It's not the solvent itself, but the solvent does play a role. Basically, the formulation of the water-free version is different and this difference makes aerial oxidation virtually impossible. The acqaeuous version isn't immune to oxidation in the same way. If you search around a bit, you'll find a thread or two that goes into the production methods of the water-free version and this also clarifies the inherent difference in keeping properties.
Elaborating on this thought, I think there is a good chance that oxidation by O
2 is catalyzed by either free hydrogen ions (H
+), or free hydroxide ions (OH
-). These ions are present in water, either as a result of dissociation of H
2O into H
+ and OH
-, or by ions from the buffer, which will usually be hydroxide ions in common developers.
Most common developing chemicals, like metol, are active in the negative ion form, meaning that that are more easily oxidized in that form. Hydroxide ions from the developer buffer convert the free developer into negatively charged developer (through acid/base chemistry), and in that form they are more susceptible to oxidation by O
2.
The new watery HC110 is water based, so it will have free ions present in solution, and if my theory is correct these ions will be available in the concentrate to convert developer into the negative ion form that is susceptible to oxidation.
The old syrupy HC110 is not water based. The form of acid/base chemistry that is available in water-based developers is not present in solvent in the old syrupy HC110. Therefore, the developer is not very susceptible to oxidation in the concentrate. To elaborate, the solvent is, I believe, diethanol amine, which is a base, but in the absence of water it does not generate free hydroxide ions, so free hydroxide ions are not available to convert the developer into the oxygen susceptible form. Once the concentrate is diluted into water to make active developer the diethanol amine molecules from the HC110 concentrate react with water to produce free hydroxide, and the free hydroxide ions then convert developer into active form.
There may be some additional subtleties involved in the chemistry, but my bet is that the description above is probably pretty close to correct.