Regarding your idea of a greater number of active users in staking, it is an interesting
hypothesis to explore, but before I could answer, I have to analyze an intermediate
file of 73 MB, now I do not think I have time.
My hope is I can give you some data tomorrow.
As promised here is the new analysis on the number of addresses that generate
block, day by day.
The method by which it is carried out influences the results obtained, so here's a
brief description of how it was conducted, sorry to be a bit boring.
The files of 73 MB of which I wrote yesterday, is a list of the blocks produced by
block 1.4M up to 1,669,266, the last block of the day 26/07/2015 (UTC time).
The blocks of the next day accounted for only a part of the day, and then I have
not included here.
The file has been sorted for the block date [1], and every day, were counted the
blocks (from which to derive the average block interval), and the addresses which
have generated at least one block in that day.
The extracted data are fairly coarse, their resolution is in fact a whole day, but I
hope that makes the idea, consider them only as preliminary data.
Perhaps it is too coarse for a precise analysis of cause and effect, if you need a
more precise let me know.
In this period of time, the number of addresses that have generated blocks is varied
between a minimum of 124 and a maximum of 162, also quite a few after all.
The number of blocks produced daily was between 3532 and 8298, far too many,
considering that should be 2880.
The average interval between the blocks is gone from a maximum of 24.5 to 10.4.
At first glance, there may be similarities between the tracks, but if we look
carefully we can see that the block interval in some cases has been dropped
even when the number of users was practically stable, or even decrease.
Perhaps it might be more likely an inverse relationship, I mean, when the block
time interval and PosDiff fall (see chart from yesterday), then the addresses
which typically fail to get into the game, producing a few small block more often.
And so shortly after the fall we see the effect of an increase in the number of addresses.
In summary, your hypothesis that a greater number of users has resulted in an
acceleration of the product cycle of blockchain not seem sufficiently supported by
data, but this is just my interpretation of these coarse data, not necessarily the
correct one.
[1] Small note:
In early attempts to process the data, I did not do the sort of lines based on the
time of the block, and I got strange results, such as "days" that contain only 4 or
5 blocks, followed by other blocks belonging to the previous day, then some of
the next, and again the previous one, and finally a series of consistent data.
This odd behavior at the edge of date change, I would have expected
for 1-2 blocks, but 4 or 5 jumps one after another seem far too many,
abnormal, rather symptomatic.
As an example you may try to control the sequence of blocks between 1652846
and 1652867.
This oddity is evident during the date change, but I think it also occurs
during the rest of the day and days.
Bye Bye
