[Just when you think all hope in a thread is lost, along comes a post with Clue] On Mon, Mar 07, 2005 at 03:10:37AM +0000, MJ Ray wrote: > Matthew Palmer <mpalmer@debian.org> wrote: > > On Mon, Mar 07, 2005 at 02:09:26AM +0000, MJ Ray wrote: > > > Arguing in favour of so-called "positive discrimination" is just > > > another case of ignoring present crimes by past-persecuted people. > > > We have to learn from the past and overcome mistakes, not relive it > > > with everyone swapping places each iteration in a vicious circle. > > An overly damped system is just as useless as an underdamped one. > > This might be a good analogy. I thought so too. > We're trying to reach a known equilibrium, but we don't know how much > damping is required. If I remember my physics, it can be fiddly to get > exactly right damping and our measuring sucks on this anyway. I don't > think we can measure damping well enough to aim for critical damping. Knowing all of the parameters of the system and the desired system characteristics, you can apply an appropriate amount of feedback (whoops, my EE background just slipped out) to produce the correct damping ratio. The correct feedback to apply depends on what you want to achieve -- quickest to reach zero, no overshoot, etc. > If we apply too little, then we never reach balance, but > the sum over all time can be zero repeatedly, yet only for > an instant each time. If we apply too much, then we tend to > balance/zero more quickly but the sum is non-zero. Too much damping means that you won't overshoot, but you'll take longer to reach the zero point, which isn't so good for the people depending on reaching equality. > Which side you err depends whether you prioritise balancing in the future > or cancelling out the past. I favour balance. Understandable, since you (and I, for that matter) have a lot to lose should there be overshoot. - Matt
Attachment:
signature.asc
Description: Digital signature