Share good ideas and conversation.   Login, Join Us, or Take a Tour!
wasoxygen's profile
wasoxygen




stats
following: 57
followed tags: 61
followed domains: 17
badges given: 21 of 23
member for: 2232 days
style: office


tags used



comments 39
wasoxygen  ·  link  ·  parent  ·  post: Sam Altman: E Pur Si Muove

This reminded me of our great discussion about slavery.

    GM Hikaru Nakamura ... called the match "dishonest" and pointed out that Stockfish's methodology requires it to have an openings book for optimal performance.

I agree that's a significant handicap, and it would seem fairer to pit the algorithms against each other at full strength, running on equivalent hardware.

Still, it would merely be an exhibition to see which bot is farther along on the road to divinity.

    "I am pretty sure God himself could not beat Stockfish 75 percent of the time with White without certain handicaps."

The future is here – AlphaZero learns chess

    Approaching chess might still seem unusual. After all, although DeepMind had already shown near revolutionary breakthroughs thanks to Go, that had been a game that had yet to be ‘solved’. Chess already had its Deep Blue 20 years ago, and today even a good smartphone can beat the world number one. What is there to prove exactly?

Would the best mobile phone chess app be a favorite in a match against Carlsen?

    AlphaZero had done more than just master the game, it had attained new heights in ways considered inconceivable. The test is in the pudding of course, so before going into some of the fascinating nitty-gritty details, let’s cut to the chase. It played a match against the latest and greatest version of Stockfish, and won by an incredible score of 64 : 36, and not only that, AlphaZero had zero losses (28 wins and 72 draws)!

    Stockfish needs no introduction to ChessBase readers, but it's worth noting that the program was on a computer that was running nearly 900 times faster! Indeed, AlphaZero was calculating roughly 80 thousand positions per second, while Stockfish, running on a PC with 64 threads (likely a 32-core machine) was running at 70 million positions per second. In spite of this insane deficit, AlphaZero crushed Stockfish 64-36 with no losses at a time control of one minute per move.

    In the diagram above, we can see that in the early games, AlphaZero was quite enthusiastic about playing the French Defense, but after two hours (this so humiliating) began to play it less and less.

    People will put financial concerns over what they have access to.

I agree, and am in favor. People should make their own decisions about how to spend their money. Making everyone pay for a standard level of service (or else go without any service) does not seem like the way to please as many people as possible.

What's so bad about a world that includes the choice of cheap, basic internet service?

    I'll remind you of this conversation once a major ISP starts banning political speech it doesn't like.

I would like to express our disagreement in the form of a prediction, so we could give it some time and come back and see who was right. But I don't understand what your language of "banning political speech" even means.

I agree that comparatively few customers have access to 25 mbps broadband, as the map on your link shows.

If, instead, the map showed that most people can get 25 mbps, but few can get 100 mbps, would you continue to claim that Americans have insufficient access to broadband? You can keep raising the number until you reach a point where most people can't get that number.

Why is 25 mbps the target?

25 mbps is fast enough for five simultaneous high-definition Netflix streams. Is it some kind of market failure if a lot of people can't do that? How many people want to? How many are willing to pay for that level of service?

Netflix is the runaway leader of streaming services, serving some 50 million U.S. households. 50% of households subscribe to any video-on-demand service.

So half the country may not even be willing to pay for the 5 Mbps needed to stream HD, or 3 Mbps for SD video.

    Now it's your turn to provide a source.

I'll start with the "For more information" link from your source. As usual, to understand what's going on you have to dig into a long, tedious PDF.

From page 6:

    In the 2015 Broadband Progress Report, the Commission increased the speed benchmark for advanced telecommunications capability to 25 Mbps/3 Mbps, up from the 4 Mbps/1 Mbps benchmark used in the previous three Reports. In setting the benchmark, the Commission took into account the needs of multiple users in the average household, as well as the speeds required to use high-quality video, data, voice, and other broadband applications.

So the FCC defines mom, dad, and the three kids each watching their own HD movie all at once as "advanced" broadband, that's fine. I think it is a lot more bandwidth than many households are willing to pay for.

What's the outlook for a more basic broadband, say enough to watch just one or two movies at the same time, or perhaps just use Facebook and do some job-hunting?

The latest Internet Access Service Report was published in April. (The FCC budget, by the way, was perfectly constant from 2014 to 2017, with some extra funds for moving headquarters added in 2016-17. It has been pretty nearly constant since 2009.)

Page 6 has a chart like this one, from a 2016 article.

The numbers in this year's report have improved.

For 10 Mbps, enough to watch three simultaneous SD movies or two in high definition, 79% of developed census blocks have three or more ISPs to choose from. A census block is not the same as a household, but I haven't found more granular data. Another 18% have two providers, for a total of 97% of census blocks with choice in ISP.

For those who are willing to give up high-definition, and are content to watch one movie at a time, or simply choose not to subscribe to video on demand, 90% have three or more providers, 10% have two.

I think you may be overestimating the level of service many households are interested in paying for.

Today must be bubblenalysis day!

The Grumpy Economist opines on Bitcoin and Bubbles, with a jaw-dropping WSJ chart comparing 2017's 1000% rise in bitcoin to other "historically huge market moves." It seems unfair because all the other assets were decades old where the chart starts; bitcoin is not yet 10.

    The first equation of asset pricing is that price = expected present value of dividends. Bitcoin has no cash dividends, and never will. So right off the bat we have a problem...

    In sum, what's going on with Bitcoin seems to me like a perfectly "normal" phenomenon. Intersect a convenience yield and speculative demand with a temporarily limited supply, plus temporarily limited supply of substitutes, and you get a price surge.

Marginal Revolution asks Is Bitcoin just a bubble? with a hypothesis, similar but briefer than mine, starting with $241 trillion in wealth in the world.

    think of Bitcoin as competing for some of the asset space held by gold and also to some extent art. Gold, too, in its hedging functions is a “bubble,” though not a bubble. It is hard to ship, but has some extra value because it is perceived as a focal asset and one that does not covary positively in a simple way with the market portfolio. The same is true of Bitcoin, yet that kind of focality-based “bubbliness” can persist for centuries.

A comment mentions Extraordinary Popular Delusions and the Madness of Crowds, which may be a good answer to your request for research on bubbles. Gutenberg link

    Another story is told of an English traveller, which is scarcely less ludicrous. This gentleman, an amateur botanist, happened to see a tulip-root lying in the conservatory of a wealthy Dutchman. Being ignorant of its quality, he took out his penknife, and peeled off its coats, with the view of making experiments upon it. When it was by this means reduced to half its size, he cut it into two equal sections, making all the time many learned remarks on the singular appearances of the unknown bulb. Suddenly, the owner pounced upon him, and, with fury in his eyes, asked him if he knew what he had been doing? “Peeling a most extraordinary onion,” replied the philosopher. “Hundert tausend duyvel!” said the Dutchman; “it’s an Admiral Van der Eyck.” “Thank you,” replied the traveller, taking out his note-book to make a memorandum of the same; “are these admirals common in your country?” “Death and the devil!” said the Dutchman, seizing the astonished man of science by the collar; “come before the syndic, and you shall see.” In spite of his remonstrances, the traveller was led through the streets followed by a mob of persons. When brought into the presence of the magistrate, he learned, to his consternation, that the root upon which he had been experimentalising was worth four thousand florins; and, notwithstanding all he could urge in extenuation, he was lodged in prison until he found securities for the payment of this sum.

The claim is "You either have Comcast, or you don't have internet." This implies the presence of a person. Uninhabited areas don't need competing ISPs, so coverage gaps in a map are not necessarily a concern.

The great majority of California is "unserved" by wired internet according to the map, as if it were Antarctica; this is obviously misleading.

If you include wireless, even with a minimum of four providers most populated areas appear to be covered. If you don't include wireless, we need a standard for what "having internet" means.

About one-quarter of the world's adults, 1.3 billion in number, have over US$10,000 each and collectively $280 trillion in wealth, which Credit Suisse defines as "the value of financial assets plus real assets (principally housing) owned by households, minus their debts."

Much of this wealth is tied up in illiquid assets like housing, leaving about $90 trillion in easily-invested broad money. How much might end up in bitcoin?

Value investors like Warren Buffett scorn a strange new asset that is hard to understand. This probably reduces potential investor interest by three-quarters to 90%. People are naturally skittish when the word "bubble" is mentioned so often. Call it $10 trillion in liquid wealth that does not dismiss bitcoin on principle.

Probably most of that wealth is held by people who have at least heard of bitcoin. Say half of them are discouraged by barriers to entry, either government censure or immature infrastructure. Access is improving rapidly: Coinbase reportedly added a million new accounts in one month, and bitcoin ATMs are present in most wealthy countries, but we might imagine $5 trillion held by people who are curious and not immediately blocked.

Many of them will do some casual research and get quickly overwhelmed by the complexity of the technology and the intensity of the hype and snake oil. Then they will notice that the price went up by $1000 while they weren't paying attention, and promise themselves to get in when the price drops again.

I reckon (very casually) that leaves about a thousand billion dollars held by people who are interested, risk-tolerant, and willing to make the effort to buy. It's quite a lot compared to bitcoin's current "market cap" of about $160 billion, though I'm not sure if that number means much for an asset at an all-time-high. If the yet-unreached bitcoin demographic invests 5% of their play money, well, it's hard to imagine what it would do to the order book, with some previous owners selling to take profits, but there is plenty of room in the Up direction. $50 billion invested at today's price would buy about a third of the existing supply of 16.7 million units, making a doubling or tripling of price at least plausible.

The most important factors are hard to gauge. At this point speculation interest far outweighs any perceived utility of the technology (which could be provided as well or better by existing alternative cryptocurrencies). The bitcoin brand and the hype are the main drivers.

It's a Keynesian beauty contest, where investors decide to buy mainly by asking if there will be a greater fool to sell to later. So the "bubble" language is appropriate, but not very helpful in making predictions without any insight into when a crash might occur.

Jamie Dimon, the CEO of J.P. Morgan, called bitcoin a "fraud" and threatened to fire employees found speculating, but did not exactly predict imminent disaster.

    I’m not saying go short… Bitcoin can go $100,000 a bitcoin before it goes down, so this is not advice on what to do, ” the chief executive said Tuesday. “I refer to it like the tulip bulb crisis.”

I reread the chapter of Tulipomania to see if there were any warnings before the bust. If there were, they were hard to spot. In the winter of 1636-37 the tulip trade was "positively booming in Holland." A contemporary made a shopping list of what you could buy with 3,000 guilders, the price of a single flower:

    Eight fat pigs, four fat oxen, twelve fat sheep, twenty-four tons of wheat, forty-eight tons of rye, two hogsheads of wine, four barrels of eight-guilder beer, two tons of butter, a thousand pounds of cheese, a silver drinking cup, a pack of clothes, a bed with mattress and bedding, a ship.

Not only prize flowers, but previously worthless bulbs that had been sold by the pound were commanding "astonishing prices" by the beginning of 1637.

    The great crash in tulip prices began in Haarlem on the first Tuesday of that February, when a group of florists gathered to buy and sell as usual in one of the city's tavern colleges. As was customary, an established member of the college began the day's trading by testing the state of the market; he offered a pound of Witte Croonen or Switsers for sale. The florist asked a fair price — 1,250 guilders — for the bulbs, and in the normal course of events he would have found several eager buyers. Slates and chalk would have been distributed, the tulips would have been knocked down to the highest bidder, and the rest of the day's trading would have continued in its usual frenzied way. On this day, however, there were no bidders for the bulbs at 1,250 guilders. The auctioneer offered them again, this time cutting the price to 1,100 guilders. Still there was no interest. Desperately now, he offered his bulbs for a third time, dropping his price to a risible thousand guilders the pound. Once again there were no bids.

There's no telling when the bottom might fall out from bitcoin. If BTC is mainly a meme for "effortless money" it seems likely to grow until demand is sated, which I think could take more than another year.

Hopeless to guess what might happen, I got some help from a nearby elementary school student. I asked him to plot a continuation of the price chart for the next year.

This looks as plausible as anything else, so I am going to copy mk (never a bad strategy, I think) and put my non-competitive guess at $35K.

(I should add, I also requested a plot on a log chart, providing some justification for the Dimon Max.)

This source ignores where people live. For much of Wyoming, zero is the correct number of ISPs.

Can I get another, please?

A good question. I didn’t think so, since there are only recently relatively easy ways to buy Bitcoin. But Google suggested “bitcoin” as soon as I typed “how to short”...

This Is How You Can Short Bitcoin

It's your analogies that seem like weak tea to me. Most any business can offer any quality of service they choose, at whatever price they want, whether or not they participate in the lobbying game. Why don't we expect ISPs to compete on price and service like every other sector?

Perhaps you can correct me, but I'm not aware of any price controls for home internet service, where most people would use BitTorrent or Tor. I should say, where most people wouldn't use BitTorrent or Tor; do even five percent of home users know what those services are?

Your worst-case-scenario sounds like something a lot of customers would appreciate: being able to save money by forgoing services they don't want.

posts and shares 0/1