I don't really disagree with a lot of your analysis Zimmah - yet I am a bit concern about part of your premise about addresses.
The longer someone owns bitcoin and the more that s/he uses bitcoin, the more addresses s/he creates, and even some older and more sophisticated users advocate using a different address for each transaction.
I've been using bitcoin about 2.5 years, and I can also be a witness to this kind of address generation. At first I only had a few places that I held bitcoins, and I barely did any transactions. My use of bitcoin has increased. I think I do some off chain transactions with various (third party holder of my coins) , yet I probably generate a new address on average about once every 3 or 4 on chain transactions. I estimate that I've probably generated more than 40 bitcoin addresses through my various usages in such a short period of time, and some addresses hold more bitcoins than others, and several addresses have fraction of coins remaining.
what's the thing that concerns you about that?
The inaccuracy of the data or are you worried we run out of addresses?
The data of course is not 100% accurate, but it gives us a general idea about roughly how many people own a certain amount of bitcoin (and it's also roughly what you would expect according to zipfs law and wealth distribution).
But lets say for example that on average an user use 50 wallets and spreads his coins between them.
If an user has 10,000 bitcoin, than he could spread them in many different ways, but if he has like >100 in each wallet, then all 50 of his wallets would show up as 10 bitcoin wallets.
That would take 50 wallets out of the 10 BTC owners table to combine into 1 large wallet of 10,000 bitcoin.
Let's do something crazy and let's say we apply this to EVERY wallet (of course this is not true, but let's just assume that it is).
So let's say everyone in the below chart actually has their money spread around in 10 wallets and for simplicity sake lets assume the money is spread evenly between those wallets. (let's ignore the 3 mega-wallets, they don't really count).
(quoted the numbers for convience) This is the chart as it is.
Now let's see what the converted chart would look like
As you can see, this barely changes the numbers at all, only the extremes (>0 BTC, 1BTC and 100,000 BTC change significantly (and those are outliers)).
Most likely, the truth is somewhere in between those values. So we can expect there is somewhere between 725,000 and 7,2 million unique bitcoin users, and between 60,000 and 130,000 of those have 10 or more bitcoin, and between 500,000 and 750,000 people own at least 1 full bitcoin. the other numbers are pretty similar in both scenarios so they are most likely roughly in the right ballpark already. (give or take a few %). (In reality obviously the people with less than 1 bitcoin most likely only use 1 wallet, that's why the lower end of the table is more inaccurate)
This is of course all pretty rough calculations but as you can see, splitting the money up doesn't matter that much for the statistics overall. Because for every couple of wallets combined to create a larger wallet, you will only move wallets between different classes and the whole chart will just correct itself and end up nearly exactly the same anyway.
We also won't run out of addresses because there's just so many possible addresses.
2^160 to be precise, and to put that into perspective for you, if 8 billion people would generate 1 billion addresses per second it would take 5.7 SEXTILLION
years before we run out of adresses. That is nearly half a billion times the calculated age of the universe.http://www.wolframalpha.com/input/?i=2%5E160+%2F+(8+billion+*+billion+per+second)