The obfuscation mechanism isn't reliable when there are two or more trains running with the original reporting number.
The train that appears as 132K (let's call it 1Q00) was manually called at 1204 and so any description in the TD feed of 1Q00 would be replaced with '132K'. This is all good until 1837, then another manual train call was made for another train running under 1Q00 elsewhere in the country. This set up a separate obfuscation mapping of 143K, so any 1Q00 in the TD feed is now replaced with 143K.
This non-determinism is fine, because obfuscation isn't a simple case of "Replace the reporting number with something else" - it's supposed to make it difficult to determine all details of the service. Knowing there is *a* train moving between signals is better than not knowing there's a train there - it paints a bad picture if you see successive signals turn to danger with no TD message.
Anyway, the best solution for this is for the industry not to treat train reporting numbers as a pseudo-secret.
Peter