A limitation here is that this approach will always only represent the change in prices of the representative items in the basket.

Why is that a limitation? If you make your list out of relevant and largely unavoidable expenses - food, housing, transport and utilities - surely your represenative list will remain representative for long periods.

So the Second way is to take the value of a set of transactions, such as GDP, and measure what would have been the value if the prices had not changed. This gives the so-called "Real" GDP. The ratio of actual GDP to "real" GDP gives the index, called a "deflator", and the rate of inflation is the percentage change in the deflator.

This also ties GDP closely to 'growth', because a lot of apparent growth will come from increased prices - and this has a worrying circularity about it, as a notion.

And so there has been a move from a fixed index to comparing each year to the previous year, which is called a "chain index".

This is now making less and less sense, and moving further and further away from anything that might pass as a coherent understanding of what inflation really means to most people, which is the fact that the same sum of money buys less than it used to.

The PC utility argument is nonsensical, because it's equating being able to watch claymation with being able to eat and keep the rain out - or in other words equating discretionary disposable income with basic living costs.  

I can buy the volatility argument, however, especially for prices with reliable seasonal variations. But it should be easy enough to average those out - it's standard practice for retail sales figure in the UK.

So I think the basket idea remains the most useful by a long way - especially now.