Yeah, I'm not sure how to model the fat+sugar problem I keep hearing whispers of. I know from the research I did with my wife's gestational diabetes is that the glycemic index of a food went down if you added just about any other food volume that with a lower glycemic index. The best example I found was that toast is like the worst possible food you can eat in terms of GI, it was higher than anything but pure glucose . But toast with jam, actually had a lower GI, since sucrose has ~half the GI of glucose (makes sense since sucrose is glucose+fructose, and fructose requires time for the liver to process). Toast with butter was even better, since butter fat has 0 GI.

I still don't know how much I trust the claims that fat+sugar is particularly bad, but I suspect it may have to do with some kind of second order insulin response. Like, the presence of triglycerides in the blood cause fat cells to be more likely to respond to insulin, storing fat, instead of releasing it. I dunno.

Also keep in mind GI is falling out of fashion as a key indicator. Glycemic Load seems to be more indicative of poor health indicators. That is to say, that the area under the curve is potentially more important than the max height of the curve. I'd say limiting both is probably important as well. I also highly suspect that the slope of the curve is very important in how we perceive hunger, which is why we tend to feel hungry shortly after eating foods with fast carbs...

FWIW, at the end of the day, in order to keep my wife healthy, we ended up giving up fast carbs, and she had to keep her carb count low in general. Her basic formula for a snack was to add the grams of fat+ grams of protein, and if that was bigger than the grams of carbs it was safe to eat.

Sorry I don't have links to this stuff, this was like 5 years ago, and I'm not sure if I ever saved those bookmarks. I can say it was really hard to find credible data sets. There didn't seem to be anything from America (draw your own conclusions :) ), most of everything I found was published datasets from Australia and New Zealand. Also, don't take this for anything other than anecdata, do your own research. You'll find very quickly a lot of really bad nutrition information. I had to plow through most of it, and might have given up if it wasn't for my pregnant wife, coupled with a friend with a digital signal processing background pointing out engineering problems with most of the current thoughts on controlling diabetic blood sugar levels.