Please do and thanks in advance for any insights you can provide -- it would be great to understand any benchmarking improvement with ada-002 from your previous findings, and whether you tested the specific OpenAI text-search-*-{query,doc} models as a comparison for large document search.
"Ten thousand might be fine if it meant saving Afghanistan from Taliban rule, as long as that as low as coalition forces could keep civilian deaths."
The same thinking could then be applied to our "war or drugs" -- the killing of several thousand innocent Colombians could be easily justified in order to eradicate the cartels, albeit temporarily.
Do you find that a satisfactory trade-off as well?
Government stated inflation based on chained CPI and other new measures to reduce inflation may be at around 2%, but previous formulas used in the past to calculate inflation would show a much higher rate (6-10%)[1]. For those actually buying or consuming anything, that official number is a farce.
Shadow stats isn't credible in any way. They can't even be bothered to take the raw data and apply the 'old' weights to them, so they just arbitrarily add a few percentage points to the CPI. The BLS has dedicated far too much effort into showing why they are an asinine source for 'data':
If "real" inflation was 7%/year, in 10 years, the price of everything would be 100% higher than it was previously. How many things in your life are twice as expensive as they were in 2004? Now how many things aren't twice as expensive?
Apropos of nothing, and since you mentioned it, a surprising number of things are at least twice as expensive now as they were in 2004.
Nearly all produce products, gas, houses, rents, cars,
and clothes. Beer, wine, and soda seem a bit less impacted.
If you want you can play this game yourself by looking at archived newspapers in Google's newspaper archive (http://news.google.com/newspapers) from 2004 and them comparing them to present day newspapers.
That would be a much more interesting set of graphs if the Fed would let you in on how they compute the index.
That is why looking at newspaper advertisements are actually pretty cool. They don't use, Detroit for example, when computing the cost of suburban housing.
Here is an example, I don't know if will work for you. Marysville Washington, a quiet little town, the paper has a 2004 edition online [1] and on page 21 of 26 are the rental listings. 2 Bedroom, 1.5 bath apartments are about $500. If you look at 'forrent'[2] a rental search service for Marysville, two bedroom 1.5 bath apartments today are closer to $1,000 a month. It may not be exactly 100% and its hard to pin it to an exact unit (would be interesting if you found the same unit in a newspaper and online) but not taking the Fed's word for it, and actually looking at papers that were published back then which is key in my opinion, shows a much higher increase in costs than the Fed reports as having occurred. That is why I wonder how they compute their indexes (but the last time I dug into it they claimed secrecy so that people wouldn't try to 'game' them.)
Not trying to argue, just getting a different result when I go to the primary sources.
Thanks for the links, what I was actually trying to argue were two things, one was that the larger the data set the more homogenized and useless the data becomes, and two that looking at data for a given city returns different results.
There are lots of ways to come up with insights into what is happening, one way is to take the prices experienced by an individual in a single location over time, then to take the population of each set of locations and compute p(I|L) where I is your experienced inflation rate, give L your location. Then you can ask questions like "What is the likelyhood that a given individual has experienced an inflation rate that differs significantly from the reported rate? (either positively or negatively)"
Charts like http://www.city-data.com/ which can show you housing pricing changes by city for example you can see the wide disparity in changes between coastal vs more central states. Some states would have experienced very little change, others quite a bit. Now overlay that on population. Then you start to get to the answer of "What do 'most' people' experience in terms of inflation." The newspaper archive is a great way to check the numbers.
Thanks again for the links, one of the criticisms for the CPI scale has always been this : "The CPI follows the prices of a sample of items in various categories of consumer spending—such as food, clothing, shelter, and medical services—that people buy for day-to-day living."
Which historically did not include changes in quantity. So a 'can of soup' which went from $1.00 -> $1.25 over 10 years seems like it has a 25% increase in cost, but it went from 16 -> 10 ounces. Which means its actual cost went from $1.00 -> $2.00 when computing the per ounce price. This is another great thing you can see/check with newspaper archives as grocery prices are exceptionally well documented.
Both are driven largely by Supply & Demand rather than inflationary pressures -- hence why economists rely on Core CPI to make decisions -- but there is good data for both;
Think about the rest of your life though, especially if you're outside the SFBay area. Are rents twice as expensive? Did an equivalent Camry cost $11k in 2004 vs. the $22k it costs today?
The data provided in my link shows 4-6% average, I was referring to the most recent spike in real inflation that is not showing up in the BLS data. At the lower end, you'd have 48% price growth compounded annually over that 10 year period, at the higher end is 79%, so not sure why you extrapolated that at 7% over ten years.
I'll admit, I didn't even click on your link, but rather went by your claimed range:
> Government stated inflation based on chained CPI and other new measures to reduce inflation may be at around 2%, but previous formulas used in the past to calculate inflation would show a much higher rate (6-10%)
I'm not here to quibble about inflation ranges, but to point out how idiotic their 'methodology' is. Don't take my word for it;
Or my favorite, via the libertarian blogger James Parsons. If you actually use Shadowstat's inflation ranges, you can 'back-calculate' the price of assets to show how insane they actually are:
* If shadow stats inflation assertions are correct, the 'real' value of housing has dropped 60% since 1980!
Agreed -- looking at the player data[1], IMO the format type is unrecognizable:
## GK / Goalkeepers
Kawashima|Eiji Kawashima, 20 Mar 1983
Nishikawa|Shusaku Nishikawa, 18 Jun 1986
Gonda|Shūichi Gonda, 3 Mar 1989
## DF / Defenders
Inoha|Masahiko Inoha, 28 Aug 1985
G. Sakai|Gōtoku Sakai, 14 Mar 1991
Nagatomo|Yuto Nagatomo, 12 Sep 1986
Uchida|Atsuto Uchida, 27 Mar 1988
Konno|Yasuyuki Konno, 25 Jan 1983
Kurihara|Yuzo Kurihara, 18 Sep 1983
H. Sakai|Hiroki Sakai, 12 Apr 1990
Yoshida|Maya Yoshida, 24 Aug 1988
Masato Morishige, 21 May 1987 ## Japan F.C. Tokyo
Comments as a double-hash, key fields are either player last name or occasionally first initial-space-last name, then three different delimiters of pipe, then comma, then tab. Choosing either a consistently delimited format or a more verbose JSON/YAML structure with clear metadata would seem to be a better approach.
I had used it to setup a test server and had no issues -- FYI, the date on the article is today, but the original article has been available for a few months so perhaps the article has been updated.
