Re: Logarithmic tail
The last 30% is always likely to be harder; I would not expect the new predictive line to be accurate either.
If I was organising it (thankfully not) I would probably start with the major population centres where technology is concentrated and leave the outliers until last to try and provide communications to the majority of users as soon as possible. Like the 90% plus coverage in the UK.
It is interesting that the graph prior to the report was almost a straight line; this could mean that there was a team running flat out at maximum efficiency and maintaining a consistently high fix rate.
It could also mean, of course, that daily reporting was too much of a faff so they just fudged the figures. In which case the El Reg report may have caused someone to take a bit more time checking the figures.
For the conspiracy theorists there may have been deliberately optimistic reporting to hit bonus targets and the El Reg report may have hit around the time when they always planned to flatten out the curve after getting the maximum money up front.