------Fritz 9--------
TOPALOV 1111111 FULL MATCH (35.Kxa2), Fritz 9 delta = 7.73
TOPALOV 1111111 FULL MATCH (30.cxb3, forced!), Fritz 9 delta = 6.97
TOPALOV 1111111 FULL MATCH (34.Qg8+), Fritz 9 delta = 6.24
TOPALOV 1111111 FULL MATCH (27.Qh6), Fritz 9 delta = 3.07
TOPALOV 1111111 FULL MATCH (28.Qxh7), Fritz 9 delta = 3.01

TOPALOV 1111111 FULL MATCH (26.Bxc5), Fritz 9 delta = 1.98
TOPALOV 1111111 FULL MATCH (33.Bxf7), Fritz 9 delta = 1.65

TOPALOV 1111111 FULL MATCH (20.g6), Fritz 9 delta = 0.42
TOPALOV 1111101 MATCH (29.Nd3), Fritz 9 delta = 0.41
TOPALOV 1111111 FULL MATCH (25.Bf5), Fritz 9 delta = 0.40
TOPALOV 1111111 FULL MATCH (19.h5), Fritz 9 delta = 0.17

TOPALOV 1111000 PART MATCH (24.Rhg1), Fritz 9 delta = +0.04

TOPALOV 1111111 FULL MATCH (22.Be6+), Fritz 9 delta = 0.03
TOPALOV 1111111 FULL MATCH (21.gxf7+), Fritz 9 delta2 = -0.05
TOPALOV 1111111 FULL MATCH (31.Rxd3), Fritz 9 delta2 = -0.11

TOPALOV 1000000 PART MATCH (32.Be6), Fritz 9 delta2 = -0.88
TOPALOV 0000000 NON-MATCH (23.b3), Fritz 9 delta2 < -2.27

------Fritz 10-------
TOPALOV 1111111 FULL MATCH (35. Kxa2), Fritz 10 delta = 8.33
TOPALOV 1111111 FULL MATCH (34. Qg8+), Fritz 10 delta = 6.08
TOPALOV 1111111 FULL MATCH (30. cxb3), Fritz 10 delta = 6.03
TOPALOV 1111111 FULL MATCH (27. Qh6). Fritz 10 delta = 3.17
TOPALOV 1111111 FULL MATCH (28. Qxh7), Fritz 10 delta = 2.96

TOPALOV 1111111 FULL MATCH (33. Bxf7), Fritz 10 delta = 1.55
TOPALOV 1111111 FULL MATCH (26. Bxc5), Fritz 10 delta = 1.40

TOPALOV 1111111 FULL MATCH (25. Bf5), Fritz 10 delta = 0.56
TOPALOV 1111111 FULL MATCH (29. Nd3), Fritz 10 delta = 0.37
TOPALOV 1111111 FULL MATCH (20. g6), Fritz 10 delta = 0.35
TOPALOV 1111111 FULL MATCH (22. Be6+), Fritz 10 delta = 0.31
TOPALOV 1111111 FULL MATCH (19. h5), Fritz 10 delta = 0.30
TOPALOV 0011111 MATCH      (24. Rhg1), Fritz 10 delta = 0.20
TOPALOV 11111ii FULL MATCH (21. gxf7+), Fritz 10 delta = 0.06
TOPALOV 1111111 FULL MATCH (31. Rxd3), Fritz 10 delta = 0.05

TOPALOV 1000000 PART MATCH (32. Be6), Fritz 10 delta2 = -0.47
TOPALOV 0000000 NON-MATCH  (23. b3), Fritz 10 delta2 = -2.45
---------------------

Besides the two non-matches, the /significant group/ is moves 
19--25 (ex. 23), 29, and 31.  In Fritz 9 this group can be argued
as pretty random, as two of the "full matches" even went negative at
18 ply /which is reachable for those moves with 2-or-4-core technology/.
Whereas with Fritz 10 they are /all/ matches /with high info gain/
(ignoring 11,12 ply for 24.Rg1).  Albeit only for 8 moves, this makes
a notable difference.  Likewise classifying 29.Nd3 and 31.Rxd3 as 
full matches in Fritz 10 changes the perception of moves 26--35.

() My long formal run on Fritz 9 is consistent with my early word
   from informal, multi-engine running on the day the allegations
   appeared, in comments to the Jan 27 item at www.chessninja.com.

() Fritz 10 is notably different from Fritz 9 for the classification
   of matches in the critical range within half-a-pawn, even though
   the evals on the whole look similar.  

The good news: these matches were notably different from many other
engines, thus promoting the idea that cheating with a particular engine
can be told apart from "cheating with engines in general".  (The math
will make the latter potentially detectable too, however.)

The bad news: we really need to assess the robustness of testing---and
these tests are time-intensive even for a single run!

---KWR, 2/14/07.