Could it simply be people switching to/from some popular nightly item on linear broadcast TV from/to whatever streaming they were consuming ? Noting the "Coronation Street kettle" effect of years gone by, if the power utilities publish real-time consumption stats with enough time resolution that may also be interesting to examine.
Having a look at what DNS queries are being sent to ISP resolvers around that time may also yield insights. Keith On 5/12/20 6:16 AM, Neil J. McRae wrote: > All things being equal Clive I agree give or take a small plus or > minus, but the operators have different market shares and not all of > the population is a mobile phone subscriber or a smartphone user. > I'm going to get a more detailed view of 9pm as the views I'm looking > at are over 24 hour period and possible to miss something. > Understanding the size of the gradient change of the dip would be > helpful. > On 12/05/2020, 11:10, "Clive D.W. Feather" <[email protected]> > wrote: >> Giles, interesting correlation ??? would be very interesting to >> find out if this was the cause. I???d find that quite surprising if >> it was the case but very interesting. 5% of the population might >> fit into this bracket but I???d expect them to be shared between >> all the operators perhaps making the size of the population on each >> operator quite small, and not sure large enough to make such >> something noticeable. > > Actually, it should be about 5% of the population on each operator, > assuming there's nothing (such as geographical distribution) to > disturb the relationship. Yes, they have to be shared out, but so > does the population!
