[ppml] Policy Proposal: Decreasing Exponential Rationing ofIPv4 IP Addresses

[Ted Mittelstaedt]

>-----Original Message-----
>From: Stephen Sprunk [mailto:stephen at sprunk.org]
>Sent: Wednesday, August 22, 2007 8:02 PM
>To: Ted Mittelstaedt; Dean Anderson
>Cc: ARIN PPML
>Subject: Re: [ppml] Policy Proposal: Decreasing Exponential Rationing
>ofIPv4 IP Addresses
>
>
>Thus spake "Ted Mittelstaedt" <tedm at ipinc.net>
>>>From: Dean Anderson [mailto:dean at av8.com]
>>>
>>>Some people think that ARIN staff doesn't have enough time to
>> look as hard as they'd like.
>>
>> ARIN has what - a surplus of 10 million bucks in the bank right
>> now?  And they can't hire more people?  Review of applications
>> can be done in parallel it is not a serial process with everything
>> funneling through 1 person.  Adding more people makes it go
>> faster.
>
>That depends on your definition of "faster".  Processing a request
>takes the
>amount of time it takes; that latency is unavoidable.  Throughput can be
>increased to a point by having as many analysts as there are concurrently
>outstanding requests, and that will reduce the latency due to
>waiting for an
>analyst to get to yours, but that's it.
>

Yes, of course keep in mind here I was
refuting the OP's assertion that ARIN's staffers don't have enough
time to review applications.

Let me write out the explanation in long hand to make it more clear,
if ARIN is indeed chronically overloaded as the OP asserted, then they have
a
backlog - since they have a lot of money, they can hire more people
and AFTER they are trained, the work will go faster.

You can draw some obvious conclusions from this.  Since ARIN can hire
more staff, and are NOT doing so (at least as far as we know) then
they obviously do NOT have a backlog.

As for the argument that adding more gerbils clogs up the wheel, sure
that is true - which is why if you have an immediate and unexpected
crisis, throwing more people at it is the worst thing you can do.
It is, however NOT true if the problem is chronic.  If there's 5
men's worth of work to get through in the day and you only hire 4
men, then you either have a backlog that gets bigger and bigger, or
you do a half-assed job on the work you got.

For a real world example of this check out Ford Motor Company's
lead time on their new 2008 model trucks.

>My understanding is that the vast majority of time is spent trying
>to figure
>out whether someone's request meets policy or not, particularly for
>first-time requestors, because of insufficient documentation.  If
>we want to
>reduce that part of the latency, we need to fix the policy that created
>it -- not hire more people to perpetuate it.

Or simply toss the application back to the requestor and tell them
to do it over and this time read the instructions rather than watching
the movie and reading Cliff's Notes?  It works in high school...

Ted