FWIW, comments on a few things written above:
KT-88's are beam tetrodes (as Ecclectique alludes to), so there is no pentode connection available as with EL-34's for example. I was under the same misimpression until very recently, and there are good reasons to be confused, because the terms have gotten tossed around somewhat indiscriminately since the time beam tetrodes were first manufactured as an alternative to pentodes. (These technical developments were under patent at the time, and from what I can infer it seems the tube company marketers might have desired the positive association customers had with pentodes).
The connection Eldartford describes actually sounds to me like Ultralinear, rather than straight pentode. I do not know whether any VAC amps employ Ultralinear connection.
As for the sound and theory regarding mode vs. mode, allow me to offer my basic take from my experience with VTL amps using 6550C and KT-88 tubes:
As I see it there are two essential phenomena in play. One is the difference in power: roughly double in tetrode what's available from triode. This changes the sound all by itself, and taken on its own, if no other properties were to change, more power from the same number of otherwise identical tubes ought to sound better, or at least more accurate, given a power supply that can keep up adequately with demand.
But this leaves out the second essential property - the presumable reason for the sonic positives associated with triode mode, despite its lower power. This has to do with how the tubes diverge from linearity, otherwise known as distortion. As I understand things, when compared within comfortable power margins, triode-connected tubes offer lower overall distortion than when run in pentode, and what distortion remains is more benign in nature, i.e., lower in order and more weighted to the even harmonic series. This greater purity is what you trade off to some degree in order to obtain higher power from the same tube connected in tetrode or pentode.
These two characteristics - output power and harmonic distortion signature - taken together, pretty well explain what I hear comparing triode and tetrode modes with my amps.
Higher power sounds like...higher power. Better speaker control, which can translate into a host of audible qualities ranging from increased soundstage size and image separation to more bass tautness and less overhang. Greater dynamic capability, which can translate into less compression, more impact, and better microdynamic expression. The ability to play loud music louder and still maintain dynamic contrast and overall authority. Greater power, considered in isolation, even seems to be able to allow for greater transparency - but the degree to which extra power will benefit the sound has to do with many other variables, including speaker sensitivity and impedance, room size, type of source material played and at what volume, etc.
Experience, though, tells us that in actual practice, many of the positive qualities associated with the higher power of tetrode in theory, are in reality somewhat compromised in their ability to convey musical enjoyment, due precisely to the less agreeable harmonic structure of the more powerful mode. Who wants to hear music played louder, with less compression, if that music has taken on more of an agressive edge? Who needs the clarity of better driver control if a good part of what it lets you do is hear more clearly the extra, higher-order distortion the amp in full-power mode is injecting into the music?
There's another variable at work as well. When we talk about push-pull amps, we're talking about a circuit that inherently works to cancel even-order harmonic distortions. This serves to leave the odd-order distortions more nakedly exposed to the ear. (This fact, rather famously, also goes to part of why people may like single-ended triode amps - despite their very low power - which leave the triode's naturally low-, even-order harmonic products intact, instead of canceling them. The lower power does mean that total distortion levels rise much sooner and higher before clipping than with higher-power push-pull amps [even if SET clipping does approach more gracefully when it comes], but the low-order, even harmonics which the ear finds more musical predominate.)
Thus, a push-pull circuit operating in triode mode gets a double benefit on the distortion front: triode generates mostly low- and even-order harmonic artifacts, which the output stage architecture then tends to cancel out, leaving both less residue, and less objectionable residue. Tetrode or pentode by contrast creates a double-whammy applied in push-pull: higher- and more odd-order harmonics are generated than with triode, and then on top of that the output stage configuration tends to cancel-out what lower- and more even-order harmonic content does exist - content which would otherwise help to mask this less-pleasing timbral quality.
These reasons explain the continuing efforts - necessarily at high cost, since more tubes are required to achieve comparable output - to build high-powered push-pull tube amplifiers wired in triode, instead of simply going for the bigger bang for the buck possible from tetrode or pentode. (Witness, for instance, Atma-Sphere, which uses only triode tubes and many of them; CAT, which uses beam tetrodes but wires them exclusively in triode for sound quality; or of couse VTL, which makes mode-switchable amps so high-powered, you can run them in triode and still have juice to spare. VAC has even gone to the expense of making high-powered push-pull amps using multiple pairs of one of the most expensive triodes, 300B's, which we normally only see used singly in SET's.)
So, especially when played at lower volumes or with music that is not as dynamically demanding, triode will often sound best, this being manifest in a variety of ways, and again dependent on several other circumstances (including the particular output tubes used). But in general, it is possible to say that triode will usually sound more timbrally pure and natural, texturally cleaner, and with less artificial emphasis on silibants and harsher overtones.
A lot of ink is spilled in the above posts trying to make generalizations about the sounds of the two modes, but I think Trelja finally comes closest to the truth when he hits upon the term 'sweetness'. (I don't believe he means to connote the word with the modifier 'cloying' implied, as in added 'sugar', but rather more an absence of bitterness, as with the way really good mineral water can taste 'sweet' in the absence of the usual contaminating pollutants, but with traces left intact that our evolutionary history informs our senses are healthful.) This is cutting to the chase of what's most advantageous about triode and weakest with tetrode, though many other qualities, some of which may be more important with any particular musical program, will always pertain. What I've tried to do here is hopefully help explain why this may be so.
It is also important not to oversell the differences. The same amp, with the same transformers and power supply, fitted with the same tubes, and playing at moderate volume - especially with music of a not too-demanding nature, or not with highly specifically-flawed or extremely revealing recorded sound - will in reality probably not sound revolutionarily different played in one mode vs. the other, and that only makes sense.
As the volume gets turned up though, triode may tend to suffer more audible compression and loss of control, although only because of the lower power available (unless maybe you have one of the monster triode amps and/or easy-to-drive speakers), and not because of any inherent shortcoming about the mode per se. So frequently you're presented, as with so many things in life large and small, with a choice, when playing an average (or averagely bad) recording: to have that kick drum really pound you in the gut the way you know it can, or to keep that high hat from cutting off your head at the ears with what feels like white noise. Or just turn it down and play lute music.
In my amps, for whatever reason, I have found that KT-88's made by Electro-Harmonix tend to thrive more, relatively speaking, in triode mode than did the Svetlana (SED) 6550C's they replaced. So I think it is possible that there are tube-specific differences as to how amps will respond in either mode, and some tubes may do better in one mode than the other, or better in one mode than another tube type or brand will do in that same mode.
