Sure computers may 'get better', but it's software we care about. If Word 2050 needs a X$ laptop then it does not matter how fast that laptop is it's still the same quality as the laptop running Word 2040.
Net result all inflation measures are off which is one reason so many things keep growing faster 'than inflation'.
http://www.bls.gov/cpi/cpihqaqanda.htm#Question_1
If a t-shirt's quality declines, then it's actually getting more expensive, and that should be tracked.
If computers get better, then that should equally be tracked, otherwise you're introducing an asymmetry, where products get better -> price index doesn't change, but products get worse -> price index increases.
In fact, computers have rapidly gotten both better and cheaper over the last decades. A new 486DX2/66 machine ran around $2500 in 1992 -- now a basic desktop shouldn't run you more than $300-400.
> gotten both better and cheaper
last decades sure. Last five years somewhat. Next five years ehh, possibly.
If you need to by X, Y, and Z and Z get's cheaper it can only save you so much money. Price increases on the other hand are unbound.
The things - real estate, education, health care - that are inflating most rapidly are doing that because they are subsidized. When you subsidize a thing, it costs more and you get more of it
Ag subsidies are different because that's a sort of Nash equilibrium - the cost of inventory underruns exceeds the mild price increase from overproducing agricultural products.
The BLS rate is the amount of money charged to insurers, not the money they charge you I'm afraid.
If you don't like the official statistics you can always compare against private measures like the Billion Prices Project.
http://bpp.mit.edu/