Quote from: gmaxwell on June 11, 2016, 05:18:54 AM

Quote from: rizzlarolla on June 10, 2016, 05:50:19 PM

If average tx is twice median tx, will this not lead to under paying tx fees?
(i am asking 21.co also)

The 'average' tx mostly doesn't exist. The sizes of transactions are highly multi-modal. There are a whole lot of small ones (most ones people make) and a decreasing number of larger and larger ones that drag up the average. If you're joe-schmoe and make a transaction, it's most likely you'll make a median sized one.

It's like saying the the average number of testicles an American has is 1... and yet relatively few people have exactly one testicle, even though it's the average.

[In no case should fees be under-paid... in competent wallets fees are configured per unit size and set accordingly.

]

Another way to think about it is that "a transaction" isn't really a great unit of capacity. Consider, what if everyone started using 4-party non-amount-matched coinjoins for all their transactions. We'd end up with 1/4 the number of "transactions" each of 4x the size. Would it be right to say that suddenly the Bitcoin network had 1/4th the capacity because now the tx were 4x larger? No-- they're also doing 4x as much.

You could say this only if you know the characteristcs of the tx size distribution (of the future!) or at least the first moments. the average or mean as you know is a very good estimation while the median is just the mid of all sizes and a just bit more stable against outliers. And the density is highest where?