bramling
Veteran Member
Heh, this is a problem that affects so many companies I've worked for and consulted for. I know someone who's literally written a book on this phenomenon.
KPIs (Key Performance Indicators) drive behaviour. They are generally designed with a particular behaviour in mind, but the Law Of Unintended Consequences kicks in rather rapidly (particularly if they're very blunt measures, badly thought through, or dogmatically applied).
So, you have a desired behaviour: minimise delays. The measure? How much your trains are delayed by. Makes sense. People working for the railway have delay minutes in mind.
But then you're looking at the trains and how their performance can be optimised. Skip stops to make up time? OK. Ignore a delayed connecting service to avoid a delay (paths pending). OK.
What would the alternative be? What would happen if you measured passenger delay minutes instead of train delay minutes. Hardly beyond the capability of railway IT to have a reasonable estimate of passengers flows, and model the total delay minutes by taking one course of action over another. More passenger friendly? Sure.
But that has unintended consequences, too. What are the knock-on delays down the line? On a congested network, what are you snarling up? Where to driver hours come into all of this?
No simple answer, really. I've no doubt NR have some data scientists looking into this, who know far more about the topic than I do. But that's what KPIs do: drive unintended behaviour. The art is balancing them, and reacting appropriately to the negative consequences.
There are still staff left in the industry who are prepared to stick their finger up at target nonsense and simply do what's right for the service. However they do find themselves on the wrong side of management for their troubles!
London Underground is another classic example - they do something called the "snapshot", which is the number of trains running on a line at a given time (normally on the hour) versus the number which are timetabled to be running. The principle is sound enough, however it didn't take long before certain managers started exploiting this to their benefit - for example arranging coverage such that any cancellations affect services which don't hit a snapshot time, bringing a train out a couple of minutes early so it comes in at XX59 instead of XX01, or in extreme cases bringing a train out of a depot to sit in a platform for a couple of minutes at shapshot time. None of this benefits the passenger, as what matters is trains being where they're timetabled to be, not just that they happen to be on the railway. Not everyone is prepared to play the game though. In a different world it might be considered as fraud, given that money changes hands as a result of these performance targets.
In the case of GTR though I'd say the issue is more that they're trying to run a service which is inherently flawed, and haven't put in enough mitigations to get it work tolerably well.