As someone who always felt mystified playing either the Catalan or Neo-Catalan, despite my successes with both openings, this game looks like a typical example of how GMs accumulate minute advantages in the most mysterious fashion and win.

Not so.

After 26.Rfd1 Botvinnik seems to have an overwhelming positional advantage to the eyes of this carbon-based life form. But if Stahlberg had found the right defense, White would have had only an insignificant advantage--basically, an even game.

That's the tricky part. Kind of wish computers would have some sort of add-on to their evaluation. Something like 0.00 but not an easy draw. Or 1.10 but not a trivial win.

After 26. Rfd1, white looks clearly better. Better rooks, better king, better N, and qside pawn majority. Plus white has the initiative. But it's equal material and black doesn't appear to have any structural weaknesses. So it's not surprising a computer could draw this.

But the computers technique might involve some sort of tactical trick or subtle positional move 15 moves down the line. Where missing that or any inaccuracy along the way might be just lost.

Guessing Stahlberg thought hard about the logical 26...Rfd8 and undertaking the difficult, unpleasant defensive task. Maybe even feeling it was possibly holding. But it's not unusual that a top player would rather go with active counter play when worse than passive defense when worse.