Google wants to use its data to provide an new way of calculating inflation.

I wonder how they'll get around the tech divide. For example: people don't sell gasoline online, but there are numerous websites that track that information. Also, any price index requires some decision over which goods are included in the basket. How would they determine this? How will they be able to determine the ratio of goods purchases online compared to those purchased offline? How will they overcome the "lowest price" problem that tends to come up (fortunately, for the consumer) when searches are done for products online. A simple average? A weighted average based on market share? How would that market share be determined?<p>Lots of unanswered technical details here. Still, it is another data point.

Mostly duped, see: <a href="http://news.ycombinator.com/item?id=1782681" rel="nofollow">http://news.ycombinator.com/item?id=1782681</a>

This is a simple idea, but one which is probably quite hard to execute. Issues that spring to mind: identifying identical products (they may be named slightly differently, or be similar variations on a product - ipod w/ 8GB or 16 GB), web prices may often be different from in-store prices (e.g chains will price goods on a store-by-store basis), plus you got multiple prices for a product (big, small, 500ml, 250ml), old prices, new prices, discounted prices - and what about geographical location? Possibly hard to identify for some sites / merchants.<p>Interesting, none-the-less.

Since about 40% of the CPI is housing, and that component is calculated in a way that reduces the volatility of CPI, and essentially ignored the entire housing bubble, then any alternative to government statistics is welcome. It is as if the Fed was driving a car using the oil pressure gauge as a speedometer.<p>There was a time when the cost of collecting economic statistics was prohibitive for all but a monopoly provider. Now that we are all beginning to walk around with smartphones in our pocket, we can look forward to using them as live economic indicators.

Distilling inflation to a single number is hard even when you have all the prices available.

Governments would never agree to this. For example, in the U.S., The CPI drives so many parts of the budget (inflation indexed bonds, Social Security payments, real GNP vs nominal GNP, etc.) that they would never give control of this to an external party [/conspiracytheory]

I think it is very important for people to know that CPI is not inflation, it is not even a measure of inflation. At best-- and that is, if it were not so greatly "massaged", it would be a measure of the impact of trailing inflation on prices with the effect of productivity increases removed. But even that ignores international money flows.<p>Inflation is increase in the money supply. So the best measure was the money aggregates that the government used to publish. (Though I'm not sure how accurate they were.)<p>This may sound like semantics, but it is this constant under-reporting of inflation that is distorting the market. If you measure monetary inflation and work against that, a capitalist system is much more effective. By replacing "inflation" with "CPI" in people's minds, so many people have been led astray and many economic calculations come out wrong.<p>But this is beneficial for government that wants to print money to support deficit spending without limitation... and so they publish an essentially completely fabricated CPI number. then when you talk about inflation being %20, people say "LOL U Stupid. Inflation %1!!!!!"<p>To predict how prices will go, you project the likely increase in productivity over the next year and the amount of inflation over the last year... but even then, this is ignoring the massive holdings of dollars and treasuries overseas -- which really represent past inflation that whose effects we avoided via the bretton woods and other agreements. If confidence in the dollar continues to drop, those dollars will continue to come home to roost, goosing price changes even higher over the amount caused by the significant deficit spending (and consequent money printing that increases the money supply) minus relatively small increases in efficiency.<p>The noose around the dollar's neck is the accumulated decades of inflation that have been exported overseas. If it becomes untenable to continue holding them, then a selling panic will likely occur as nobody wants to be the last one holding the bag.<p>This is what happened with greece. The US is approaching greece in terms of debt levels and given that it appears the federal reserve is buying a large chunk of the treasuries at each auction (under the table, of course, as if this word got out it would immediately cause what I'm describing) the game can go on for only so long.<p>2008 was not the financial crisis. The financial crisis is in the dollar, which is the mother of all bubbles. I believe, that less than %10 of treasuries have a greater than a year, which means that %90 of the debt is being rolled over each year.