[ppml] Combining Forecasts

[Geoff Huston]

michael.dillon at bt.com wrote:
>> This is a recent change in Huston's predictions. A few months 
>> difference from his previous prediction in July.  Google 
>> cache of this page from Aug 4, shows April 16th, 2010.  Tony 
>> Hain now says April 2010.
>>
>> This seems to be a lot of variation from July, when both said 
>> March 2010.  This deserves further investigation.  I suspect 
>> that in Huston's case, if he doesn't update the data from 
>> ARIN, etc, his program will see this as no new allocations 
>> and compute a new (but wrong) exhaustion date. I'll see if 
>> this is the case.


The model used for http://ipv4.potaroo.net is based on daily data
gathered from the 5 RIRs, the IANA IPv4 registry and the Ipv4 BGP
routing table. The prediction changes over time as the predictive
model is a sliding window model that uses data collected over the
last 1095 days as the foundation of the prediction. In recent months
the pace of IPv4 consumption has slowed down a little so the predictive
model stretches out the predicted dates a little further  - perhaps 
related to the summer break in the northern hemisphere?

regards,

   Geoff