Listen to Elgar's Nimrod:

compare to a typical modern version

We know which Elgar preferred, but he's dead. How do the different tempi change how YOU perceive the music? The emotional impact. Then think about it in the context of what comes before and after...

Compare Strauss conducting Heldenleben-

not just the tempos, but the very straightforward approach that seems almost nonchalant to our modern ears. Here's a typical modern version with Jansons
; 8 minutes longer (and some out there can be yet another 8-10 minutes longer than this- anyone remember last summer?) Are the differences more in the slow or the fast parts, or in the amount of rubato? How does it affect your perception of the music?

A bit more- Ravel plays Ravel:

; Volodos plays Ravel:

Saint Saens plays Saint Saens:

Rubinstein plays Saint Saens:

You see the pattern... I would be interested if anyone knows of a composer that played slower than people today play his works. I don't know of one.

That is not to say that the composer is right simply because the music is theirs- George Lucas ruined Star Wars for many fans- but I think there is something to think about in the forward and directness of their approach in comparison to the interpreters', particularly because the pattern is seen again and again. (other ones to check out: Rachmaninoff on his own works, Prokofiev playing Pno Cto No. 3, Shostakovich, Debussy...)

Performers today, more often than not, play slower than the composers would have chosen. How many composers have you heard must have had a 'broken metronome?' We need to think about why, and how this affects the audience's perceptions of the music...

Let me leave you with this passage by Harvey Sachs' discussing of Rubinstein's 1929 recording of the Brahms Second Piano Concerto:

"... this version is worth hearing, inasmuch as most of its tempos are shockingly fast compared to those that have become the norm in our day. Brahms did not often put metronome numbers in his works, but in this case he took special care to do so. If one were to follow his indications inflexibly-which, of course, was not his intention-the first movement would last about 16:30, the second and third about 7:30 each (including the repeat in the second and allowing a great deal of leeway for the slower parts of the third), and the finale about 8:50. Here are the timings for two thoughtfully and beautifully played modern recordings: Alfred Brendel with Claudio Abbado and the Berlin Philharmonic (Deutsche Grammophon)-17:52, 9:20, 12:15, 9:21; Vladimir Ashkenazy with Bernard Haitink and the Vienna Philharmonic (Philips)-18:40, 9:26, 13:07, 9:29. The timings for every movement of Rubinstein's 1958 recording are also slower (16:53, 9:05, 12:39, 9:02) than Brahms indicated but not as broad as Brendel's or Ashkenazy's-except the third movement, which is slightly shorter in Brendel's version. If one listens to the well-known Horowitz-Toscanini-NBC Symphony recording of 1940, one hears a performance that is closer still to Brahms indications (the timings are 16:15, 8:06, 11:05, 8:25) and-in the case of the first and fourth movements-a jot faster than the composers guidelines suggested. But to contemporary ears, Rubinstein's 1929 recording (timings: 14:35, 8:10, 9:09, 7:54) sounds rushed in every movement except the second, and absurdly so in the third, which, however, is not played nearly as quickly as Brahms indicated.

...When one listens, today, to a recordings of the Brahms symphonies conducted by Felix Weingartner or Toscanini, one hears tempos that are considerably faster, on the average, than those to which we have become accustomed in recent decades. And yet Weingartner had conducted Brahms for Brahms (and had received the composer's praise) and Toscanini had modeled his approach on that of Fritz Steinbach, who was one of Brahms's favorite conductors. Barth, Rubinstein's teacher, had played Brahms's music in the composer's presence and the young Rubinstein had heard Brahms played by many other proto-Brahmsians. I would be willing to bet that, despite its defects, Rubinstein's 1929 recording of the concerto was closer to what Brahms had in mind when he wrote the work than is Rubinstein's 1958 recording or any of the later ones by other pianists."