Again, your system is too open to uncertainties and not practical to implement. And even supposing it was possible (its not),
How is it impractical or impossible? I don't know if it's economically sound, but the mechanism I described is very simple to implement.
Bitcoin has changed price a lot mainly because its a very small market. Trying to adjust inlation because of some exchage rate wont change it too much. Any big influx of money will make it go up and any big sell off will make it go down quick. As Bitcoin grows this will disappear.
I don't consider fluctuations inherently a problem. Like you say, that's the nature of a small market. The problem is the wild-eyed speculation that coins will be worth 1000x in a few years if we just push this thing hard enough - I think that's undermining Bitcoin terribly.
My goal isn't to prevent all price fluctuation. I'd definitely want a very heavily-smoothed inflation algorithm, to let the market do its thing (arbitrage, market making, forex, etc) over the short to mid term. The inflation is to limit to long-term price speculation to prevent bubbles during deployment - training wheels until it's big enough to ride on its own.
Fiat currencies are not "trusted", fiat currencies are accepted because the government impose them. If the governments were ot forcing their fiat currencies nobody would accept them.
That's a fair argument. Let me take a different angle: Why is this less trustworthy than Bitcoin? It's preprogrammed inflation, not at the whim of a central authority - which is a lot like Bitcoin - but it adds a stabilizing outside influence. It also adds uncertainty, but is that uncertainty bigger than that currently surrounding Bitcoin's future?
And there is lot of people trusting bitcoins (including me).
You do, but take a look at the market cap: that's a pretty direct measure of the global trust of Bitcoin.
I trust Bitcoin to function as designed, but the design may be economically flawed. How are we going to achieve widespread trust if we can't get past the current small-scale problems?