Endgame tablebase and chess curiosities

Interesting thing happened today when I solved some studies with one chess player from Lund.
We tried to solve pretty complicated study that you can see on the diagram, and although we seen abundance of ideas in very limited time, we failed to solve it correct.
Well, nothing special about that, til the moment when we looked at the solution.
Right in the beginning something was fishy there.
As in the final position (which you can find if you decide to read this article) there were only six pieces on the board, we could consult very useful tool.
An endgame tablebase!

I am not going to explain the whole solution, which is quite complicate, and I will give only the main line with some nice points.
The task is "White moves and draws".

1. Nf2+, Kh4!
2. a7, Bxa7
3. Bxd6, Rd5
4. b8Q, Bxb8
5. Bxb8, Rd2+
6. Ka1

We have following position.

Now, if Black takes on f2:

6...Rxf2
7. Bg3+, Kxg3 stalemate

So Black needs to play:

6...Bd5

Which almost traps the knight on f2.
White has cunning move here:

7. Bg3+

If Black takes on g3, then Ne4+ forks king and rook and leads again to a stalemate.

7...Kh5

The solution is over, and according to it, the position is draw.
Well, if you look at online tablebase and set up this position, you can see that White is dead lost, and in the best case for him, he will be mated in 30 moves.
Here is the link: Online Endgame Tablebase

This proves that the study is incorrect.
There is no way to save a day.
However this motivate us to look further at some positions which can be interesting.
For example:

The exact position is not so important. The main thing is that neither side can change material proportion in couple of moves.
So the tablebase teach us that White is winning after 148 moves.
You can go through the solution, and see that it is a way behind human understanding.
The position with this material proportion had Shirov against Karjakin in 2007 in The World Cup.
It finished in a draw.
What I am going to do is to play a little with other combinations of a minor pieces pair.

So I exchanged one of black knights for a light squared bishop.
The final verdict of did not changed.
White is still winning, and this time even quicker.
With the best possible play White is going to mate Black after 71 moves.
I am not so familiar with the rules, if you are allowed to play more than 50 moves in this exceptional positions, but that is not subject of this article.
So, bishop+knight is the better combination then knight+knight?

Not really!
This position is just a draw.
I just changed the colour of Black bishop.
So bishop+knight gives a draw (and two knights loses) in a case of same colour bishops on the board.
In the case of opposite colour bishops White is winning.
This is not strange, and this is not opposite to the rules.
With an opposite colour bishops you can obtain the complete domination on one colour which gives you wonderful (in this case winning) attacking chances.
While good to know, I do not thing that these examples has a real practical value.