On Mon, Jul 31, 2000 at 12:14:36PM -0500, Weisberg wrote: > "vinton g. cerf" wrote: > > > The trouble is, it is too little, too late - we're already over the > > top in terms of what we can handle in a reasonable time frame, taking > > our funding (now expended) into account. More time is more cost and more > > delay - it doesn't add up. > > It is a bad situation. No good will come of it. Whatever is done, > now, will poison the well in the coming elections. > > Hindsight is clear. The membership drive should have occurred sooner > and there should have been a "shake-down" cruise before the first "race," as> had >been suggested. You mean like the first couple of months, while the system worked fine? > That water is under the bridge. > > So, what is the best bad choice, now? > > I suggest extending the deadline; accepting registrations online, > only; and distributing PINs by email. That would be silly beyond belief. Whatever security there may be in the system depends on the fact that pins are shipped to a physical address. Changing that rule midstream would completely alter the requirements for voting, and would certainly destroy whatever validity the system might have. > If necessary, you can bump the election a few weeks. That would cause > minimal harm. Extending the deadline sounds good, but probably won't change things one bit -- it will only delay the period before people start screaming about being disenfranchised. The news about the ICANN elections has been filtering out slowly -- more people will know about it tomorrow than know about it today; the rate of registrations has been increasing steadily, with no upper limit in sight. It is quite possible that the number of registrations could grow to the millions, given more time. Moreover, there are obviously many many people such as Mikki Barry and Dan Steinberg who have known about this for months, but put off registering until the last few days. Heavy load on the system during the last few days is predictable no matter how it was sized to begin with. Finally, a moments thought indicates that the feedback loop is intrinsically unstable: as soon as the system starts to slow down, people start retrying. The retries clog the system even more, which leads to more retries. It gets into catastrophic failure mode very quickly, where the system is completely clogged, and no registrations can proceed. Consequently, it doesn't make much difference how you sized the system, or how long you delay -- similar problems are very likely to occur. Once again, hindsight is wonderful.... -- Kent Crispin "Do good, and you'll be [EMAIL PROTECTED] lonesome." -- Mark Twain
