I was listening to NPR this morning in my car. The first time in years I’d heard a broadcast. Some might say it was concern trolling over nothing.
The story I caught was about how evil salad companies in California were throwing away salad in land fills. Apparently the packaged salad industry isn’t able to operate with zero production waste.
It was bad enough that salad was going uneaten, but the worst part was it wasn’t being composted. You see, when lettuce is thrown away in a landfill, it rots and gives off methane, which is bad. However, when it’s composted, it…doesn’t rot or something… Whatever though, the important thing is that composting is good, it’s like going to confession or saying your prayers was in 1650, if you don’t do it, you can’t be saved.
Big Salad causing big problems.
I’m unsure why America has been so out to get Bashar Al Assad all these years. We are told he’s a “brutal dictator”. So what? Is he worse than the Ottomans? Than the Umayyads or Abbasids? Than the Romans? Than Alexander the Great? Than the Achaemenid Persians or any of the other rough types who’ve ruled Syria? Do Americans or their government know a goddamm thing about the near east? Have they earned the right to an opinion?
Al Assad is a member of a watered down, laid back, minority sect of Islam. He is seen as a heretic by Sunni Muslims, the branch behind every major Islamic attack on America and Europe. He is allied with the remnant Orthodox Christian population in Syria as well as many of the minority branches of Islam that characterize Syria and make it so different from other Muslim areas. In a sane world, a Western power like the U.S. might begrudgingly back Assad, or at least leave him be. However, a group of German and Polish nationals decided to play the Iron Age Barbarian card and conquer Syria’s neighbor, the old Roman province Palaestina Prima, starting at the close of the First World War, and culminating the late 1940s. The German and Polish conquerors colonized these already long-populated lands with Europeans for their exclusionary, razor-wire-ringed, ethno-nationalist state. Hengest and Horsa would have approved. Syria, which lost a few colonial wars to the Poles and Germans, was now stuck in a subordinate stalemate.
America, apparently, must bend over backward to do whatever the Polish and German colonists want. This is because it’s important to have a “western state in the middle east”. You see, it’s bad to support western states in South Asia (Portuguese colonies conquered by India in the 1960s), or East Asia (Hong Kong, Macau, Malaya) but it is good to have Western representatives in West Asia where there are deep roots….unless we’re talking about Patriarch of Constantinople Bartholomew I, then we shouldn’t lift a finger to save a 1700 year old continuous tradition, at the very heart of Christianity and the West, even if it’s being strangled for no reason, by the West’s oldest living foe, over whom we have enormous influence. And so we must root out Al Assad…or something! “HE KILLS HIS OWN PEOPLE LOL OMG!!!1”
My sources tell me that the Syrian government forces are waning. ISIS marches to new victories. Weapons flow from the Turk, the Saudis and as far as I can tell, from the drooling dolt Uncle Sam (but only to the good ISIS, because there are always good guys, who love freedom). Russia and Iran send weapons to Assad, but weapons are not the problem, it’s men that the Syrian forces need.
If Assad should fall, and Syria should fall under ISIS’ neo Caliphate, America will have presided over a genocide. Every liberal and every nonSuni in Syria will be at the mercy of ruthless barbarians. Our tax dollars and geopolitical incompetence will have witnessed the transition from a flawed dictatorship (not new to Syria in the last 10k years) to a nightmare, a living hell for one of the oldest Christian communities. I emphasis the Christian Syrians because I think as Westerners, even if we don’t go to church or even believe in God, we share a special memetic conection to these Arabic speakers. Had the continents been ordered differently, it might be us who faced the Caliphate’s blade and they who looked on, without understanding from the distance of the cable news program. The heretical Muslims of Syria of course also face a grisly future worthy our sorrow and shame, but it is especially disgraceful that we leave our spiritual cousins to the dogs. The Chinese and Russian historians will not be kind to us.
If Syria falls to ISIS and all that we fear comes to pass, America will have participated in an irredemeable act of fratricide, we’ll be kin-slayers, and beyond forgiveness.
I just saw the film Ex Machina. I hadn’t heard of it before, but then that donation-funded, era-defining journalist in Los Angeles whom we all pretend not to read called it “high brow” so I thought I’d better check it out.
It’s a good movie. I think I enjoyed Mad Max: Fury Road more, but that’s only because Mad Max was an action movie for smart people, and I have a safe, boring life and high testosterone.
I didn’t actually want to write about the movie though. Instead I thought it was a good spur write a post I’ve had brewing in my brain for a long time now. It’s a common thing these days to see a headline like “Elon Musk shits pants at thought of AI: Will we all be killed by Google robots?”. Now, it could be that Elon is privy to higher-level technology than I, yet I remain skeptical. I’ve seen little improvement in my life in functional AI.
Take tonight. I moved to my current city about seven months ago and still don’t know it so well. I’d never been to the destination movie theater, it wasn’t far but I used Google Maps on my phone all the same. As you may know, phone app GPSes are sometimes a bit slow to tell you where to go. It said “turn left” but I was already turning right, wrongly anticipating the directions. Then it tells me to “turn right” at the next intersection. I obey, thinking it is just going to have me keep turning until I’m back on the original path. Instead, it sends me on this absurd route along the highway and then through a suburban neighborhood, taking much longer than the original route which it turned out was just within the city grid. On my way back, I realize that only about 5 or 6 turns are needed to get from my building to the theater. Google Maps failed me because it can’t think. GPS programs get a little better over time, but only because more and more contingencies get built into the code, it’s not actually thinking, it’s ‘just’ good programming, and the improvements face diminishing returns.
Another area where AI is demonstrably not progressing is in video games. I remember being so impressed with the AI in games like the original Half Life when I was a kid. Then I started building maps for the game and realized that it was all scripted. The characters in the game were only fiendish-seeming in their efforts to kill you because the developers had built in various ‘invisible’ trip wires and ‘action zones’ that would trigger different responses by the AI. When I play video games now, with an awareness of programing and how the environments are built, I can see things haven’t changed much. Similarly for strategy games, AI might be good, but if it acts at all like a human, it’s because it’s been given an excellent ‘cook book’ by the developers. It’s essentially no different in kind from the automata which have existed since the 10th century. The popularity of online multiplayer games, which are fun because other people make better teammates and foemen, bespeaks the failure of AI.
The same idea applies for translation engines. Anyone who has a solid grasp of at least a second language will know that you can’t trust Google Translate any more than Google Maps. It’s a useful tool, but when I really need to translate something, I type what I think the clause should be into Google/Bing in quotes, and see if I can find it used by a native speaker. Or I ask a native speaker. Google Translate can more or less handle the Germanic languages, particularly Dutch and the Scandis which are grammatically similar to English, but it still misses loads of subtext. Translating from more distant languages is harder, the Latin languages are sort of readable in English translation, but it chokes on the highly inflected and analytical yet folksy Slavic languages. Forget anything non Indo-European.
The reason I point to these mundane examples of AI making the same mistakes AI has always made is because I think consumer-level AI is pretty close to the cutting edge. Unlike Ex Machina‘s world, I don’t think it’s likely that massively advanced AI could be cooped up in Sergey Brin’s basement. Sure, it could be that some new paradigm for generating AI has been hidden and in the works for years and will come out next week. But given the poor progress we’ve seen in consumer industries to-date I’m doubtful.
I still think AI is inevitable, if only because eventually we’ll be able to scan the brain at such a high resolution that a computer simulation of the scanned activity will lead to contentiousness. But it might take 200 years a team of cloned John von Neumanns to do it. I think if the singularity happens in my life time, it will be because they’ve found effective aging treatments and I’m very, very old.
The efforts to set up an NGDP futures market on HyperMind are worthy. I have an account and trade on the 2015 contract now and then. Indeed, last weekend I spent over half my remaining HyperMind dollars to move the 2015 y/y % change market from 3.3% to 3.9%. 3.9% was the figure my market-informed forecasting engine was giving for 2015Q4, though the figure I’m presently showing on Efficient Forecast is the y/y % growth rate for 2016Q2 (a year from the current quarter 2015Q2).
As of now (01:00 CDT) the market is selling at 3.4%, which is still below what my system predicts for 2015Q4 (it’s below 3.9% now though, having fallen this week). I’ll have to wait for the Q2 NGDP Advance release before I get any more Hypermind cash, as I’d also bought shares on that Q/Q contract.
Hypermind’s big weakness is that the main contract is frozen at 2015Q4, the same as at the project’s inception. Efficient Forecast on the other hand, because it runs off other financial markets’ prices, and doesn’t need seed money, is always set to forecast for the year-ahead quarter. It occurred to me that I could start reporting the 2015Q4 figure on Efficient Forecast, to help the HyperMind traders out. However, I am going to keep it for myself, it would be disreputable of me not to.
If I really believe in my system (I do), I’d be a fool and a fraud to share it with you all. Now that there is a platform to make money off Efficient Forecast, it makes more sense to keep the methodology to myself, and direct you all to the public market. This is how you know I’m serious about it. Keeping it to myself honors the Efficient Markets Hypothesis (don’t you dare draw pictures of my Prophetic hypothesis!). Hoarding the forecast also gets to why we shouldn’t put much stock in the expert economic forecaster types we see in media. They don’t put their money where their mouths are.
We don’t yet have a true, year-ahead NGDP market. So I will keep Efficient Forecast up, but surely its days are numbered, at least as an NGDP forecasting site. Although the coming NGDP markets (the New Zealand deal, plus hopefully future Hypermind contracts) will mean less attention for me, it will also probably mean easy money for me, at least until the market is properly trained. I have quite possibly done more work forecasting NGDP than anyone else alive, which largely reflects the misdirected nature of economic forecasting, rather than my own abilities. Once I get a few forecasting cycles in to drown out randomness in the grossly inaccurate first NGDP print, I should be able to make some money trading Hypermind. I can also use the NGDP market forecast to run a suite of satellite forecasting equations, so there’s lots still to do with real time economic forecasting. Huge potential to shake up that industry.
I’m getting away from what I wanted to say though, so back on task. Hypermind is great, but the current contract is increasingly less interesting. A time will come, probably in early Q3, when the contract becomes completely uninteresting from a monetary policy standpoint. I like Sumner’s ship analogy. You want to steer the ship (NGDP) so that we’re always on course, always expected to be on the most efficient sea lane. However, this is a big-ass ship, like an aircraft carrier or a Korean oil tanker. The captain has very little control over where the ship will be in the next 5 minutes—that’s predetermined by the ship’s current position and inertia—though he has a great deal of control over where the ship will be in 3 hours or 3 months.
I am confident that Janet Yellen & co can keep the root-mean-squared-error of NGDP, about a target trend line, quite low over a 3 year time frame. I am not confident that Yellen can do much about the current quarter NGDP, or even really next quarter’s NGDP. The vagaries of weather and random disasters also mean that there can be ‘one off’ NGDP shortfalls for a given quarter, that tend to be made up in following quarters.
As we approach the due date of the Hypermind NGDP contract (2015Q4), and learn the NGDP levels for 2015Q2 and 2015Q3, the contract payoff will be increasingly due to nonmonetary factors, that is, things that don’t matter. In fact, I was disappointed when I first looked at the Hypermind contracts and saw that they’d spent money on quarter-over-quarter growth contracts for 2015 quarters. Who cares what Q2 NGDP growth is? That money could have been used to fund y/y contracts for 2016Q1 and 2016Q2. We care about the expected path a year to two years out.
The Fed should look ahead, and not just ahead, but about a year ahead. In my experience working in macroeconomic forecasting, a year is about as far ahead as I’m comfortable forecasting. Maybe as much as three years for some variables, but much further and the errors compound and it’s frankly counterproductive. Similarly, ‘current quarter’ or “nowcasting” approaches, which try to forecast very near term figures aren’t terribly interesting, near-term growth is mostly due to predetermined factors, like last quarter’s level and random events.
For these reasons I don’t put much stock in the first print of the BLS Employment Situation or the BEA’s GDP reports. By “don’t put much stock”, I mean I don’t even look at them. There is so much measurement error in early statistical estimates, and the economy is so noisy that it’s not usually worth the bother. I look at the markets. The truth, that no one want’s to hear, is that economic statistics are really only useful after they’ve been out for a while, when they’ve been revised and future data have been released, allowing the figures to be put into some context. I check economic statistics maybe six times a year. I can remember so many times, in 2011, 2012, and 2013 when it seemed big things were happening, but it was just measurement error and randomness; the U.S. macroeconomy has been remarkably stable and blessedly boring since 2010. No, about a year or 18 months ahead is the really interesting forecast.
This realization is what spurred me to move out of macroforecasting and into credit forecasting and stress testing. There’s just more action, higher resolution, little or no measurement error. The golden age of the macro econ blog is over. Level target NGDP, use markets, treat citizens like they club-own the country. Those are the lessons.
So to close this meandering post, I will keep trading Hypermind and hoping that that a well-heeled benefactor buys us a 2016 contract this summer.
P.S. it occurs to me that in the Efficient Forecast system, I define NGDP to be the mean of GDP and GDI. That might be why my system is reporting stronger growth. Still, NGDI is a better predictor of the revised NGDP levels which the Y/Y growth rate will be calculated off.
Efficient Forecast is back up.
It turned out that I hadn’t backed up the latest version of the .html file for the chart when I hastily took down the old server. This was for the best though, as it compelled me to rebuild the chart with a new API, it looks better than ever.
I still had my data gathering script running most days over the past two months, so I’ve been able to backfill the forecasts for that period. Go check it out.
For such a great man, Michael Bloomberg sure isn’t represented well by Bloomberg News. (Update, I meant by this, that BN doesn’t live up to the name, the standard.)
Actually I’m too hard on them. Bloomberg News is fairly ok, and they supply the only quasi-intraday 5-year TIPS yield on the net, which is useful if you’re running webscrapers to gather those data.
But today they had a misleading headline:
I’ll be damned if I’ll let that stand unopposed!
Go ahead and read the article. It’s short. Basically it seems that someone was writing about Canadian housing prices, probably that they’re under threat from falling oil prices. Then, I am guessing here, an editor had the author weave it into the unexpected rate cut story. There’s a special spot in hell for editors.
For the record—and I’m just writing this to keep my mind limber, I know most of you already understand it—you don’t cut rates to save a housing market. You cut rates, or rather move rates below the heretofore expected rate path, to boost nominal income. At leas that should be your reason, because that’s what you’d be doing, directly.
House prices are incidental. Hell, higher house prices are undesirable by themselves, as they hinder affordable family formation. The important thing is to keep the realized nominal income path as close as possible to the expected nominal income paths from the recent past, lest you cause a debt crisis. Today’s rate cut seems reasonable to me, and without looking into it further, my suspicion is that the Canadian economy will be fine, because the BoC seems to be implicitly following Lars Christensen’s Export Price Norm. Look at this graph from Yahoo! Finance:
The CAD is getting hammered, as well it should. I’m impressed the BoC is allowing their currency to weaken like this. The only alternative would be to run the country through a recession for the sake of ‘currency prestige’, which while really stupid, happens all the time. Sort of different shocks, but this reminds me of how Greenspan allowed the USD to similarly weaken when the U.S. was hit with the ’87 stock market crash
The Canadian dollar weakening is a great example of the benefits you get from having a small, local currency. Texas and North Dakota are almost certainly headed for local recession, because they’re in the ‘USD zone’. Canada, blessedly, is an independent nation, and will thus have the option of responding to local circumstances with a weaker currency. Being part of a currency empire is not a good thing.
The plot above suggests Switzerland is far from overheating. NGDP growth has steadily risen since the ceiling was put in, in 2012, but is still well below the 4% per year that’s probably about optimal.
All I can think is that the SNB has been forced to buy lots of foreign currency since this Russia trouble began and rumors of Greece leaving the EMU started circling. That is, much of the pressure on the CHF had been speculative (does the SNB have the will to print?) whereas now there’s genuine haven demand-folks really want to hold CHF and Swiss assets. Maybe the SNB is uncomfortable with its recent surge (that I’m assuming happened) in currency holdings.
If this is the case, it seems like a big mistake to me. I’m blogging on my phone and can’t update myself on Switzerland until late tonight, so take my analysis with some salt. I look forward to learning now the Brazilian and Danish market monitarists read this.
I hope it was clear that by allowing the franc to appreciate, the SNB has tightened monetary policy. The drop in Swiss share prices suggests this is going to be bad for NGDP. I wish I had had a webscraper set up to gather Swiss financial market prices.