It works, that's why. Others have explained it better than I can. But at higher dilution, Perceptol changes its personality, acts more slowly, and allows more grain growth without need of a separate special developer. And it does so in a semi-compensating manner. The proof is in the pudding. I've got a set of 6X7 TMX negs I'm going to print this afternoon based on it. Rodinal etc would do something quite different, and not to my liking. The whole point is, Perceptol diluted 1:1 would render TOO FINE a grain in TMX;
plenty of detail, yes, but optimal edge acutance, no. But at 1:3, it behaves quite differently. The effect with Delta 100 is somewhat analogous; but like I already stated, I prefer D100 souped in PMK pyro instead.
I shoot Delta 100 at 50 simply to boost the shadow values further up onto the straight line, regardless of developer type, in order to better simulate what TMX100 does at full box speed due to its shorter toe and longer straight line. In general, staining pyro developers render better highlight control than ordinary developers. Therefore, nowadays I only use Perceptol for TMax 100, and only at 1:3. For all other pictorial films, including TMY400, I prefer pyro. There was a time when I tried numerous developers for all kinds of films, including FP4, but struggled with highlights back in those graded paper days, and often had to resort to masking if I wanted a full range of crisp tonal separation.
But I do keep a number of other developers on hand, or the chemicals to mix them, for various lab purposes or special applications.
The results of Perceptol can allegedly be replicated by higher dilutions of Microdol (like Nicholas noted). But Perceptol is a popular developer which can still be picked up at the local camera shop if needed. Lots of locals have defaulted to it in preference to D76.