and previous experience we can make reasonable decisions on what to do next.
Below are the answers i received:
Tester 1:
b. - Process is controllable if it is measurable and controllable. Measurable means that we can gather useful information from particular situation (how was it done, what happened) and controllable means that we can reproduce that situation with certain action and it was not generated accidentally.
Even with simple process we can not assure that all variables are considered and because of that we can not state that our process is completely measurable or controllable.
Tester 2:
a) … it is measurable
If there is nothing that can be measured in the process then there is no way to control it.
Tester 3:
b) ... it is measurable and controllable
For the process to be controllable you'll have to be able to control it with an input signal. You will also have to be able to measure its parameters to determine if the process has achieved the desired state.
The control & measuring processes do not have to be absolute. In fact - they cannot be. There will always be a measurement error and therefore the current state of the process can only be estimated from a series of measurings.
The required accuracy for measuring depends on the process. Tolerance for measurement error is much lower in nuclear power plants than say... consumer thermometers. However both of these processes are controllable.
Tester 4:
My choice is b).
In real world only very limited processes can be completely measurable and controllable. Measuring is not enough to achieve control over a process.
Tester 5:
A-Process is controllable in range where it's measurable. It's not necessary that process should be completely measurable. B and D are not a correct because 'controllable' cannot be in same time a precondition and a result. (Similar example: Shape is _square_ if it is _square_ and have 4 right angles.)
Tester 6:
Correct answer is B. Process must be measurable to get the info
about it and controllable to be able to change the input based from measurement info.
The correct answer can not be D because no process can be fully measurable or controllable because of faults, time constraints or lack of accuracy.
Tester 7:
Process is completely controllable if it is completely measurable and completely controllable. If the process is fully measurable and controllable, the control system can use signals to send process to a desired state from any of the proccess's states, including initial state.
Tuesday, August 11, 2009
First Question
Only then is a process controllable if:
a) … it is measurable
b)… it is measurable and controllable
c)… it is completely measurable
d)… it is completely measurable and completely controllable
a) … it is measurable
b)… it is measurable and controllable
c)… it is completely measurable
d)… it is completely measurable and completely controllable
Conclusion time!
I was pretty cruel on the testers asking them to after 1-month vacation what they remember about the questions and answers. And more importantly what they think was important for them. And did this one-by-one as I told them what had thought of their answers.
Most of them did not remember much from the first look - but after reminding of the questions they seemed to recall the ideas behind them. Which is very good. I did not expect them to remember everything but to recall it when something triggered it.
Due to the voting of the best answers (to win a case of beer) one important trend was noticed:
Which in my mind goes a long way. When situation is complicated and the solutions are all not-so-good and/or confusing - the most understandable version is used.
For example:
When somebody asks You
"Where are the long arm and short arm of the chronometer pointing?"
Would you know to tell the time?
On the whole everyone agreed that these questions are good exercises:
*) Reminding to look stuff from another angle
*) Getting out of daily routine
*) Making their mind work on different things than usual.
I hope this has been good experience also to those following this blog.
New questions are coming soon...
PS. Changes for new "season" no winner will be selected each time, all votes that testers answers get over the 8 questions will be counted and whoever gets most - wins.
(Secret vote, naturally).
Most of them did not remember much from the first look - but after reminding of the questions they seemed to recall the ideas behind them. Which is very good. I did not expect them to remember everything but to recall it when something triggered it.
Due to the voting of the best answers (to win a case of beer) one important trend was noticed:
"The winning answer does not have to be the most correct on, but it needs to be very well expressed."
Which in my mind goes a long way. When situation is complicated and the solutions are all not-so-good and/or confusing - the most understandable version is used.
For example:
When somebody asks You
"Where are the long arm and short arm of the chronometer pointing?"
Would you know to tell the time?
On the whole everyone agreed that these questions are good exercises:
*) Reminding to look stuff from another angle
*) Getting out of daily routine
*) Making their mind work on different things than usual.
I hope this has been good experience also to those following this blog.
New questions are coming soon...
PS. Changes for new "season" no winner will be selected each time, all votes that testers answers get over the 8 questions will be counted and whoever gets most - wins.
(Secret vote, naturally).
Subscribe to:
Posts (Atom)
