$40M being unclear isn't, IMO, a reason to discredit the whole data set. On the contrary, it's a reason to go deeper and explore the data further.
a) Isn't that an assumption? The data simply shows the net result of inflows and outflows on that account at given periods of time based on USD valuations of SBD and STEEM. Since we know that account is acting (at least in part) in an exchange capacity where people buy and sell STEEM/SBD for other currencies, then I don't see what the problem is. It sounds like you want 100% transparency before any of these numbers are meaningful
b) As I see it, the other numbers may be off only if an account is used as an exchange account but isn't included in the query.
I think you're looking for comparative meaning between accounts, and that meaning may not be there. The data for the blocktrades account is what it is. If blocktrades wants to release their own public financials for how much of their activities are not exchange-related, things might be improved, but I don't think that fundamentally changes how, for example, the binance exchange is being used.
Well we just disagree (also it isn't the only reason as I noted). IMO $40 million is just too large of an uncertainty. If it were $400 and we didn't know what was going on then I would say, sure $400 = small error bound. $40 million = large error bound.
That's fine, but it would make this a work in progress with huge uncertainties. Nothing wrong with that.
It may to the extent that amounts sent to/from binance are aggregated with amounts sent/to from blocktrades (which may or may not be exchange), and are disaggregated with other unknown exchange accounts.
You didn't break down account X sent $Y to binance. That would be a totally objective and accurate number. You aggregate account X set $Y to all 'known exchanges', including binance. Example. Account sent $100 to binance but received $200 from unknown exchange. The amount shown is -$100 but the 'real' total sent to exchanges (known and unknown) which is the actual economically meaningful figure is actually $100.
I would expect that it may never be possible to 100% identify all exchange accounts and activity but the idea is clearly to get as close to that as possible. This isn't close.
I've been tracking exchange accounts for over two years. I've had an open call to the community to contribute known exchange accounts to improve the reports I do. Many have contributed. It seems to me like you're using the blocktrades situation to argue there are many shadow exchange accounts on Steem, and I just don't see it that way. Again, wallets like Vessel have known exchange accounts also.
If you go right now and do a deposit or withdrawal from any major exchange and note which accounts are used, you'll see the accounts I've used in the code. Yes, those exchanges may have other hot and cold wallets and such, but ultimately that doesn't matter much to me as that would essentially be internal accounting anyway. The transfers to the public-facing exchange accounts are what matter most as constituting a deposit or a withdrawal.
Maybe we should take this conversation a different direction. Do you think the numbers associated with your accounts are inaccurate? If so, let's discuss them specifically and see if we can find improvements to the code to be more accurate. All the data is there.
You're saying "this isn't close" but you seem to only be using one account to make that claim (along with some hypotheticals). What if all the other accounts are close?
As I said in the post, I'm fully open to this data being wrong. What I'm not as open to is claims it's wrong without evidence relating to specific accounts.
One thing I could do is track power ups on the blocktrades account to other accounts. That would probably bring the $40M number down a bit for all those who bought Steem Power using SBD or Steem (something that would otherwise look like a withdrawal to an exchange).
I honestly don't know. It would be a significant effort for me to track down all of the accounts and correlate it with my exchange records. Again, I'm hardly disputing that your list totals the transfers to and from the listed accounts. I'm just less certain about what that actually means.
They may be they may not be. I don't see how to determine that with $40 million of potential error. I understand where you are coming from but with a lot of experience analyzing large amounts of data, and unfortunately having learned the hard way how hidden errors can seep in, I have learned to become somewhat paranoid. When I see something surprising or unexplained, my reaction is to try to dig in and understand it and not reach conclusions until and unless I can do so. I see too many unexplained loose ends here to be comfortable with the model.
I gave you evidence regarding the steemit accounts. They have had almost three years of significant headcount, business and web site operational costs (the latter reported at $2 million/year at least recently), probably capex, etc. as well as stating that all of their funding comes from selling Steem (apart from I believe a modest amount of seed funding pre-launch) and that they have accumulated additional fiat which is claimed to be a sufficient runway and is being used to cover some costs now (beyond current revenue from Steem selling). And further one can get similar back-of-an-envelope estimates coming at it from the opposite direction and looking at their rate of stake reduction (where it go?). All of that does not add up with the numbers here. I consider that a basic sanity check.
Maybe @blocktrades can comment on that. I've seen a number of transfers involving @alpha that I do not understood which may or may not relate to blocktrades providing services to Steemit outside of exchange services. Here's an example of almost a million STEEM last month:
last month Received 942325.000 STEEM from steemit Owed payment to Blocktrades (@alpha)There are many more.
Tracking transfers on Steemit, inc controlled accounts would be quite interesting.
I appreciate your skepticism and bringing up concerns about how the data is interpreted and what conclusions people come to. My hope in presenting this information is to get people thinking about these numbers and about the rewards system itself.