In message

, at 16:13:34 on Thu, 7 Nov 2013, Recliner
remarked:
1. No unresolved journeys. The way I would work this is the same way as
many other systems do it, such as Singapore - touching in charges the
maximum Oyster single fare to the card that could apply from that station
(subject to cap if appropriate for London), and touching out refunds back
the difference back to the journey you actually made. If you don't touch
out, you don't get it back, tough. That is powerful motivation, and far,
far less complicated.

But isn't that exactly what Oyster does?

No, because you can phone them up and argue about it.

The reason you can argue is if the system has gone wrong (eg, gates not
working, or train failed/delayed excessively), not because the basic
algorithm is wrong.

It seems to me that the "less complicated" solution being proposed would
not have the possibility to argue in those circumstances. You'd just
lose the money. That's an operational failing, not an algorithmic one.

2. OSIs (out of station interchanges) seem to be the biggest cause of
this. I've posted about ways these could be tidied up before - one way
is to always close the journey on touching out, but reopen it when
touching back in at an OSI location. Leaving journeys open was a silly
piece of design again asking for a need for intervention.

But isn't that exactly what Oyster does?

A slight variation on this... Isn't one of the known problems that when
you travel A-B complete your business rapidly and then travel B-A, when
B has OSI? In other words the initial exit doesn't complete the journey,
and when you re-enter the network and go back where you came from it gets confused.

I think it 'provisionally' completes the journey, but reopens it if the
station is re-entered through another exit within a specified time.

That's right, but if you end up back at A (or a station C, near A) it is
then likely to penalise you because your journey A-C apparently took
"too long", and the current algorithm wishes to penalise slow-coaches,
presumably because they see it as evidence of some form of fare-dodging.

A better algorithm (but it requires more hardware too, and makes the
system more complex to navigate)) is some sort of validator at B which
allows the traveller to say "please force a completion of journey A-B".
Only then would the person be charged [a pair of] correct fares (which
also work within the cap), rather than a penalty fare (which I believe
are outwith the capping regime).
--
Roland Perry