Interesting point OP brought up.
I mentioned about high SR on another forum post
Basically, if you have a capable system, by all means record at the highest SR possible - HOWEVER, for best results you should
a) down-sample your consolidated tracks to the fidelity of the lowest SR quality of any time based plug in (delays and reverbs) you have in your session. This is because processing super fidelity on reverbs or delays can produce noticeably unpleasant audio fidelity artifacts (but sometimes cool sounding depending on the type of music)
b) Insert a Low Pass Filter at about 18khz/24dbo on each track that is sending to a reverb or delay based VST. This method does not require you to reconsolidate and resample your tracks, but you will need a very HQ Low Pass filter. Inserting LPF before the super fidelity signal is sent into a reverb or delay will greatly reduce nay super fidelity artifacts.
For your super fidelity recordings or session track, DO NOT DITHER. Dithering super fidelity tracks essentially "scrambles" super frequencies that occur about or below the noise floor or your mastered recording - but you don't want to do this yet, I have my reasons but basically, its better to AVOID using dithering until you are processing your final version.
Recording at 88k is fine, however you should strongly consider making a "Master" version of your track and a "Internet Ready Version". Your Internet ready version should be downsampled to 44.1/16bit with POW-r (or equivilant) dithering. The file version you want to convert into MP3 is the "Internet Ready Version", ther is 0 benefit to converting a 88k file to MP3. When converting a super fidelity recorded file into MP3, be aware that supersonic frequencies will distort in the audible range on your MP3 rendering (depending on how good your MP3 conversion program is.)
Also professional monitor systems use components designed to reproduce a 20-20 linear response curve on biamp/triamp or single amp enclosures. As such, the transistors inside these components are not designed to reproduce frequencies above 20khz or so. Supersonic frequencies hovering at or above the noise floor being sent through these systems usually distort the tweeter component, and it is quite noticeable. This is why most professional monitors have a built in crossover with a low pass filter at 18-22khz.
So whats the moral of the story?
Record as high as you want, but be sure to apply LOW PASS FILTERS, Reconsolidate as appropriate, and convert different versions of your master "Internet version" and "Your Version" before converting to mp3 or any internet ready format.