To be more precise: inflation is a margin of safety against deflation.
Imagine a medieval world where the money supply is fixed. New money is never created, it can only be transferred back and forth. Now imagine somebody invents and rents out a steam engine. The world is now richer by this new technology. Your money's purchasing power just went up, since it can buy something useful that it couldn't before. Since the total amount of money is fixed, other unrelated goods will go down in price. If this becomes a common occurrence people will hoard money rather than buy stuff. This conservatism reduces investment in new ventures, and the overall growth of the economy suffers.
In an ideal world we'd be able to measure just how much the economy grew by and add just that much money into the economy. But we don't know how to do that without avoiding a command economy. And perhaps the question is ill-posed anyway.
Inflation is hard to control, and it can be 'captured' by special interests, leading to tragedy of the commons. We're still looking for a better way to avoid deflation.
The problem is that the word "deflation" has been corrupted to refer to two entirely different phenomena:
1) It can refer to generally falling prices due to increasing productivity. This is entirely beneficial. Video game consoles, transistors, etc, experience deflation and it results in consumers getting better products.
2) A contraction in the money supply, or a shock to the demand for money. The classic example is the bank run that wipes out people's deposits. This type of deflation is terrible because it results in falling aggregate demand, idle factories, and unemployment.
The first type of deflation is beneficial is good, and the second kind is very bad.
The Fed and mainstream economists believe that both kinds of deflation need to be stopped. Economists thus argue for constant inflation to prevent even the slight possibility of slipping into deflation.
To me, this is like arguing that a balloon should be constantly inflated to prevent any chance of it ever deflating. That's not a recipe for preventing deflation, its a recipe for blowing a bubble and then having a disastrous popping. Which is exactly what has happened happened.
I don't see why the first type is beneficial. It creates the same sorts of incentives to be conservative as the second.
Your technology examples have two properties: they're non-essentials, and they're seldom more than a small fraction of most people's assets. If productivity caused more widespread deflation it would cause the same ill-effects and compromise money's major role: as an enabler of investment.
Here is my broader case against monetary dilution ( what you call inflation):
1) In the modern U.S. dilution happens as follows: First banks loan out the same money multiple times. Eventually, years later, they get called on this and there is a bank run ( example, the run on the money markets a few months ago). The Fed then steps in and guarantees all deposits. A guarantee by the Fed to print out money to back a deposit is no different than actually printing money. Thus the whole cycle is equivalent to the government taxing savings and spending that money on subsidized loans to borrowers that have been approved by the nationally recognized rating agencies.
2) From 1995 to 2008, the money supply was diluting at ~9% a year. Just look at MZM ( http://research.stlouisfed.org/fred2/series/MZM ). This is an extraordinary rate of dilution. As a result, Americans trying to save lost a lot of money in real terms. An investor who ten years ago put their money in 50% CD's/50% stocks would have lost money over the past ten years. Even 100% CD's would have lost money. The only investment that can protect against that kind of dilution is gold. But investing in gold is hardly productive.
3) The actual loans the government printed money to fund, went to McMansions, leveraged buyouts, commodities speculation, and credit card debt. These were neither productive nor in either of our interests. They are the classic case of Austrian malinvestment.
4) The rapid dilution forced people to put their money in any place but dollars. However, this did not produce "investment". It produced massive asset bubbles. People were not carefully considering business fundamentals. They engaged in herd behavior, putting their money where everyone else was putting it. They turned everything from stocks to housing into psuedo-money (ie not valuable because of dividends, but because everyone else thinks it is valuable). Sure there were a few gems that came out of this process, like Google. But most of the money was burned up in Vegas real estate and companies like pets.com.
5) Since I define economic growth as "producing more of what people want", taxing people to pay for subsidized loans does not result in economic growth, even if it increases the GDP numbers. Subsidized loans might result in more oil pipelines being built. But if people wanted more oil pipelines, they could have signaled that to the market by buying more oil. Perhaps what people really wanted to was to retire early and spend more time with family and friends. Taxing their savings to fund oil pipelines would thus make it harder to retire and result in less of what they wanted - that's economic decline, not growth. The government does not know what the people want better than the people, so it should not tax their savings to force them to spend money in certain ways.
6) The asset bubbles have made the prices of everything more expensive, from housing and oil to bushels of wheat. Some of these prices are now starting to come down, but a return to dilution will send them right back up again. As a result, for the past 35 years real standard of living has been stagnant. While Moore's law has brought some new technological gadgets, higher housing and commodity prices have forced us to work longer hours just to make ends meet.
7) Investing is not something that just magically works for everyone. Investing takes skill. A stock market investor is an owner of that company. Most people are not smart enough to be owners of companies. They are not even smart enough to pick out a mutual fund that can properly manage a company. Thus the net effect of the mass entry into the stock market from 1980 to 2008 has been to almost completely destroy it. People make crappy investment decisions, then more regulations are put in place to protected them. These regulations then bring the stock market into a state of permanent stagnation. Just look at SarBox.
8) If a fiat currency could be run with a dilution rate of 2-3%, it would be acceptable to me. We came close to this before Bretton Woods ended. I'd prefer 0%, which would be even better than a gold standard. But I am skeptical that we will ever return to that kind of stable fiat currency. The political pressures to dilute are too great. The U.S. has come perilously close to destroying its currency twice in thirty years ( 1979 and early 2008).
You're absolutely right about all those issues with dilution run amuck. I was thinking about McMansions and LBOs when I mentioned the tragedy of the commons.
My major disagreement is with 8. 0% dilution would be bad for the reasons I mentioned before. Some small rate of dilution would probably be the sweet spot, but better people than me need to figure out whether it's 2-3% or something else.
More tentatively and subjectively, I think you're mischaracterizing regulation in 7. There's been bad regulation, yes, but the response isn't lack of regulation.
If the economy suddenly switched from a 3% steady inflation to 3% deflation, than yeah, that would be very bad for debtors. Conversely, a sudden switch from 3% deflation to 3% inflation would be very bad for creditors and holders of dollars. But I'm talking about what would be good in steady-state. In steady state, borrowers take expected inflation/deflation into consideration. So the net result would be the ratio of home prices and mortgages to income would fall (which would be a good thing, IMO, because I think Americans have too much debt).
