I read that he said ‘It’s not magic’ rather than it didn’t work, as in there the logical reasons behind it that make sense; DCS is outputting a smaller amount of pixels, so can do it faster. The counterintuitive bit is that SteamVR super sampling making it look just as good I guess.
I haven’t watched it either. 
I put a table here: VR Optimization Guide - DCS 2.5.6.60 - #33 by fearlessfrog
| SteamVR SS % | SteamVR Resolution | DCS PD | DCS Resolution | Notes |
|---|---|---|---|---|
| 150% | 2696x2640 | 1.0 | 2696x2639 | My usual |
| 150% | 2696x2630 | 0.5 | 1348x1319 | Looks like lego |
| 400% | 4404x4316 | 0.5 | 2202x2158 | Looks pretty good |
| 500% | 4924x4824 | 0.5 | 2462x2412 | As good as My usual better fps |
The thing of note is really the DCS Resolution column. My ‘usual’ (first row) SteamVR at 150% (I know it’ll be different for you on a G2, but same principle) with PD 1.0 works just as nicely as the 500% SuperSample (SS) PD 0.5. I think Jabbers is saying ‘That’s not magic’ because the underlying DCS Resolution is a bit lower, which is why it is smoother/faster. You can see that 0.5 x 4924 = 2462, which is a smaller number than 1.0 x 2696 = 2696, so it all makes sense. DCS does less stuff = better framerate. That’s the myth bit he was debunking, albeit after I had already posted this table. 
The answer is, by using the info above, because DCS is outputting less pixels, but the super sampling is making them look just as nice as before (but with slightly better performance).