Hoogovens 33rd / Wijk aan Zee (1971) — MillBase vs Tata Steel

I found a way to more efficiently download the games from the Tata Steel site (using Firebug). Now, with this as a reference data set I can directly compare against MillBase. See my previous post for a few more details, here’s the data:

MB vs Tata

Herein, we allow the possibility that Tata's site could be wrong as well.

(Changed one abbrev - RS = baskchess.rs)


1. Andersson--Mecking (R10) 1-0   Tpos 4-5:  4.Nc3 vs 4.Be2

     3.Nf3 d6 4.Nc3 Nf6 5.Be2 O-O 6.O-O Bg4 7.Be3 e5 8.Qd2 Nc6 9.dxe5 dxe5
   > 3.Nf3 d6 4.Be2 Nf6 5.Nc3 O-O 6.O-O Bg4 7.Be3 e5 8.Qd2 Nc6 9.dxe5 dxe5
                |

     Right: Tata, BR, RS, CT, rhp, 365chess
     Wrong: MB, CG

     (Confirmed by 25-01-1971 -- "De Tijd : dagblad voor Nederland" game score)
     
     http://kranten.delpher.nl/nl/view/index?query=schaken+wijk+aan+zee&coll=ddd&image=ddd%3A110612309%3Ampeg21%3Aa0189&page=1&maxperpage=50&sortfield=date&cql[]=%28date+_gte_+01-01-1971%29&cql[]=%28date+_lte_+31-01-1971%29&cql[]=%28content+all+schaken+and+wijk+and+aan+and+zee%29#image
    

2. Donner--Langeweg (R4) 0-1    Last Move: 44.Qxc4 vs 4.Qxc3+

   > 42.Kc1 Qf1+ 43.Kb2 Qd3 44.Qc1 Qxc4
     42.Kc1 Qf1+ 43.Kb2 Qd3 44.Qc1 Qxc3+
                                   |

     Right: MB, CG, CT, BR, RS, 365chess
     Wrong: Tata

     (No question here)
     

3. Huebner--Petrosian (R1) 1/2-1/2      Move: 46...Nh7 vs 46...Nh5

     45.Qd3 Kg7 46.Ke1 Nh5 47.Rd2 Qa1+
   > 45.Qd3 Kg7 46.Ke1 Nh7 47.Rd2 Qa1+
                       |

     Right: Tata, CG, CT (dup), BR
     Wrong: MB, CT, RS, rhp, 365chess

     (Engine prefers 46...Nh7 by ~0.4, but would like to play 46...Nd8)


4. Ivkov--Najdorf (R2) 1-0     Move: 17...Qe7 vs 17...Qe6

     16.Qxd5 Rad8 17.Qg5 Qe7 18.Rd2 h6 19.Qh5 Na5 20.Rad1 Rfe8
   > 16.Qxd5 Rad8 17.Qg5 Qe6 18.Rd2 h6 19.Qh5 Na5 20.Rad1 Rfe8
                         |

     Right: Tata, BR, CT (dup)
     Wrong: MB, CG, RS, CT, 365chess, rhp

     (Engine prefers 17...Qe6)


   
5. Korchnoi--Huebner (R15) 1-0   Move: 25...Re8 vs 25...Rf8

     24.Bxf1 Qf6 25.e4 Re8 26.Qd2 Qf3 27.Bd3 Bf6 28.Ne1 Qh5 29.Be2 Qg6
   > 24.Bxf1 Qf6 25.e4 Rf8 26.Qd2 Qf3 27.Bd3 Bf6 28.Ne1 Qh5 29.Be2 Qg6
                        |

     Right: Tata, BR, CT (dup)
     Wrong: MB, CG, RS, CT, 365chess, rhp

     (Engine prefers, and positionally seems clear, 25...Rf8)


     
6. Korchnoi--Olafsson (R6) 1-0  Move

   > 11.Nc2 Nc5 12.Be3 Be6 13.b4 Rfd8 14.Be2 Qc7 15.bxc5 dxc5
     11.Nc2 Nc5 12.Be2 Be6 13.b4 Rfd8 14.Be3 Qc7 15.bxc5 dxc5
                   |

     Right: BR (12.Be3,19.Nxb5,23.Nba3), CT (dup)
     Wrong: Tata (12.Be2)
     Wrong: MB, CG, RS, 365chess, rhp (12.Be3,19.Bxb5,23.Na3)

     (Engine prefers 12.Be3 - all games but BR have big oversights)
   

   
7. Kuijpers--Ivkov (R7) 0-1    Move 71.Ke2 vs 71.Kf2

     69.Kf2 Rh2+ 70.Kf3 Rh3+ 71.Ke2 Rg3 72.Rh8 Kg4 73.Rh7 g6 74.Rh6 Nxf4+
   > 69.Kf2 Rh2+ 70.Kf3 Rh3+ 71.Kf2 Rg3 72.Rh8 Kg4 73.Rh7 g6 74.Rh6 Nxf4
                                |
                                
     Right: Tata, BR, CT (dup)
     Wrong: MB, CG, RS, 365chess
    
     (Engine prefers 71.Kf2 by a pawn)                            



8. Langeweg--Andersson (R13) 1/2-1/2    Move: 41.Ke1 vs 41.Kf1

     40.Rxd5 Rxc2 41.Ke1 Nc5 42.Rxb6 Nxa4 43.Ra6 Rc1+ 44.Kf2 Re7
   > 40.Rxd5 Rxc2 41.Kf1 Nc5 42.Rxb6 Nxa4 43.Ra6 Rc1+ 44.Kf2 Re7
                     |
                     
     Right: Tata, BR, CT (dup)
     Wrong: MB, CG, RS, chess-db, CT, 365chess, rhp

     (Engine gives complete degeneracy here)



9. Langeweg -- Ree (R7)     Move: 15...Ndf6 vs 15...Ngf6

   > 13.Bc1 Ng4 14.Nge4 Bh6 15.Bh3 Ndf6 16.Nxf6+
     13.Bc1 Ng4 14.Nge4 Bh6 15.Bh3 Ngf6 16.Nxf6+
                                   |

     Right: MB, CG, RS, CT, 365chess
     Wrong: Tata, BR
                                   
     (Engine prefers 15...Ndf6, but this is quite speculative)


10. Mecking--Donner (R13) 1-0  Move: 15.Rfc1 vs 15.Rac1   

   > 13.Bb2 Nc6 14.Qa4 Bb7 15.Rfc1 Bd6 16.Qh4 Ne5 17.Nxe5
     13.Bb2 Nc6 14.Qa4 Bb7 15.Rac1 Bd6 16.Qh4 Ne5 17.Nxe5
                              |

     (This game's ending could be analyzed further)

     Right: Tata, CG, CT (dup), BR, chess.com
     Wrong: MB, CT, RS, sachista, 365chess, rhp
     

Now, I will admit deciding what was “Right”/”Wrong”  would better be decided by cross-checking with the tournament book. But I don’t have one (and really don’t want to spent too much more time mucking with this tournament). Let’s just say “we” left it to our non-expert panel of one – with help from an engine, and a little chess common-sense.

What is chess sense? Well, take a look at <Donner–Langeweg (R4)>. The Tata Steel site gives 44…Qxc3+, which I hope we can all agree is wrong.

Another game worth examining is <Korchnoi–Olafsson (R6)> which is a nice study, even if finding the right score is difficult. Perhaps <BrasilBase> has the right score, maybe because a dedicated Mecking fan got the tournament book, and decided to benefit us all by transcribing it into PGN for public release.

I’ve talked about epsilon-squared reduction of data entry in <CG> posts – where a small, unavoidable, error rate (after all – we’re all humans, unless we’re a web-bot in these parts) – can be mitigated against by comparing two independent data sets. Perhaps, <BrasilBase> and <Tata Steel> are two such data sets.

Regardless, let’s allow my comparison and determination of correctness, for the sake of argument. Then <Tata Steel> has 3 games wrong (two of which are more certain). That amounts to an error rate of 2.5%, which I think is about right for a careful transcriber. (Remember – the error really should be measured per move, and not per game – but we’re averaging).

The epsilon-squared argument allows this error rate to be reduced to ~ 0.06%, if we are willing to duplicate efforts, and if they can be assumed independent.  

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s