Why have capacitors improved so much over the years?

The best are still vacuum dielectric or air dielectric. They’ve been around forever, but they come in small values at high cost. I use these exclusively in my phono / pre.

Film and foil are next best; styrene is pretty neutral, teflon a bit bright, polypropylene a bit dull. Some examples are ’very’ rather than ’a bit’.

Metallized are constructed by depositing metal onto the dielectric. Quality appears to depend on the thickness of the metal and the junction of the metal to the wire leads. They can be almost as good as film and foil, or almost as bad as electrolytic. Mica and teflon are the best.

Ceramics tend to be inferior to metallized, the quality seeming to depend on the dielectric constant of the ceramic, lower being better.

I don’t have strong opinions on paper in oil, but I really don’t like leaks, so I avoid them.

Good luck!  YMMV

The big difference is in manufacturing tolerances.  Back in my early days in electronics (60s) 5% or even 10% tolerances in values were frequently used because it cost money to get to 1% tolerance or better.  MilSpec parts often had to measure far more accurately than consumer electronics.  The same was true for inductors and resistors.  The simple truth is that a 50 microfarad capacitor that accurately measure as 50 mfd will sound the same as any other that accurately measures as 50mfd.  Depending on the values of the other components in the circuit you cannot hear the difference, assuming they are all accurate to a given tolerance.  Usually, the differences in rated voltage and tolerance are used for the purposes of obtaining longevity and reducing failures due to arcing which results in a short circuit.

The other truth is that snake oil salesmen are far better psychologists than most university PHDs.  They have been around a long time because they know their craft and it works.  

WOW!  Short Answer!  "manufacturing process and technology, engineering !"  Hmm, Why are micro processors, HDD's and now SDD's Smaller Nowadays?? or the iPhone has a zillion more processing power than the computer that took us to the Moon..  Hmm, the Capacitor question is a Great one, I'll do a tad more research on that.. 

Barjohn, that's the engineer's answer from 1980.

The book that changed all that was Horowitz and Hill, 'The Art of Electronics', which had a table of capacitors and their actual characteristics. With real information about the physical effect known as 'Dielectric Absorption', or DA, applied to everyday electronic components, the stage was set for a genuine improvement in sound quality.

There are a few topologies which sound good and don't much profit from better components, but the best ones can be much improved thereby. In my experience, anyway.

Build yourself a breakout box in which you can change an important cap from one technology to another with the flick of a switch. Get a book on psychology experiments to learn how to do a single-blind experiment, and recruit your lady as a test subject. I think you'll get quite a surprise.

Here is a person who has done a lot of subjective testing with many different brands and types of capacitors. Might be worth reading before diving headlong into capacitor rolling. It saved me a lot of time and money.

http://www.humblehomemadehifi.com/Cap.html