1) Eat food. This is basically saying eat stuff that is traditionally considered food in different cultures (i.e., stuff that has been eaten for centuries) because doing nutritional studies is hard, but the fact that certain cultures have thrived for centuries and even millenia eating "food" provides decent evidence that it works.
2) Not too much. This is necessary as a restriction to #1 because in the past it wasn't easy or even possible to eat "too much" since you didn't have modern conveniences that make it easy to prepare food and you couldn't just order it for delivery. So you have to explicitly ensure you don't eat "too much", something that was implicitly enforced in the past.
3) Mostly Plants. If there is one thing all nutritional studies have found, and the one thing that pretty much all the science in this area can agree upon, is that eating plants (i.e. vegetables) is good for you, and for the most part, the more the better. But also note that he is saying "Plants" and not "Veggies" in a nod to the idea that it should be largely simply prepared and as close to its natural formation as possible, so buying a bag of "veggie straws" does not count.
Plenty of cultures heavily process their plants, and for good reason: the raw form is less nutritious. Plants != vegetables. Vegetables exclude fruits and uncultivated plants; wildflowers are delicious and definitively plants but debatably veggies [1].
Chocolate is an example of a food that is highly processed just for it to be reasonably less bitter and unpromising as it is in raw form.
This study gets around this because the trans-fat ban (which took a long time to come because of the above factors) was a nationwide thing and so lack of access means we can make assumptions. Even then they had to figure out what was this ban, what was smoking becoming less common, and what was increased use of statins, which is statistics that depending on input assumptions.