Additionally with two pair heads up you need to get it in, the stuff about making reads gives amateur players the license to go nuts because they feel like it, again +EV. Dolly knew the odds better than almost anyone for a very long time, objectively we've learned a lot about poker in the last few years, and it's not fair to judge his play vs today's standards.
Is this outdated in general? From the talk below about GTO-and-heads-up is the assumption that it'd be dangerous now to assume anyone who's still in at that point doesn't have something better than you?
Situations like this are akin to an "I cut, you choose" situation. Hypothetically, you want to size your bet such that all possible responses lead to the same expected negative EV for the opponent and positive for you, with a layer of fuzziness so that your bet size doesn't always give away what you hold. That's what is meant by non-exploitable game strategy. Doyle's original advantage was knowing that bet sizing and fuzziness more precisely than anyone else, from a combination of intuition and calculation, and also in reading the opponent's propensity to inaccurately calculate and take advantage of that too. What's changed in the last few years is that now everyone and their bots know the proper calculations as Doyle always did - and they're also good at reading your bet sizing to correctly guess that you have the two pair.
Your description is still right for a single situation: with a stack advantage, yes you want to apply pressure. What's changed is the aggregation over many such situations, where your opponents have two pair just as often and play it just as well as you do.