MtnGoat Member
Joined: 17 Dec 2001 Posts: 11992 | TRs | Pics Location: Lyle, WA |
|
MtnGoat
Member
|
Mon Feb 26, 2007 2:43 pm
|
|
|
guy: the engine design is running 10C too hot because of the emissions controls
boss: where's the data on the engine without it?
guy: we don't have any for the engine since we changed it, and can't get any.
boss: why?
guy: because we're corrupting the baseline data with the emission control, and can't remove it
boss: when did you last get a baseline for the engines normal temp
guy: oh, sometime around say, 1960.
boss: and you claim you know the engine is running hotter than it should because of what we added?
guy: well we're certain our simulation of the temperature it should run at is correct, though we have no way of testing the simulation empirically, we know it's running too hot anyway because the simulator says so.
boss: haven't you iterated the simulator a bunch of times?
guy: yes
boss: how did you empirically verify the simulator output in the last five decades of changes?
guy: we haven't. but our guys are really smart they're certain they don't need to
boss: didn't these smart guys already find errors in the simulations that led to the need for simulator iterations in the first place?
guy: yes, but this time they're sure they're really really right
boss: how much control over the system under test's inputs do they have?
guy: none
boss: how well do they understand these inputs and factors?
guy: well, we have really high certainty on two or three, and low to medium low on the rest
boss: how many is that?
guy: 5 more
boss: you're telling me you have a peak of medium to low understanding for most of the factors influencing the engine and a model baseline you haven't empirically validated for fifty years
guy: yes. It's OK though because our simulators say all those other factors can't make much difference.
Boss: Did you empirically validate that claim?
guy: no, we can't.
Boss: why not?
Guy: because we can't control the inputs and we have a low understanding of how to judge those areas anyway.
Boss: have all these external factors you can't control been changing on their own, understood or not, measured or not, looking for or not, during that fifty years?
guy: yes
boss; OK, this is giving me a headache. let's cut to the chase, maybe it doesn't matter much..how much will these changes based on a simulation you can't actually verify empirically, cost us?
Guy: We need to spend a trillion bucks or fifty and permanently change all our customers lives based on our simulation results and the measurements of the engine as it is, because the simulator says so. We'll need to spend much of that money simply to replace things with other things that have the same output, but create it differently.
boss: won't that mean all that money won't be available for other future investments?
guy: yup
boss: and at the end we'll have much of the same stuff, just done a new way, at the opportunity cost of all the investment in other things that don't just replace what we already have doing it?
guy:yup
boss: OK...Just so we're clear. You have a low to medium low understanding of most of the factors in a system you cannot control.
Guy: yes
Boss: your baseline hasn't been empirically verified for fifty years in spite of numerous software revisions, external factors you don't fully understand control all the inputs, and the areas you don't understand comprise the majority of factors by your own admission
guy: yes
boss: your solution is to OK spending trillions of bucks and permanently reorder the life of every customer on the planet. We'll not spend on all kinds of other new things in order to replace equipment which already does what we need and has already been built, installed, and paid for.
guy: yes
Boss: All this based on simulations you know are right when you were wrong before and had to change the code.
Further, you haven't empirically tested even the latest revision and there is no way to do so today, and you say you have proof we're above baseline.... consisting of simulations for the 'normal' temp of the engine which haven't been empirically truthed in five decades. During which time all the factors changed on their own as you simultaneously iterated your code.
guy: yes, but it doesn't matter because the simulation shows...
boss: get out of my office, you're fired.
Diplomacy is the art of saying 'Nice doggie' until you can find a rock. - Will Rogers
Diplomacy is the art of saying 'Nice doggie' until you can find a rock. - Will Rogers
|