Originally Posted by wstar

We agree on this, although I think the new knock sensors are supposedly picking it up much earlier, when it's a tiny hint of a knock. You don't actually have to get to where you're risking damage to have the sensor "see" it and adjust.

Here's where we potentially differ (not that I'm solidly in the other camp either, I'm just playing out their side of the argument as best I can, and it's reasonably sound, but has question marks):

The way that the ECU responds to knock is not to "throttle back" literally with the throttle, but to retard the timing, which eliminates the knock at the cost of performance and efficiency. You mentioned this yourself in an earlier post, but I just wanted to be clear on that issue:

Similarly, if you put 110 race gas in a car that was designed and tuned for 93, there's lots of "knock headroom", so to speak. With the higher octane fuel in the tank, a tuner can go into the ECU parameters and advance the ignition timing more aggressively than stock to create more power than was possible on 93 octane due to the higher resistance to knocking the 110 fuel has (the inverse of what we described above, where retarding it reduces power and helps avoid a knock). There's still only X MJ/liter of energy in the higher-octane gas, but by being more knock-resistant, it allows a more aggressive spark timing advance, which results in increased horsepower.

I *think* we both agree on the above - that by tuning the ignition timing manually with an ECU programmer up as far as you can without knocking, on any modern-ish car where that is possible, you can get more horsepower out of 110 gas than 93 gas, even though they both have the same MJ/liter of potential energy within them.

The question mark is whether the ECU is smart enough to do this by itself. Some have claimed that when it detects absolutely no hint of knocking, it slowly advances the timing more aggressively until it finds the first hint of it, and thus self-adjusts upwards (more aggressive spark timing, more power) for higher-octane fuel in much the same manner that we know it self-adjusts downwards for crappy fuel.

Nobody has put out any hard evidence that this is true, but people have mentioned it, and the 110 race gas dyno results seem to indicate that *something* is going on, and this explanation kinda fits the bill.