Actually engines are famously bad at endgames. For example, Houdini would probably consider this a win for white (I need someone who actually has an engine to confirm this):
On the other hand, if an endgame is drawable, chances are that a top-level pro player knows how to draw it. Professional players have pretty much all useful openings and endgames memorized, so letting them have access to texts probably won't help them much.
Top professionals believe that Houdini 3 is “unbeatable”. What professionals are overlooking is one simple edict, “He who writes the rules wins.”. Once you realize this the top chess programs can be beaten. For the past 50 years we have written rules favorable to computers. Isn't it time we started writing rules favorable to humans? It is critical, in my opinion, that chess professionals start beating chess programs again. We are losing our humanity byte by byte to computers; if, with simple, fair adjustments of the rules we could defeat computers wouldn't that be worth the effort?
What is critical is that any rules changes seem “reasonable” to the public. If we are to turn chess into the most popular game in the world it is critical that the rules changes seem reasonable to the average person.
Rule change #1---any professional may consult any opening or endgame text; computers have in their memory millions of games and millions of lines of opening theory, and have analyzed simple endgames to completion; this will seem reasonable to the public,
Rule change #2---consultation games will be encouraged so that we can step inside the brains of the top professionals as they discuss strategy and tactics. It is highly unlikely that two pros discussing the position together will miss simple tactics. Verbose, assertive chess athletes will be preferred---this should be highly entertaining. Spectators on their laptops can improve on their strategy and tactics sort of like being able to return a tennis shot by Serena Williams on your laptop!,
Rule change # 3---go back to 120'40 followed by adjournment, the “old” rules, a suggestion of GM Lev Alburt; this will eliminate fatigue errors,
Rule change # 4---eliminate time delay and implement the one second rule. The one second rule holds that computers must take at least one second on every move---or face a severe time penalty. This will permit human professionals to run computers out of time in simple endings. This requirement is reasonable because computers can never make a touch move violation,
Rule change # 5---if all else fails, then limit the size of the computer to the volume of the average human brain that is actively involved in analyzing chess positions. In other words the parts of the brain controlling such things as sight, sound, taste, smell and touch or our hearts beating or lungs breathing are not included.
The most controversial rules change is the one second rule, but think about it. Why should humans be forced to calculate the move, pick up, move or capture a piece or pawn and punch the clock when all a computer has to do is calculate? It will require enormous programming advances for a computer to calculate a move, pick up and move a piece or pawn without touching another piece or pawn and punch the clock all in under one second.
Without the one second rule, human consultation teams with the above rules changes will achieve at least an ELO 3200 rating against computers and a 3300-3500 rating against computers if the one second rule is implemented.
A simple test of my proposal is to have consultation teams in the privacy of their own homes try out the rules changes. If they are successful they can issue a challenge to Houdart, Houdini 3 Pro's father, in a public contest with appropriate stakes---there should be no difficulty finding sponsors. If the humans win it will make news world-wide.
Think of what it would mean if two professionals could analyze together prior to a contest and consult opening books during the contest (World Champion Kasparov lost to Deep Blue because he stepped into an opening trap).
In "simple" drawn endings like Bishop and Rook versus Rook a computer will always draw if its King can get to the "right" corner, a pro will often lose this ending. With access to an endgame book, wouldn't they be far more likely to draw? In a slightly favorable ending won't two pros working together have better winning chances?
If we could reverse the sense of inevitability i.e. once a computer achieves intellectual supremacy in a discipline it will be permanent champion in that discipline, this would permit us to reverse this sense of inevitability and permit us to regain, in a small way, our humanity.
With the right rules changes, the right consultation teams and wins by top professionals, chess has the potential to become the greatest spectator sport of all times and all that implies for the financial futures of even mid-level professionals and teachers!