Part of it is that SpaceX is very carefully managing reentry speeds and profiles. If you budget more fuel mass for de-orbit at slower speeds and better angles, you can get rid of the mass if Shuttle tiles or similar materials.
Thermal protection systems are dead weight except for a few minutes of maximum frictional heating, compression, and thermal radiation from plasma etc.
So being clever and getting rocket parts that are doing some semblance of "rocket stuff" during launch, that can also be pressed into service as thermal protection is a very good thing for mass fraction, payload, and the bottom line.
So if there's cryogenic fuel, pressurizing helium, or maybe even LOx that's going to be used for landing burns anyway, using it for regenerative cooling before it comes out of the engines is getting kind of a two-fer. The shiny stainless will also simply reflect a large portion of the thermal radiation from the compressed plasma shock radiation.
Interesting side note, both the Apollo Capsule and the Space Shuttle were designed wrong, for their intended angle of attack and center of gravity/stability on their chosen reentry paths. Because the ideal gas law that's good for most aircraft breaks down at hypersonic speeds, and at temperatures in excess of 500 degrees F, and really diverges with plasma at over 2000 deg. F at Mach 20+. Because the exact figures for the "real gas law" that's accurate under even extreme conditions, the computing power and correct math didn't exist during Apollo or the final design phase of the Shuttle, so they had to use approximations which turned out to be off more than predicted.
Think like the difference between Newtonian gravitation, and Einstein's space-time/relativity. Newton is good enough, very accurate even in long range circumstances like a transfer orbit from Earth to other planets in the Solar System. But it diverges in extreme conditions like large/steep gravity wells, high speeds at fractions of light, or where extreme precision is needed like the GPS satellites.
So I am guessing that SpaceX, starting from scratch, with zero 1960s-70s legacy hardware to constrain them, can get by with much closer to exactly what they need for thermal management on reentry, with a safety margin, than earlier craft.
As to the leaking and holes in the fuselage, there's precedent for that too. The SR71 had razor slits that leaked fuel. It was just for the thermal expansion of the titanium at Mach 3+ not active cooling, as far as I know, but it's a similar idea.
So far, SpaceX's iterative physical fast-failure development model, more akin to the "Space Race", if not actually more aggressive, definitely seems to be outpacing the old school big-Aerospace/NASA model of exhaustive paper and computer analysis, and multiple layers of bureaucracy to manage it.
