i really do not understand what you guys are trying to tell me
You posted anecdotal evidence, ie very particular situations where the knight is better.
There are other positions where the bishop is better, and it happens that on the average, the master's evaluation is that the bishop is better on the long run. However, it highly depends on the position, so you will have something like 40% of the positions where the knight is better and 60% where the bishop is better. (the numbers are made up for the sake of the argument)
I don't get the point of the post in the first place, White was lost even without the minor pieces on the board in your opening post. You put this in the endgame forum, so I assume we are talking about the relative strenght of minor pieces in the endgame, but a single bishop or a single knight cannot produce checkmate alone. Knight and pawn versues bishop has drawing tendencies and if the diagnole is long enough the position of the weaker sides king might not matter. Bishop and pawn versues Knight also has strong drawing tendincies because the king can set up a blockade on a square that the bishop cannot touch and the presence of the knight means he cannot be forced away. Zugzwang is possible of the knight is cut off from the king. You never brought any of this up, you produced two positions with entombed bishops and pointed out that the knight was superior to an entombed bishop. I am sure everyone already knew that, so this thread had very little content when you started it.
This month's column by Karsten Muller on chesscafe.com is about "Andersson endgames".
"Endgames with rook and knight against rook and bishop, where the knight has the advantage are sometimes called Andersson endgames because of the many famous victories of the Swedish grandmaster with this material configuration."
The analysis is way too much for me to play through, but it's definitely related to this thread.
While not dispositive consider this game I played in the Wilkes-Barre against Houdini 3. You will note that my Knights are on the first rank but vital to my advantage of +-
What I'd like to do now is provide a more sophisticated example of development based on a game I played against Houdini 3. Moody-Houdini 3, Match 2013, Game 1: 1.e4 e5 2.Nf3 Nc6 3.Bc4 Nf6 4.Ng5 Bc5 5.Bxf7+ (5.Nxf7 is also good, but I was able to win a nice game on the Black side of the 5.d4 variation: 5. d4 d5 6. Bxd5 Nxd4 7. Nxf7? Qe7 8.Nxh8 Bg4 9. Qd3 (Bf7ch Kf8 10.Qd3 Be2 11.Qh3 Bg4 12.Qd3 Be2 =) Nxd5 10. c3 Nb4 11. cxb4 Bxb4+ 12. Nc3 Qd7 13. Kf1 Bxc3 14. f3 Bxf3 15. gxf3 Qh3+ 16. Kf2 Bb4 17. a4 Rd8 18. Bg5 Bc5 19. Qc3 Nb3+ 20. Qxc5 Nxc5 21. Bxd8 Ne6 22. Rag1 Nxd8 23. Rxg7 Ne6 24. Rg3 Qh6 25. Rg8+ Ke7 26. Rhg1 Qxh2+ 27. R1g2 Qh6 28. a5 Qd2+ 29. Kg3 Qxa5 30. Kh2 Qc5 31. R8g4 Nf4 32. Rg7+ Ke6 33. Rd2 Qe3 34. Rdd7 Qf2+ 35. Kh1 Qxf3+ 36. Kh2 Qxe4 37. Rde7+ Kd5 * 0-1) Ke7 6.Bc4!? (While technically this is not a Novelty, it is practically unknown; White anticipates an exchange sacrifice on f3 so the retreat Be2 dulls the effectiveness of that sacrifice):
6...Na5?! (Houdini tries to con me into winning the exchange: 7.Nf7? Qe8 9.Nxh8 Nxc4 =/+) 7.Bd3! (White demonstrates that the Knight on a5 is misplaced; White holds the e4 pawn in anticipation of h6/Nf3/Nxe4) d6 8.Nc3 h6 9.Nf3 Be6 10.Be2! Nc6 (There is nothing better. From the above examples the reader will readily assign a value of -1 to this move, but the move sequence Bd3/Be2 is worth 0 because White can reach d3 or e2 in one move.
11.d3 +/=a7 12.Na4! (Not to attack the Bishop so much; the main reason is to free the c-pawn) 12...Ba7 13.O-O Kf7 14.h3 Rf8 15.b3! (To free b2 for the Knight) 15...Kg8 (Look how many tempos Black lost by castling by hand) 16.Be3! Bd4?! 17.c3! (Houdini doesn't like this move) Bxe3 18.fxe3 b5 19.Nb2 d5! (Houdini “knows” it has to attack, yet this straightens out White's pawns) 20.exd5 Nxd5 21.Qc1 (Holding c3 and e3 and not blocking the d-file) Qd6 22.Nd1 Rab8 23.c4 bxc4 24.bxc4 Nf6 25.Nf2 Qc5 26.Rd1 (Preparing d4) Qa5 27.Rd2 Bf5 28.d4! e4 29.Nh2! Kh8 30.Nf1! Qb4 31.Nd1! Qd6 32.c5 Qe7 33.Rb2 Rxb2 34.Qxb2 Nd5? (This is a programming error; Houdini should force me to win the a-pawn, not just give it away for nothing.) 35.Bxa6 +- (The threat is Bishop retreat, a4/a5/a6, but this is over the computer's horizon so it thinks that other moves give White a lesser advantage; in a game between GM's the phrase you will hear is that, “White's pressure nets him a pawn.”)
Now that White has a clear edge, the Knights' will go on the offensive. Unfortunately I made a series of fatigue errors and lost, but this does not alter the fact that White gained ground the entire middlegame primarily because of his Knight maneuvers.
While not dispositive consider this game I played in the Wilkes-Barre against Houdini 3.
As Orwell put it, that kind of structure become less attractive when you think of the sentence : a not unblack dog was chasing a not unsmall rabbit across a not ungreen field.
As for the game, it does not deserve comments, much less giving a diagram.
I sometimes like knights better because they can fork easily.
+1 for jonnin's explanation.