Thursday, January 25, 2018

Optimizing queue size

Stripe, a payment company, complaining about bitcoin.  I do the theory
 (By the time the transaction is confirmed, fluctuations in Bitcoin price mean that it’s for the “wrong” amount.) Furthermore, fees have risen a great deal. For a regular Bitcoin transaction, a fee of tens of US dollars is common, making Bitcoin transactions about as expensive as bank wires

OK, we have volatility over the transaction interval and the implied congestion price.  Where is the optimum? Price it so there are exactly two in your queues, and multiple queues make a structure, a generator. We are back to the model of a Walmart checkout, if you have "five or six items, take line 3', and the pit bioss can alter the number items per basket to keep all queues equally at two customers.  The queues are really the match of clerk queues and customer queue, the implied customer wait is 1.5, the implied clerk wait is .5; their combination meet the bandwidth conditions, Shannon match in the information sense where both SNR and Signal we quantized to provide the most flow with the least number of transactions.

The problem is similar to Uncle Milt's thermometer analogy, the system quantizes basket size in units of measurement uncertainty. We get a set of relative pricings, nice quants to pack a sphere in remembrance of a thermometer. Those bitcoin quants, right now, seem to fit the price setting needed to plan a hotel and flight for a few days. Folks would wait yen minutes, and pay a $30 dollar fee if they could share in the gains of a central bank neutral coin for trip planning.

No comments: