Well, I said I'd like 'less' anomalies, not 'none'. No rule system will be perfect, but after the twentieth airless rock with a pop of 9 billion and TL3, my imagination doesn't want to be 'stimulated' any more. One or two problem worlds is fine.
What I'd do would be to give each problem world a quick 'think-over' to see if I could come up with a quick, facile explanation that'd work for that time and place. If I couldn't, I'd give it a "saving throw against weirdness" (difficulty depending on how weird it was -- e.g. anomalous world stats would be weirder outside the stomping grounds of the Ancients than inside it). If it made the saving throw, I'd dedicate 10-15 minutes to working out an explanation that didn't involve Ancients or once-in-the-lifetime-of-the-universe coincidences or lamp-hanging ("It's a Mystery" explanations). . If I still couldn't think of anything I'd give it a second throw and if it made it, I'd try again, this time not barring Ancients or once-in-the-lifetime-of-the-universe coincidences or lamphanging. But the explanations have to work on another level, making up in fun for what they lack in plausibility! And if I
still couldn't come up with anything, I'd change something -- preferrably in a way that enhanced the RPG adventure potential.
This way, half the worlds would take ten seconds apiece to vet (nothing wrong in the first place); most of the rest would require maybe five minutes, and a few would require a quarter of an hour.
My knowledge of the OTU is sketchy, but I thought quite a bit had been UPPed, and any of these worlds that couldn't occur under the new system would need to be retconned - making quite a lot of work. As you say, if it works, don't fix it, but I don't know how much of the OTU works.
There are parts of the OTU in dire need of a retcon, true. But that has nothing to do with any revised system that might be developed to make sure new regions won't suffer from the same problems. And there remains the problem of the worlds that'd work just fine when considered in isolation. One huge problem are the statistical anomalies. Those are hard to argue against, because you can't point to any one world with corrosive atmosphere and high population and say "That's wrong". There are ways to explain such worlds, and indeed, a few such worlds is a Good Thing and make the universe a more interesting place. But you can say that having exactly as many crappy worlds with high populations as you have nice worlds with high populations is wrong. There should be SOME correlation between habitability and population. The reverse (having exactly as many nice worlds with low populations as you have crappy worlds with low populations) is less of a problem, because you can always say that this otherwise Terran-norm or Terran-prime world has something wrong with it that doesn't show up in the UWPs, or have some strong Imperal organization keeping squatters from moving onto a nice world that has been reserved for some purpose or other. Still, there's a limit to how many such worlds you can have. (Actually, in some ways a nice world with a low-medium population is more unlikely than an ostensibly nice world with no population at all).
Another statistical anomaly is the stellar classes. My guess is that they were assigned randomly according to the stellar distribution of the universe without regard to the UWP of the system. But the systems weren't random. Some of they were 'pre-selected' as being systems where worlds with breathable atmospheres existed. That ought to have skewed the distribution (in those systems) towards class F, G, and K stars. Breathable air implies a biosystem, and a biosystem implies an orbit with a life zone. Note that by the Book 6 rules, Class M stars (other than M0) don't HAVE orbits in the life zone. And when you allow worlds orbiting close enough to Class M stars to be in the life zone, as FIRST IN does, you get tidelocked worlds. As with other extreme situations, a few tidelocked worlds are fine, but not as many as the OTU has right now. IMO Earthlike worlds simply have greater potential for gaming fun. I won't say that if you've seen one tidelocked world you've seen them all, but I do think there's a definitely limits to how many a universe really needs.
A third anomaly are all those independent member worlds with low populations. Now, as a European, I can't argue that political entities too weak to defend themselves can't exist. Not with Monaco and and San Marino and Lichtenstein as examples to the contrary. But I still think that such places would be rare. Mostly, a low-population world would be an outpost of some more powerful entity, not an independent political entity. And those that were ostensibly independent would require a Big Brother to defend it (or a reason why the Imperium accepted it as a full member).
So I'd examine the statistically anomalous worlds and make a few judicious changes. Not too many, and making sure that each change makes for a more interesting universe. Worlds that have already been written up would be left alone on the principle that text is more authorative than mere UWPs. For instance, Alell and Fornice would be left with their low diameters and breathable air (in both cases explaind by "It's a mystery" (The best I've come up with even after hours of cugdeling my brain)) whereas Feri would have its size increased.
Another example is Rorise. It's a world with a population of 50 that the Imperium, according to
Fighting Ships donated a 50,000 T monitor to. Sure, they got it for free, but how can 50 people possibly afford just the
upkeep of a 50,000 T monitor? They couldn't even man it even if they all lived in it! When I tried to figure out an explanation, the best I could come up with was that the 50 people were billionaires (the Rorise family, owners of a trading company that was founded on Mora around Year 100; limited coverage, but very big in Mora and Trin) and that Rorise was their personal property (That saves the government code too

). The monitor is only partly manned by their security service, and was acquired after an attempt to kidnap all the Rorises at their annual family party. So the population figure of 50 is a fiction, the members of the family itself. But there's a whole bunch of servants and security guards who live there on a temporary basis, and the economic activity would be based on their number, so the proper population level is actually 3.
(Incidentally, if you base economic rules on population size, the figure you use should be the effective population size, regardless of their legal status; this IMO crocks up the 'transients are not included in the population figure' explanations).
Hans