shermanmccoy

Yes! You can control for human risks in the tail scenarios. You can't control for the tail scenarios in AI because you don't know what on earth they are, they are a nasty surprise, thrown up by a model which, in the tail, we as engineers don't fully understand

He only resubmitted 10 days ago. Presuming there was something missing with his first application, something like this would need to go through a legal area internally at Amazon to get a considered opinion. Maybe they are busy?

Best of luck to OP.

>they're selling them on USDT-USD markets.

My hypothesis is they get dripfed onto these markets so as not to spook them. Or just transferred off to collateralize other trades - hey it's free money to them, why not set up another trade elsewhere while they are in?

It is not safe to assume anything based off market price action. You seem to be assuming there is some essential relationship between volume traded and price. These are market prices, they can be totally irrational, especially if there is some fake news (Tether is fully backed) behind them.

I think the ponzi continues to work for now, the arbitrageurs can cash out, while the illusion that tethers are backed prevails. So indeed if no one does redeem these USDT as I keep reading could the arbitrageurs just slowly buy the cross to USD over a period of time after the trade? Alternatively USDT seems to be a proxy-currency to move across exchanges and networks, so it may indeed be transferred off to other pairs/tokens to fund alternative trades...

Thanks, that is very enlightening. So Tether are all about maintaining the apparition of a 1:1 backing of USD to USDT. It's the lynchpin of the whole scam. Otherwise the step above you mention around the arbitrage bots buying btc-usd no longer occurs, and the game is ruined for them.

Do you have some modelling, or any credible reference to back up your assertion that the energy used by banks would be higher than a Bitcoin blockchain?

Ah good insight, thanks.

Precisely! So picture some, however oversimplified, code which searches a customer's orders, determines which is a phone product id, remembering they sell other stuff which we would not want to update in flight order details on, then updates the embedded customer address on the order.

I would like to hear some original thoughts on why this is so difficult that a human is doing this work instead.

Obviously maintaining a different behaviour for a class of product ids is painful and not beautiful code to look at, but it is the real world.

While I disagree with the positions put forward by lolc and flukus, that Librem 5 should have the customer address modelled analogously to some 'fast moving goods' e-commerce implementation, I appreciate their comments for being constructive and actually adding something to the discussion.

Well the Librem 5 is not another 'widget in a warehouse'. It was always going to be a long wait to delivery, due to the enormity of the undertaking. This was not a surprise to anyone. Addresses will need to be updated after order in this case! Every implementation is different.

It's precisely the "this is the way it's done" approach you mention, which means they now have to pay a human to do the work their web stack ought to do!

I smell a NoSQL database keeping those addresses.

Cheers

Two simple questions.

How long does the battery last, while largely sitting on my desk not being used?

Can it make and receive calls and texts?

Does screen sharing fall into one of those categories?