There are a few wars in there, so this gets messy. In general, if you take a look at the non-reentrant spikes in unemployment, you'll see the outline of recessions. (They're all recessions, but not useful for pattern-matching.) That's great and all, but we need to identify a few things, like when is a new recession coming?
You'll notice that the less-chaotic of these (the ones that didn't have crashes or wars in the middle) spike up after what looks like a level trough (horizontal movement). It looks like something you might want to call "flat" there even though it's curved because there's an inflection point: the slope changes.
Get a straight edge.
On the downslope of, say, the dot-com bust, follow its regression. You'll see a point where it stops following your straight-edge, veering off to the right. Two years after that, the housing market shit itself, and the 2008 recession started.
You can see that in the mid-90s, and the mid-80s, and 1978. You can see the abortion around 1995, where something strange happened as a new recession looked like it might shoot out of nowhere, and then didn't.
So here's the trick: generally, you bottom out on unemployment when a recession ends. You have a recovery period where unemployment steadily goes away, and then it stops going away as fast (inflection point). Two years later, unemployment starts creeping back up. You need to identify the inflection point coming out of a recession, then identify the inflection point entering a new one.
Unemployment isn't a straight line. It'll wobble a bit. It may take 3-4 months to see if you inflected. So you can go up.1%, then continue on your way down, carrying the same general regression trend. If 3 months later you're going sideways or, worse, starting to climb slowly, that's the next recession. It's starting and, if something doesn't change, it's going to come full-swing.
We're coming up on about the right time. Unemployment just ticked up--that might be the flag, but I need to see if it goes back down 0.1%, then comes back up 0.1%, or if it otherwise shows that it's done falling and is ready to start climbing again. I had predicted September to be about the right time, and figured on a recession kicking in between September 2017 and April 2018. Note that you can't make this prediction until a few months after April 2015; and even then it's a shot in the dark. The marker should be just about here, and the recession itself should be 6-12 months further out.
Some people are actually pretty good at this shit. They're mostly using stock market reads and economic factors like GDP from various sectors. I'm using blunt tools. To the uninitiated, it all looks like voodoo, because there's always a lot of people screaming that the sky is about to fall in just a few more days and we all need to take cover; somehow the weatherman actually knows when a storm is coming, or at least he knows one's actually out there headed this way right now, even if he can't say precisely what time it's gonna get here. Yes, there's always a storm in the future; there isn't always a storm physically formed right over there, making its way over here, to arrive some time depending on how fast the wind blows.
... oh my god. You're trying to use ECON100 from high school!
Hey, what's bringing the price of LCD monitors down? A boom year for LCD monitor crops?
Putting aside the attempt to lean on the broken ideal that supply-and-demand is the primary factor in prices, no, I'm not confounding factors. I'm using the DEFINITION OF INFLATION. You're saying a penis is a long, narrow body part, like the five penises on your hand; and I'm saying a penis is exclusively the thing between your legs. One of us is wrong.
You say things like, "When inflation is low...", but you miss something: central banks keep inflation at a specified rate. In the US, we try to keep 2% inflation per year. Technical progress constantly causes deflation; we create an inflation benchmark--a measure of the change in prices of various groups of goods--and try to adjust that to be +2% per year on average.
So our CPI inflation is, ideally, 2%. Let's say we nail it one year, and the CPI considers $102 2018 dollars to be equivalent to $100 2017 dollars. Does that mean the central bank issued enough dollars to drive prices up by 2%?
No, not exactly.
Most likely, technical progress has brought the inflation benchmark goods down by 0.5% or so. That means the central bank has had to drive up those prices by an average 2.5%, creating 2% inflation.
That "2%" number is the reduction in purchasing power of a fixed sum of money (e.g. $100). The portion of that inflation caused by the increase in money supply is 2.5% (i.e. more than that 2%).
Inflation isn't the change in price caused by some source; it's the change in prices. Full stop. The definition of inflation doesn't care about the source, just like the definition of a meter doesn't change whether you're using a ruler or a tape measure to measure it.
You want to know something fun? We can actually remediate most of the problems in the welfare system and get a big tax cut on everybody with the top tax bracket ending up at 34.8%. Corporate income taxes (35%) become 32.5%; payroll taxes drop by a marginal 0.9%.
A two-adult household starts out with $17,502/year of untaxed income (as in, sans-job); a single-income household at $200,000 has about $2,700 more income after taxes than today, and every household with more or less income than that comes out further ahead. 75% of qualified HUD housing assistance applicants land on a waiting list and never receive benefits; instead, these households would have supplemental income, and HUD can top them up with a much-smaller subsidy besides. Unemployment insurance becomes unimportant; OASDI will never have solvency problems (real or imagined); taxes to support this will never need to increase (beware politics); and TANF, WIC, and SNAP can focus their mission on supporting children of low-income households.
Everyone's ideal is to go tax the rich as their first action. Nobody plans. This above plan? With 485,582 adult residents in Baltimore City at an average income of $25,707, there's an extra $3.16-$3.66 billion of spendable income coming to Baltimore. That elevates the average income to $32,220-$40,785, median $48,754-$57,319. This represents a sudden demand for employment equivalent to 190,000-220,000 full-time minimum-wage jobs (fewer relative to the average wage).
With Seattle's 595,882 at an average $45,673, they're looking at $4,500-$5,000 per person single, about $5,800 per person married. Range it as an extra $2.68-$3.46 billion. That money's not supposed to be taxed as an income tax; it gets spent, which is suddenly income, and what have we here... a revenue source?
It's not quite $140 million (not unless you capture 4% in taxes off this new money), but it's there.
There's a recession on the horizon and the whisperings of whatever bubble is popping are coming out. The narrative recently has been a start-up bubble; I haven't quantified it. All I know is we're consistently entering recession, at peak recession, or recovering from recession for years and years; and then we settle, coast for two years, and enter a new recession. It's about time for the next one,
so I guess I'm not the only one who noticed; folks want to be able to tell you they saw it coming.
Nonsense. That's not how inflation is defined or described. The Federal Treasury doesn't issue a 2% increase in money per person every year; they issue enough money to create a 2% increase in inflation. That's more than 2% more money.
Inflation is a general increase in prices and a fall in the purchasing power of money. Again: that differs from region to region, from product to product, and even from store to store. "General" is a handwavey term for the reasons I described.
Shrug. You consistently reject things that don't meet your ideals, such as the fact that raising subsets of wages (e.g. minimum wage) concentrates income into fewer hands and thus contributes a reduction in employment. So I look down on you as an ignorant child; you laugh at me because I have knowledge you don't and so your ignorance leads you to conclude wrongly, and you reject the possibility that you could be wrong because you're a child.
I got here by being you, realizing I was a dumbass, and then ceasing to be you. Catch up to me.
An object has discrete weight. The problem is you can't quote a number for "weight".
Apples have gone up 0.3%. Pears have gone up 0.1%. Steak has gone down -0.9%. Cereal is up 1.4%. Clothing is down -1.1%. Cars are up 1.3%. Computer parts are down -15% due to new manufacture technology.
You can't say, "Oh, there's X% inflation!" There isn't, because that's not a thing. When you look at the price of discrete things, they won't follow whatever inflation you're quoting. Likewise, you can't measure and compare the price of every sale of everything that occurs everywhere in a time frame to even get a baseline average.
You can measure the inflation of Snickers Ice Cream bars sold at the Rite-Aid on the corner of Lolly and Hutch; you can produce an approximated model of the inflation of food; you can create an inflation benchmark using any of numerous subset models, such as basic needs, middle-class expenditures, and so forth. You can't say "Inflation has changed by X%" any more than you can say "Weight has changed by X%".
More employment without printing up more money means less money available per person, and deflation.
Obviously, we're printing or loaning more money into existence. That's where jobs come from: consumers demand (buy) with money they have, and businesses can't keep up with the demand; they hire additional labor (jobs) to increase their capacity to keep up with demand. When capacity no longer increases linearly with labor, you have scarcity (you have to hire 3 people to produce twice the output of 1 person = costs going up when supplying more).
You can't really measure inflation. Inflation isn't real: it's a concept about currency losing discrete purchasing power over time. The prices of various goods aren't directly related, however, because the costs aren't directly related: you can find a way to farm cheaper wheat (GMO wheat injects a barley gene to cut the growing time by 20% and increase the yield by 50%--the wheat costs 40% as much!) but not cheaper potatoes. You might find a way to make fertilizer cheaper, impacting wheat and potatoes; and then wheat costs 37% as much and potatoes cost 93% as much.
In other words: prices don't just go up (or down). The price of discrete goods changes; this may impact supply chains and affect other prices, which are themselves affected by other factors as well and, in total, are not necessarily bound to change in price strictly-proportional to any given supply input in any given time frame. Other shit is going on.
We measure an inflation benchmark by selecting subsets of goods and modeling inflation that way. From that benchmark, and from population growth, we issue more money so as to create a 2% annual inflation benchmark. If our benchmark shows that goods have become roughly 0.3% cheaper to produce, we issue enough additional money to cover population change, and then increase the total money supply by 2.3% to target a 2% increase in the price of goods.
U6 is a bastard number. It measures underemployed, and it peaked at 16%-ish.
I don't like U6 because it's not a good measure of anything. Underemployment means you have 10 people and 5 jobs, so 10 people work half a job (40 hours). They're only underemployed if they desire more working time and work less than full-time. Thing is 10 people working 20 hours each is 10 underemployed; 10 people working 10 hours each is 10 underemployed.
For underemployment, I want new metrics.
The first metric is to measure only the U6 underemployed--they want more hours, but can't get them. Count their hours. Every 40 hours is one job. Give us the number of full-time jobs available as 40 labor-hours per week (2,080 per year) and the number of underemployed. That tells us how many people are fighting over how many jobs.
The second is that, plus people who are content and working less than full time. That gives you an accurate count of all available working hours, excluding any overtime worked. It lets you see how many people are working and content (UN2 persons - UN1 persons) and how many full jobs are available among them.
The third is a full count of all hours including overtime hours. That lets you count the number of full-time jobs against the number of employed plus any UE metric you want. (UN3 jobs)
Now you know precisely how much work is available, how it's distributed, and how our employment market really looks.
Disability means you can't work. It should, technically, be U5: you're willing but not able to work. U5 includes single mothers who can't afford daycare and so aren't working because they can't afford to work, for example.
Actually, the peak labor force participation rate was 66.4% in 2007. Adjusting for peak labor force participation rate, our current U3 unemployment rate of 4.7 at 62.8% would be 4.97%. This is versus a peak unemployment rate of 10.0% at 64.4% in October 2009, which would be 10.31% at 66.4% labor participation rate; our current U3 would adjust to 4.82% at 64.4% labor force participation rate.
So if you want to use the labor force participation rate numbers as a normalization, unemployment fell from 10% in October 2009 to 4.82% today; or it fell from 10.31% in October 2009 to 4.97% today.
The old labor force argument was that LFPR went up because the economy was so bad that two people had to work to make enough income to survive. That was also a bogus argument. We're in a really long labor force bubble, and we're so wealthy now that more spouses and domestic partners are electing to not work and have a single breadwinner in the family again. This comes after a prior cultural shift around WW2 where men went to war and women got jobs, and the women kept on working.
In any case, you tried to handwave labor force participation rate without running the numbers. Running the numbers proves you're wrong.
I told you, no data, no conjecture. If I wanted to make shit up, I would say there must be outliers on one end, and that extra advertisement and season finales must draw more eyes than usual, and try to spin a narrative on wholly-imagined numbers.
Of course, if I tried that, I'd have to point out that the deviations are based on the same audience affinities, and would only represent normal variations in existing viewership; whereas deviations from other events draw normal viewership away. That suggests that imagining there may be an inverse-superbowl might be stupid. Note that a deviation within 1.5 times the interquartile distance from the first and third quartiles is not a statistical outlier.
Again: I don't have data to back up the claim that there is or is not a positive outlier effect, or how it compares to negative outliers in frequency and magnitude.
It's a continuing pattern. Our economic system isn't yet stable; I'm trying to improve on that, but it's difficult politically.
In high-school economics, they talk about the business cycle, with peaks and troughs. It'd be nice if the world was like that--I think I can achieve that, actually. The problem is we actually get most of our "good times" battered and bleeding, in a bad economy, recovering from a recession; there's a very short low point where the economy's stable, and then it sharply drives itself into recession again. They're usually small; the last one of the magnitude of the Great Recession of 2008 peaked in the last quarter of 1982.
An economy with better stability would recover from recession faster, and would react more-quickly. That means the ramp-up would be slower (constant stability = constant recovery, including recovery during the fall), while the reaction of the economy would be faster (we'd move from setting in to recovery sooner), and the long-term recovery would also be faster (we'd return to baseline at a faster rate). As a result, recessions would reach lower peaks in severity, recover more-rapidly, and normalize sooner. The economy would sit in a better state and wobble, instead of having repeated seizures.
The downside is the real world would start to look like ECON 100. People already think they understand economics because they had a class about it in high school.
Yeah but the press should have a thermal protection circuit that cuts the motor if it overheats, and shouldn't be designed to push so hard that its own parts can't take the strain.
You won't have a strong political bias on many issues. Most people aren't rabid Amazon fanboys or Amazon haters; they use Amazon because it's convenient, until someone tells them Amazon is fucking them and that they should be outraged. Most people donate to the American Red Cross because they've been hit up for money, and because they hear the name a lot--until someone throws them a line about how ARC is a corrupt money-wasting organization.
You might have a political ideal of whether charity or big business is good or bad, but not much of an opinion on a particular business. You're not all that vested in it. We can turn you from a passive actor without a strong bias into a politically-polarized reader with a manufactured opinion pretty damned fast if you're not actively defending yourself against stupidity.
I'm schizoid. My dad's visibly schizotypal; I've watched him rewrite his memories. When something changes that doesn't damage his political beliefs but does make the past messy, he retcons it into history. Republicans supporting Social Security? Hey, remember when Bush was pushing back against the liberal assault on Social Security because they wanted to privatize it so they could get kickbacks from Goldman Sachs and steal all your money?... yeah, me neither; Bush was the one trying to privatize Social Security ("so the Liberals can't steal the money anymore"). Now that that's flipped, though, he's flipped history around. We've always been at war with Eastasia.
Distorted thinking is dangerous. I don't want to go along for the ride and just swallow whatever any idiot writes, or cling to my own political bias without re-evaluating it in the face of new facts, and end up screaming from the rooftops about FEMA's coffins and the plague they've developed to exterminate 95% of Americans and march the rest into forced labor camps. I like my psychosis, and I like it compartmentalized away from the real world.
I say a lot of things that are uncomfortable, and people sometimes attack me for it. That's okay, as long as I'm otherwise largely-distinct from Hans Reiser's delusional bullshit. On the plus side, people sometimes stop and think when I talk, because I say a lot of things that are uncomfortable for a lot of reasons, some of them too disturbing to simply bury away in the back of your mind and ignore.
Season finales and well-advertised episodes would stand out, yeah. I'm not sure by how much. It might not be enough to make the outlier criteria. This is, again, why robust data analysis is better than a single number, even though we have statistical methods for improving that single number's meaningfulness.
Anyway, next time some episodes airs during the superbowl, advertisers can trust the predicted number of viewers being correct
The problem is you have six numbers: 890, 870, 845, 865, 520. The average here is 798. That last number is during the Super Bowl.
So the number you're looking for "during the superbowl" is 520. The number you're looking for as a benchmark for the series is more like 867.5. If you don't have the 520 number but do have something like an understanding of the viewership of the superbowl and its demographics impact, you can use the 867.5 number to predict the 520 number, with some loss of accuracy. If you have the 798 number, you can try to do this in reverse to predict the 867.5 number, with some loss of accuracy.
The number you want accurate is the 867.5 number.
It gets better: what happens when that 520 number is "9/11 happened today and everyone turned off Smallville and turned on CNN"? Now you have a meaningless sample polluting your sample base, and you have to figure on if that sample is there (by analyzing history) and how to factor it out (for which you have minimal data--it happened one time).
Even the set of all individual samples isn't really more-accurate than a properly-analyzed conclusion. Your pile of samples requires methodology for analysis. That requires work. There's limited effort available, and repeating that work thousands of times is infeasible; either you pay Arbitron to do it for you or you build your own mini-Arbitron to do it in-house at a much-higher cost and with less-reliable results. You can take a third option and do qualitative analysis: cherrypick what your gut tells you is good, then analyze those against each other; ignore the rest because it's too expensive to get good data.
Tough problem, huh? Read what I said up top. We need to be able to ask more questions about broad spans of data if we want useful information.
Actually, unemployment numbers count discouraged employees in U4. U4 is currently 4.7. This year, starting in January, it has been 5.2, 5.0, 4.8, 4.7, 4.5, and 4.7. U3 has been 4.7, 4.5, 4.4, 4.3, 4.4.
How would you get positive outliers? Are there positive outliers?
We can conjecture about things all day; the facts here, however, are that data confounded by outside variables to the point of becoming a negative outlier is removed from the system. They may not be statisticians over there, and their reasoning might be incorrect; the action taken is the correct action. I don't have any facts about any other actions taken. Bringen sie mier data.
The count of eyeballs is not correct because it suggests that the count is different for the other airings of the show.
You do realize that not removing this distorted data in any professional capacity will get bad things on your doorstep, right? For example: if you have a deviation from the median greater than 1.5x the inter-quartile distance and you keep it in your data to show safety and efficacy of a drug, you are committing fraud and the FDA will fine you and pull your drug from the market when it turns out to be more-dangerous and less-effective to the general population than you said it was.
I took statistics. My teacher got pissed off because I was failing. I never did my homework, napped in class, and misanswered four problems on my exams all year--and every one of those was due to misreading the problem. I got a perfect score on the AP Statistics Exam. When I tell you that data impacted by outside variables pollutes the model and gives you invalid results, that's an expert opinion.
It's amazing how fucking horrendous people on Reddit and Slashdot are at statistics. A few months ago, I had to deal with a guy claiming some research was invalid because the researchers used people of different backgrounds and so, because the participants were not all substantially-similar, the statistical trend could have been caused by some other variable that differs between them. That's how bad you people are at this shit.
I can use real numbers to commit fraud by not removing data that needs to be removed. They removed data that needed to be removed for their statistical model ("Ratings") to be valid. If doing it the "right" way gets you an incorrect result, then that's not the right way to do it.
tl;dr the right way to handle this would be for the Nelson system to have automatically removed those numbers by identifying them as statistical outliers. It's generally-trivial math. That leaves a smaller problem that things which can hold their popularity even against elections and shit still incorporate confounded data and so come out less-wrong instead of corrected.
Because it's novel, thus unknown, and you can get people to see you as a source of security by making them feel threatened by something else. They see that you saved them with knowledge, and come back to you.
Unethical journalism to construct stories of villainy from big organizations people trust is a staple of ProPublica. They've repeatedly released American Red Cross Lessons Learned documentation--stuff that says what went wrong, what went right, what they'd do different if they did it again, why it went wrong, and what needs more review--with headlines and fanciful stories claiming that ARC is ineffective and is trying to cover up and ignore the horrendous mismanagement of their relief efforts.
In case you missed it: the existence of LL documentation demonstrates the exact opposite.
ProPublica also attacked Amazon's search because it shows a lower price+shipping for an Amazon product based on the free shipping Amazon offers (on orders over $25), which they demonstrated by tacking on the cost of expedited shipping to Amazon-sourced products. So you can get it for $23 and free shipping from Amazon, or $21 + $3.99 shipping from BuyOurBooks; ProPublica says it's more-expensive from Amazon because non-Super-Saver-Shipping orders have to pay $23+$6.21, and thus concludes Amazon should have ranked the other option as the top, cheapest option, and is only putting their products first to scam you into sending them your money.
Who doesn't have an entire backlog of stuff they want to buy at some point? I just tack something else I didn't need right away onto my order to pad it out.
When you sit down and weed these things out, you start noticing that a deep inspection makes it look like a whole lot of yelling about nothing. The presented facts are organized to tell an amazing falsehood, and suggestions are made to do things "better" which would be objectively-worse. The point is most people don't sit down and debunk the story; they apply their own political bias and accept or reject it. If they don't yet have a strong bias, then they let you spoon feed your opinion to them.
Now you have an audience who swallowed your bullshit about how someone they trusted has betrayed them. You are their savior. They come back to beg you for more information, to tell them more of the wisdom they have missed in their vulgar, miserable, wretched lives. They give you ad revenue.
Of course we push hundreds of attack pieces against anything new and trendy. It's new and trendy, and most people are either against change or indifferent; we can convert those people to returning revenue!
If you put a 7-cent part in a machine where a 3-cent part typically falls within the lifespan of the machine, you waste money (and labor). Likewise, you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts, and so costs 4-5 times as much.
what you say here, thus, is only valid if one would replace a few 3-cent parts with 7-cent parts, AND LEAVE OTHER KEY-COMPONENTS (aka, which are critical to the working of the machine) with 3-cent parts.
These are the same statement. I made the first; you responded with the second.
Just for redundancy's sake:
you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts, and so costs 4-5 times as much.
*lifespan* of you machine goes up when you put high-quality parts in it. Ergo, when you replace the 3-cent part components with 7-cent part components, then what 'falls within the lifespan of the machine' is exactly that, which the higher components have elevated it to. It's not a fixed lifespan, based on 3-cent parts.
Wow, you're a genius. I told you that a more-durable machine uses more-expensive (and more-durable) parts, and from that you reasoned that a machine will be more-durable if built out of more-durable parts!
You originally said: "If you put a 7-cent part in a machine where a 3-cent part typically falls within the lifespan of the machine, you waste money (and labor)." This is obviously only true if there are more than one key component and you *only* changed one of those components to a higher quality one, and all the rest are 3-cent parts
So, if
the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts
then the use of more-durable, more-expensive parts throughout the machine results in an increased lifespan? You know, what I fucking said right after the sentence you cherrypicked.
Here, let me un-fuck your stupid blathering.
And if you're arguing *you* already said that: you did! You originally said: "you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts." This is obviously true, as the more-durable machine is more-durable because it uses more-durable critical components. There's no other way for a machine to become more-durable, after all.
Now there's an accurate statement: I already said all that shit before you started talking.
you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts
Thus, I repeat: if you change the 3-cent parts (note the plural) with 7-cent parts, your machine WILL get more durable.
Good job repeating again what I said way the fuck up there before you got involved, like you thought of it first. I'm not going to let you try to rip off my post and pretend you invented the thing.
Slashdot Mirror
I haven't been predicting the next recession on the horizon since 2009; the Zero Hedge guy is a raving lunatic.
Take a look at month-to-month unemployment, go back a while, like 1970.
There are a few wars in there, so this gets messy. In general, if you take a look at the non-reentrant spikes in unemployment, you'll see the outline of recessions. (They're all recessions, but not useful for pattern-matching.) That's great and all, but we need to identify a few things, like when is a new recession coming?
You'll notice that the less-chaotic of these (the ones that didn't have crashes or wars in the middle) spike up after what looks like a level trough (horizontal movement). It looks like something you might want to call "flat" there even though it's curved because there's an inflection point: the slope changes.
Get a straight edge.
On the downslope of, say, the dot-com bust, follow its regression. You'll see a point where it stops following your straight-edge, veering off to the right. Two years after that, the housing market shit itself, and the 2008 recession started.
You can see that in the mid-90s, and the mid-80s, and 1978. You can see the abortion around 1995, where something strange happened as a new recession looked like it might shoot out of nowhere, and then didn't.
So here's the trick: generally, you bottom out on unemployment when a recession ends. You have a recovery period where unemployment steadily goes away, and then it stops going away as fast (inflection point). Two years later, unemployment starts creeping back up. You need to identify the inflection point coming out of a recession, then identify the inflection point entering a new one.
Unemployment isn't a straight line. It'll wobble a bit. It may take 3-4 months to see if you inflected. So you can go up .1%, then continue on your way down, carrying the same general regression trend. If 3 months later you're going sideways or, worse, starting to climb slowly, that's the next recession. It's starting and, if something doesn't change, it's going to come full-swing.
We're coming up on about the right time. Unemployment just ticked up--that might be the flag, but I need to see if it goes back down 0.1%, then comes back up 0.1%, or if it otherwise shows that it's done falling and is ready to start climbing again. I had predicted September to be about the right time, and figured on a recession kicking in between September 2017 and April 2018. Note that you can't make this prediction until a few months after April 2015; and even then it's a shot in the dark. The marker should be just about here, and the recession itself should be 6-12 months further out.
Some people are actually pretty good at this shit. They're mostly using stock market reads and economic factors like GDP from various sectors. I'm using blunt tools. To the uninitiated, it all looks like voodoo, because there's always a lot of people screaming that the sky is about to fall in just a few more days and we all need to take cover; somehow the weatherman actually knows when a storm is coming, or at least he knows one's actually out there headed this way right now, even if he can't say precisely what time it's gonna get here. Yes, there's always a storm in the future; there isn't always a storm physically formed right over there, making its way over here, to arrive some time depending on how fast the wind blows.
Hey, what's bringing the price of LCD monitors down? A boom year for LCD monitor crops?
Putting aside the attempt to lean on the broken ideal that supply-and-demand is the primary factor in prices, no, I'm not confounding factors. I'm using the DEFINITION OF INFLATION. You're saying a penis is a long, narrow body part, like the five penises on your hand; and I'm saying a penis is exclusively the thing between your legs. One of us is wrong.
Mabye you can read up on inflation.
You say things like, "When inflation is low...", but you miss something: central banks keep inflation at a specified rate. In the US, we try to keep 2% inflation per year. Technical progress constantly causes deflation; we create an inflation benchmark--a measure of the change in prices of various groups of goods--and try to adjust that to be +2% per year on average.
So our CPI inflation is, ideally, 2%. Let's say we nail it one year, and the CPI considers $102 2018 dollars to be equivalent to $100 2017 dollars. Does that mean the central bank issued enough dollars to drive prices up by 2%?
No, not exactly.
Most likely, technical progress has brought the inflation benchmark goods down by 0.5% or so. That means the central bank has had to drive up those prices by an average 2.5%, creating 2% inflation.
That "2%" number is the reduction in purchasing power of a fixed sum of money (e.g. $100). The portion of that inflation caused by the increase in money supply is 2.5% (i.e. more than that 2%).
Inflation isn't the change in price caused by some source; it's the change in prices. Full stop. The definition of inflation doesn't care about the source, just like the definition of a meter doesn't change whether you're using a ruler or a tape measure to measure it.
You want to know something fun? We can actually remediate most of the problems in the welfare system and get a big tax cut on everybody with the top tax bracket ending up at 34.8%. Corporate income taxes (35%) become 32.5%; payroll taxes drop by a marginal 0.9%.
A two-adult household starts out with $17,502/year of untaxed income (as in, sans-job); a single-income household at $200,000 has about $2,700 more income after taxes than today, and every household with more or less income than that comes out further ahead. 75% of qualified HUD housing assistance applicants land on a waiting list and never receive benefits; instead, these households would have supplemental income, and HUD can top them up with a much-smaller subsidy besides. Unemployment insurance becomes unimportant; OASDI will never have solvency problems (real or imagined); taxes to support this will never need to increase (beware politics); and TANF, WIC, and SNAP can focus their mission on supporting children of low-income households.
Everyone's ideal is to go tax the rich as their first action. Nobody plans. This above plan? With 485,582 adult residents in Baltimore City at an average income of $25,707, there's an extra $3.16-$3.66 billion of spendable income coming to Baltimore. That elevates the average income to $32,220-$40,785, median $48,754-$57,319. This represents a sudden demand for employment equivalent to 190,000-220,000 full-time minimum-wage jobs (fewer relative to the average wage).
With Seattle's 595,882 at an average $45,673, they're looking at $4,500-$5,000 per person single, about $5,800 per person married. Range it as an extra $2.68-$3.46 billion. That money's not supposed to be taxed as an income tax; it gets spent, which is suddenly income, and what have we here... a revenue source?
It's not quite $140 million (not unless you capture 4% in taxes off this new money), but it's there.
3D, VR, and AI.
There's a recession on the horizon and the whisperings of whatever bubble is popping are coming out. The narrative recently has been a start-up bubble; I haven't quantified it. All I know is we're consistently entering recession, at peak recession, or recovering from recession for years and years; and then we settle, coast for two years, and enter a new recession. It's about time for the next one, so I guess I'm not the only one who noticed; folks want to be able to tell you they saw it coming.
Nonsense. That's not how inflation is defined or described. The Federal Treasury doesn't issue a 2% increase in money per person every year; they issue enough money to create a 2% increase in inflation. That's more than 2% more money.
Inflation is a general increase in prices and a fall in the purchasing power of money. Again: that differs from region to region, from product to product, and even from store to store. "General" is a handwavey term for the reasons I described.
Shrug. You consistently reject things that don't meet your ideals, such as the fact that raising subsets of wages (e.g. minimum wage) concentrates income into fewer hands and thus contributes a reduction in employment. So I look down on you as an ignorant child; you laugh at me because I have knowledge you don't and so your ignorance leads you to conclude wrongly, and you reject the possibility that you could be wrong because you're a child.
I got here by being you, realizing I was a dumbass, and then ceasing to be you. Catch up to me.
An object has discrete weight. The problem is you can't quote a number for "weight".
Apples have gone up 0.3%. Pears have gone up 0.1%. Steak has gone down -0.9%. Cereal is up 1.4%. Clothing is down -1.1%. Cars are up 1.3%. Computer parts are down -15% due to new manufacture technology.
You can't say, "Oh, there's X% inflation!" There isn't, because that's not a thing. When you look at the price of discrete things, they won't follow whatever inflation you're quoting. Likewise, you can't measure and compare the price of every sale of everything that occurs everywhere in a time frame to even get a baseline average.
You can measure the inflation of Snickers Ice Cream bars sold at the Rite-Aid on the corner of Lolly and Hutch; you can produce an approximated model of the inflation of food; you can create an inflation benchmark using any of numerous subset models, such as basic needs, middle-class expenditures, and so forth. You can't say "Inflation has changed by X%" any more than you can say "Weight has changed by X%".
More employment without printing up more money means less money available per person, and deflation.
Obviously, we're printing or loaning more money into existence. That's where jobs come from: consumers demand (buy) with money they have, and businesses can't keep up with the demand; they hire additional labor (jobs) to increase their capacity to keep up with demand. When capacity no longer increases linearly with labor, you have scarcity (you have to hire 3 people to produce twice the output of 1 person = costs going up when supplying more).
You can't really measure inflation. Inflation isn't real: it's a concept about currency losing discrete purchasing power over time. The prices of various goods aren't directly related, however, because the costs aren't directly related: you can find a way to farm cheaper wheat (GMO wheat injects a barley gene to cut the growing time by 20% and increase the yield by 50%--the wheat costs 40% as much!) but not cheaper potatoes. You might find a way to make fertilizer cheaper, impacting wheat and potatoes; and then wheat costs 37% as much and potatoes cost 93% as much.
In other words: prices don't just go up (or down). The price of discrete goods changes; this may impact supply chains and affect other prices, which are themselves affected by other factors as well and, in total, are not necessarily bound to change in price strictly-proportional to any given supply input in any given time frame. Other shit is going on.
We measure an inflation benchmark by selecting subsets of goods and modeling inflation that way. From that benchmark, and from population growth, we issue more money so as to create a 2% annual inflation benchmark. If our benchmark shows that goods have become roughly 0.3% cheaper to produce, we issue enough additional money to cover population change, and then increase the total money supply by 2.3% to target a 2% increase in the price of goods.
Fiat money protects us from a lot of nasty shit.
U6 is a bastard number. It measures underemployed, and it peaked at 16%-ish.
I don't like U6 because it's not a good measure of anything. Underemployment means you have 10 people and 5 jobs, so 10 people work half a job (40 hours). They're only underemployed if they desire more working time and work less than full-time. Thing is 10 people working 20 hours each is 10 underemployed; 10 people working 10 hours each is 10 underemployed.
For underemployment, I want new metrics.
The first metric is to measure only the U6 underemployed--they want more hours, but can't get them. Count their hours. Every 40 hours is one job. Give us the number of full-time jobs available as 40 labor-hours per week (2,080 per year) and the number of underemployed. That tells us how many people are fighting over how many jobs.
The second is that, plus people who are content and working less than full time. That gives you an accurate count of all available working hours, excluding any overtime worked. It lets you see how many people are working and content (UN2 persons - UN1 persons) and how many full jobs are available among them.
The third is a full count of all hours including overtime hours. That lets you count the number of full-time jobs against the number of employed plus any UE metric you want. (UN3 jobs)
Now you know precisely how much work is available, how it's distributed, and how our employment market really looks.
Disability means you can't work. It should, technically, be U5: you're willing but not able to work. U5 includes single mothers who can't afford daycare and so aren't working because they can't afford to work, for example.
They won't show up in U4.
Actually, the peak labor force participation rate was 66.4% in 2007. Adjusting for peak labor force participation rate, our current U3 unemployment rate of 4.7 at 62.8% would be 4.97%. This is versus a peak unemployment rate of 10.0% at 64.4% in October 2009, which would be 10.31% at 66.4% labor participation rate; our current U3 would adjust to 4.82% at 64.4% labor force participation rate.
So if you want to use the labor force participation rate numbers as a normalization, unemployment fell from 10% in October 2009 to 4.82% today; or it fell from 10.31% in October 2009 to 4.97% today.
The old labor force argument was that LFPR went up because the economy was so bad that two people had to work to make enough income to survive. That was also a bogus argument. We're in a really long labor force bubble, and we're so wealthy now that more spouses and domestic partners are electing to not work and have a single breadwinner in the family again. This comes after a prior cultural shift around WW2 where men went to war and women got jobs, and the women kept on working.
In any case, you tried to handwave labor force participation rate without running the numbers. Running the numbers proves you're wrong.
I told you, no data, no conjecture. If I wanted to make shit up, I would say there must be outliers on one end, and that extra advertisement and season finales must draw more eyes than usual, and try to spin a narrative on wholly-imagined numbers.
Of course, if I tried that, I'd have to point out that the deviations are based on the same audience affinities, and would only represent normal variations in existing viewership; whereas deviations from other events draw normal viewership away. That suggests that imagining there may be an inverse-superbowl might be stupid. Note that a deviation within 1.5 times the interquartile distance from the first and third quartiles is not a statistical outlier.
Again: I don't have data to back up the claim that there is or is not a positive outlier effect, or how it compares to negative outliers in frequency and magnitude.
It's a continuing pattern. Our economic system isn't yet stable; I'm trying to improve on that, but it's difficult politically.
In high-school economics, they talk about the business cycle, with peaks and troughs. It'd be nice if the world was like that--I think I can achieve that, actually. The problem is we actually get most of our "good times" battered and bleeding, in a bad economy, recovering from a recession; there's a very short low point where the economy's stable, and then it sharply drives itself into recession again. They're usually small; the last one of the magnitude of the Great Recession of 2008 peaked in the last quarter of 1982.
An economy with better stability would recover from recession faster, and would react more-quickly. That means the ramp-up would be slower (constant stability = constant recovery, including recovery during the fall), while the reaction of the economy would be faster (we'd move from setting in to recovery sooner), and the long-term recovery would also be faster (we'd return to baseline at a faster rate). As a result, recessions would reach lower peaks in severity, recover more-rapidly, and normalize sooner. The economy would sit in a better state and wobble, instead of having repeated seizures.
The downside is the real world would start to look like ECON 100. People already think they understand economics because they had a class about it in high school.
Yeah but the press should have a thermal protection circuit that cuts the motor if it overheats, and shouldn't be designed to push so hard that its own parts can't take the strain.
You won't have a strong political bias on many issues. Most people aren't rabid Amazon fanboys or Amazon haters; they use Amazon because it's convenient, until someone tells them Amazon is fucking them and that they should be outraged. Most people donate to the American Red Cross because they've been hit up for money, and because they hear the name a lot--until someone throws them a line about how ARC is a corrupt money-wasting organization.
You might have a political ideal of whether charity or big business is good or bad, but not much of an opinion on a particular business. You're not all that vested in it. We can turn you from a passive actor without a strong bias into a politically-polarized reader with a manufactured opinion pretty damned fast if you're not actively defending yourself against stupidity.
I'm schizoid. My dad's visibly schizotypal; I've watched him rewrite his memories. When something changes that doesn't damage his political beliefs but does make the past messy, he retcons it into history. Republicans supporting Social Security? Hey, remember when Bush was pushing back against the liberal assault on Social Security because they wanted to privatize it so they could get kickbacks from Goldman Sachs and steal all your money? ... yeah, me neither; Bush was the one trying to privatize Social Security ("so the Liberals can't steal the money anymore"). Now that that's flipped, though, he's flipped history around. We've always been at war with Eastasia.
Distorted thinking is dangerous. I don't want to go along for the ride and just swallow whatever any idiot writes, or cling to my own political bias without re-evaluating it in the face of new facts, and end up screaming from the rooftops about FEMA's coffins and the plague they've developed to exterminate 95% of Americans and march the rest into forced labor camps. I like my psychosis, and I like it compartmentalized away from the real world.
I say a lot of things that are uncomfortable, and people sometimes attack me for it. That's okay, as long as I'm otherwise largely-distinct from Hans Reiser's delusional bullshit. On the plus side, people sometimes stop and think when I talk, because I say a lot of things that are uncomfortable for a lot of reasons, some of them too disturbing to simply bury away in the back of your mind and ignore.
Season finales and well-advertised episodes would stand out, yeah. I'm not sure by how much. It might not be enough to make the outlier criteria. This is, again, why robust data analysis is better than a single number, even though we have statistical methods for improving that single number's meaningfulness.
Anyway, next time some episodes airs during the superbowl, advertisers can trust the predicted number of viewers being correct
The problem is you have six numbers: 890, 870, 845, 865, 520. The average here is 798. That last number is during the Super Bowl.
So the number you're looking for "during the superbowl" is 520. The number you're looking for as a benchmark for the series is more like 867.5. If you don't have the 520 number but do have something like an understanding of the viewership of the superbowl and its demographics impact, you can use the 867.5 number to predict the 520 number, with some loss of accuracy. If you have the 798 number, you can try to do this in reverse to predict the 867.5 number, with some loss of accuracy.
The number you want accurate is the 867.5 number.
It gets better: what happens when that 520 number is "9/11 happened today and everyone turned off Smallville and turned on CNN"? Now you have a meaningless sample polluting your sample base, and you have to figure on if that sample is there (by analyzing history) and how to factor it out (for which you have minimal data--it happened one time).
Even the set of all individual samples isn't really more-accurate than a properly-analyzed conclusion. Your pile of samples requires methodology for analysis. That requires work. There's limited effort available, and repeating that work thousands of times is infeasible; either you pay Arbitron to do it for you or you build your own mini-Arbitron to do it in-house at a much-higher cost and with less-reliable results. You can take a third option and do qualitative analysis: cherrypick what your gut tells you is good, then analyze those against each other; ignore the rest because it's too expensive to get good data.
Tough problem, huh? Read what I said up top. We need to be able to ask more questions about broad spans of data if we want useful information.
Can't be. It's early. Inflection was Sep, 2015; I'm looking at Aug, 2017 to Mar, 2018 to see the next recession.
I need more time, dammit!
Actually, unemployment numbers count discouraged employees in U4. U4 is currently 4.7. This year, starting in January, it has been 5.2, 5.0, 4.8, 4.7, 4.5, and 4.7. U3 has been 4.7, 4.5, 4.4, 4.3, 4.4.
How would you get positive outliers? Are there positive outliers?
We can conjecture about things all day; the facts here, however, are that data confounded by outside variables to the point of becoming a negative outlier is removed from the system. They may not be statisticians over there, and their reasoning might be incorrect; the action taken is the correct action. I don't have any facts about any other actions taken. Bringen sie mier data.
The count of eyeballs is not correct because it suggests that the count is different for the other airings of the show.
You do realize that not removing this distorted data in any professional capacity will get bad things on your doorstep, right? For example: if you have a deviation from the median greater than 1.5x the inter-quartile distance and you keep it in your data to show safety and efficacy of a drug, you are committing fraud and the FDA will fine you and pull your drug from the market when it turns out to be more-dangerous and less-effective to the general population than you said it was.
I took statistics. My teacher got pissed off because I was failing. I never did my homework, napped in class, and misanswered four problems on my exams all year--and every one of those was due to misreading the problem. I got a perfect score on the AP Statistics Exam. When I tell you that data impacted by outside variables pollutes the model and gives you invalid results, that's an expert opinion.
It's amazing how fucking horrendous people on Reddit and Slashdot are at statistics. A few months ago, I had to deal with a guy claiming some research was invalid because the researchers used people of different backgrounds and so, because the participants were not all substantially-similar, the statistical trend could have been caused by some other variable that differs between them. That's how bad you people are at this shit.
I can use real numbers to commit fraud by not removing data that needs to be removed. They removed data that needed to be removed for their statistical model ("Ratings") to be valid. If doing it the "right" way gets you an incorrect result, then that's not the right way to do it.
tl;dr the right way to handle this would be for the Nelson system to have automatically removed those numbers by identifying them as statistical outliers. It's generally-trivial math. That leaves a smaller problem that things which can hold their popularity even against elections and shit still incorporate confounded data and so come out less-wrong instead of corrected.
How do you design a machine so bad it breaks?
Because it's novel, thus unknown, and you can get people to see you as a source of security by making them feel threatened by something else. They see that you saved them with knowledge, and come back to you.
Unethical journalism to construct stories of villainy from big organizations people trust is a staple of ProPublica. They've repeatedly released American Red Cross Lessons Learned documentation--stuff that says what went wrong, what went right, what they'd do different if they did it again, why it went wrong, and what needs more review--with headlines and fanciful stories claiming that ARC is ineffective and is trying to cover up and ignore the horrendous mismanagement of their relief efforts.
In case you missed it: the existence of LL documentation demonstrates the exact opposite.
ProPublica also attacked Amazon's search because it shows a lower price+shipping for an Amazon product based on the free shipping Amazon offers (on orders over $25), which they demonstrated by tacking on the cost of expedited shipping to Amazon-sourced products. So you can get it for $23 and free shipping from Amazon, or $21 + $3.99 shipping from BuyOurBooks; ProPublica says it's more-expensive from Amazon because non-Super-Saver-Shipping orders have to pay $23+$6.21, and thus concludes Amazon should have ranked the other option as the top, cheapest option, and is only putting their products first to scam you into sending them your money.
Who doesn't have an entire backlog of stuff they want to buy at some point? I just tack something else I didn't need right away onto my order to pad it out.
When you sit down and weed these things out, you start noticing that a deep inspection makes it look like a whole lot of yelling about nothing. The presented facts are organized to tell an amazing falsehood, and suggestions are made to do things "better" which would be objectively-worse. The point is most people don't sit down and debunk the story; they apply their own political bias and accept or reject it. If they don't yet have a strong bias, then they let you spoon feed your opinion to them.
Now you have an audience who swallowed your bullshit about how someone they trusted has betrayed them. You are their savior. They come back to beg you for more information, to tell them more of the wisdom they have missed in their vulgar, miserable, wretched lives. They give you ad revenue.
Of course we push hundreds of attack pieces against anything new and trendy. It's new and trendy, and most people are either against change or indifferent; we can convert those people to returning revenue!
Yeah this is bad journalism. Can we get the whole quote added to the summary?
No.
If you put a 7-cent part in a machine where a 3-cent part typically falls within the lifespan of the machine, you waste money (and labor). Likewise, you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts, and so costs 4-5 times as much.
what you say here, thus, is only valid if one would replace a few 3-cent parts with 7-cent parts, AND LEAVE OTHER KEY-COMPONENTS (aka, which are critical to the working of the machine) with 3-cent parts.
These are the same statement. I made the first; you responded with the second.
Just for redundancy's sake:
you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts, and so costs 4-5 times as much.
*lifespan* of you machine goes up when you put high-quality parts in it. Ergo, when you replace the 3-cent part components with 7-cent part components, then what 'falls within the lifespan of the machine' is exactly that, which the higher components have elevated it to. It's not a fixed lifespan, based on 3-cent parts.
Wow, you're a genius. I told you that a more-durable machine uses more-expensive (and more-durable) parts, and from that you reasoned that a machine will be more-durable if built out of more-durable parts!
No, I said a bunch of shit first, and then you repeated half of it while telling me I'm wrong. You're a dumbass.
You originally said: "If you put a 7-cent part in a machine where a 3-cent part typically falls within the lifespan of the machine, you waste money (and labor)." This is obviously only true if there are more than one key component and you *only* changed one of those components to a higher quality one, and all the rest are 3-cent parts
So, if
the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts
then the use of more-durable, more-expensive parts throughout the machine results in an increased lifespan? You know, what I fucking said right after the sentence you cherrypicked.
Here, let me un-fuck your stupid blathering.
And if you're arguing *you* already said that: you did! You originally said: "you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts." This is obviously true, as the more-durable machine is more-durable because it uses more-durable critical components. There's no other way for a machine to become more-durable, after all.
Now there's an accurate statement: I already said all that shit before you started talking.
you may find that the expensive part can go in a more-durable machine, and that said machine often replaces 6-cent parts with 35-cent parts
Thus, I repeat: if you change the 3-cent parts (note the plural) with 7-cent parts, your machine WILL get more durable.
Good job repeating again what I said way the fuck up there before you got involved, like you thought of it first. I'm not going to let you try to rip off my post and pretend you invented the thing.