The Myth of On-Time Performance

I received a comment from Karem Allen in Durham that belongs in its own thread: 

A friend asked me if I knew why there would be an empty bus following closely to a full bus and my anwer was — so the empty one would be able to jump ahead and pick up riders.

He told me at one time they could leapfrog and be able to help the other drive but are now strangled in policy.
He told me that if a Driver gets 2 early’s in a month he is suspended.  So instead of jumping ahead and taking the riders and let the full one continue the empty one will hang back so as to not be early and of course the stop is empty of people.

Is this still in force?

I did not think buses were on a schedule to be early anyways.

There are a few things going on here worth talking about.

First of all, there is nothing wrong with buses playing leap-frog to handle passengers when they are bunched.  This can even out the load and the buses actually make better time going down the street.  Sometimes, however, the following driver will let the poor sod in the first bus take all the load.  Not fair, but it happens.

Having said that, the TTC does have a fetish for on time performance that can have bizarre results from the customers’ point of view.  This is driven by a measure, reported monthly to the Commission, that was introduced by former CGM David Gunn:  what proportion of all trips operated within 3 minutes of their scheduled times.  This sounds laudable, but like many corporate targets, it skews the very process it is intended to measure.

On routes with frequent service, customers don’t care about the schedule, they care about service reliability.  The schedule is important from the drivers’ point of view because they like to get off work on time, but that doesn’t create a requirement to be +/- 3 minutes over their entire shift.  However, if the staff are going to be measured on keeping buses on time, then that’s what they do.  Vehicles will be short-turned.  Drivers will dawdle along a route that has too much running time for the prevailing conditions.

Once upon a time, Inspectors (now called Route Supervisors) would, among other things, space out service by holding vehicles that had bunched up, or where they knew a gap was coming down the line.  Unfortunately, the information displays the drivers have from the central monitoring system tell them only how they stand relative to the schedule, not relative to vehicles ahead and behind them.  Indeed, if the goal is to stay within 3 minutes of schedule, holding a car to space service may actually violate the goal even though it would provide better service.

The situation Karem describes — the penalty for running ahead of schedule — has been perverted into an operating practice that works against both the schedule and the provision of good service.

For several months, I have been working on an analysis of the operation of major routes using data from the TTC’s vehicle monitoring system, CIS.  There’s a lot to dig through, and developing the programs to automate a lot of the analysis took up a lot of my spare time.   I don’t want to prejudge this effort, but two major points are starting to emerge:

  • Service is not managed well.  Yes there are short turns and attempts to fill gaps, but there is an overarching problem that vehicles don’t run on time even on days where traffic is light.  Christmas Day was particularly illuminating in this regard.  On a route with frequent service (say every 8 minutes on a holiday), a variation of +/- 3 minutes makes a huge difference in the gap riders see between cars, and makes the service feel unpredictable even when it should run like clockwork.  Achieving the ontime goal is a snap, but the quality of the service can be appallingly bad.
  • Congestion does exist and certainly has its effects.  However, it is important to distinguish between congestion that can reasonably be anticipated (it happens in the same place at the same time most if not every day) and random events caused by accidents, local construction projects, etc.  The TTC uses congestion as a catch-phrase to describe all of its problems, and I am not convinced that it is the only problem.

If any TTC operators who read this feel like commenting anonymously, I will be happy to post your feedback here.  In particular I would like to know how the desire for on-time performance is actually being implemented on the street and if the practices, including disciplinary ones, are counter-productive.

Sunday, March 4, 2:26 pm.  Here is a feedback I received.

As a T.T.C. operator for 18 years, I can tell you that there is a big bone of contention from the drivers about C.I.S. and its accuracy.  Besides the use of the trump unit, we are expected to know our timing points as well.  However in many cases the timing points do not mesh with what is being displayed on the trump unit.  You can be on time according to the timing points but the trump will often say that you are early.  Conversely, sometimes the trump will say you are on time but you know according to the timing points that you are early.

There are many spots along the various routes where you will be driving along on time and suddenly the trump will jump 2 to 3 minutes.  Before you know it, you receive a call wondering why you are early.  When you point out the fact that the unit jumped, you are told not to go by it as you should know your timing points!  Yet they will beep you and tell you to ease back according to what they see on the screen even if you know that you are on time according to the timing points.

This leads to many disagreements between the drivers and the supervisors monitoring the line through the C.I.S.  If we are more than 3 or 4 minutes early we get told to ease back as we may be creating a gap behind us.  However if we are more than 3 or 4 minutes or more behind schedule we rarely hear from the supervisors.

In many cases when we are running very late, we have to call C.I.S. and ask if we are going to be turned in order to get back on schedule.

You are right to point out that the drivers don’t seem to pass each other as much any more (unless it’s a line with very frequent service or the lead bus is extremely late).  Years ago, on a line with a 5 minute headway for example, if you passed a bus that was a 2 or 3 down and as a result became early yourself, the supervisor allowed it and asked the next bus behind to move it up a bit to fill the gap.  Now with the rules the way they are, there have been many times when I am 3 minutes early and will not pass the bus in front of me that is late for fear of being “documented”.

Usually it is 3 times of being documented and it is in to see the divisional superintendant and it goes on your record.  I know it is frustrating to the public that we often drive slow or kill a few lights but that is what is expected of us.

15 thoughts on “The Myth of On-Time Performance

  1. I’m really looking forward to your analysis and insights into CIS… and I applaud TTC’s Surface Ops DGM for releasing it to you without a FOI request.

    I’m shocked to know that a driver gets a suspension for 2 earlies in a month—now I finally understand ATU President Bob Kinnear’s and driver’s anger about ATU members getting 3-day suspensions, going home without pay for “petty” reasons (which I never understood before).

    What a way to demotivate and demoralize the TTC’s front line “retail” employees! With respect to “leap-frogging” what ever happened to David Gunn’s oft-repeated motto: “It’s easier to ask forgiveness than permission?” to encourage employee initiative and creativity? I guess it’s just for TTC Staff…. not Union members!

    Going up and down Bathurst 7 to/from Bloor/Steeles I’ve seen 5 northbound buses bunched together south of Eglinton on a merely rainy day, and CIS up to -33 minutes behind due to accidents at Bathurst & Finch.

    CIS calls… another 7 short turn is announced by the driver who apologizes profusely to passengers, and then tells me s/he doesn’t understand how CIS Supervisors make their decisions, “…they make no sense” to the driver… “I just do what they tell me!” How to produce an automaton 101!

    During the Thursday snowstorm the Bathurst 7 driver said it took 1 1/2 hours to get from Bathurst Station to Dupont… and 4 hours 48 to get to Sheppard (without a bathroom break) where he picked me up around 8:10 pm—12 hours into his driving day!

    He was going home at the end of his regular shift when CIS called him and told him to take over the 7 north route. He finally got to go home after my run at 8:30 p.m. He deserves a medal of thanks for going above and beyond the call of duty… not the risk of future suspensions for 2 earlies a month!

    No wonder labour relations when into the crapper after DLG left… common sense seems to have left with him! Operators are the TTC’s best ears and eyes on the street, they know the job better than any CIS monitors or Route Supervisors—who should be there to empower… give permission for driver initiative, creativity—not to demotivate, discourage and alienate—by rigidly sticking to a schedule that’s lost its purpose.

    Imagine what service could be on a route, if drivers, with CIS assistance, were in charge of smoothing out daily service, adjusting headways for better service reliability on the route each shift; and actually had input with Service Planning in adjusting route service, trip times each Board period… wow!

    First, however, we need a major TTC cultural change… to value operators and proactively use them to make service more reliable (100 Service Planning “Checkers” can only do so much…and can’t possible survey each route each Board period).

    The goal is to give riders what they want: reliable service… so transit is cheaper, faster and more convenient than the car. Happy, engaged TTC drivers who feel their input counts and that they do make a difference is a good start to more reliable TTC service!


  2. Allowing empty buses to hop-scotch past full buses is not incompatible with on-time performance. In Montreal, buses do this very frequently on busy routes. (I’ve always assumed that the drivers do not on their own volition.) The STM has tighter on-time tolerances (-1 min + 3 min) but achieves significantly higher on-time performance. (Montrealers would stare at you blankly if you mentioned the term ‘short turn’.)

    Of course, on-time performance MUST be measured from the customer’s POV – i.e. does a vehicle show up when expected. The time that a given vehicle arrives back at the garage is irrelevant to the customer.


  3. It’s interesting that the TTC has all this data — it sounds like they have the raw materials to build a much better metric. Seems you’d want something that totaled waiting and travel time for riders along the entire route, and compared the actual service to what they’d get with perfect schedule adherence.

    And I think Bob is really onto something: if you can find a good passenger-centric metric, apply it to lines, not individual vehicles, and let teams of drivers and supervisors do whatever works best. (If they can deliver shorter waits and faster trips by breaking the three minute rule, more power to them.) Give meaningful recognition and rewards to the teams running the best and most-improved routes. Peer pressure sometimes works wonders… 🙂


  4. Steve, I’m glad you pointed out the folly of organizations relying on overly simplistic indicators of service. Like many unintended consequences of the so-called “new public management” it is ridiculous to think that a single number can tell you how service actually is.

    Granted I’m mostly a passive observer and don’t know the nuances of the issues, and I mostly work and play in the downtown core (and environs close by), but my experience is that being “on-time” is actually meaningless in Toronto especially in places where you’re talking about target headways of 15 minutes or less. At that level, a lot of people don’t even check schedules and just go to the stop and wait. No one cares if the streetcar or bus is 3 minutes late. They just don’t want to wait half an hour for it.

    Of course this is different in places like Calgary or even Ottawa where some routes come only once or twice an hour. But it seems to me that the principle of managing routes in a large city like Toronto is quite a different kettle of fish.


  5. I spend alot of time thinking about just this subject while sitting on buses crawling along at 40 kph when traffic is going 60 or waiting at a green light or stopping and waiting at a bus stop when noone is getting on or off. I have often wondered if it seemed as stupid to the drivers as it did to me. And what about the other people on the bus. Does anybody actually involved find this to be an improvement in service? To me it is witholding service to improve some perverse metric to please some bean counter who doesn’t have a clue about service.

    I can’t help thinking that people driving past in cars are thinking “Look how slow that bus is. You’d never catch me on the TTC”.

    I stongly agree with Bob’s thoughts on what might be called “employee empowerment”. They seem to be a rich but untapped resource.


  6. I’m also looking forward to the CIS analysis, but there is one place where it will fail to tell the whole story. CIS tells you whether a vehicle is on schedule, but it won’t tell you whether an operator is having to creep along well below the general speed of traffic, and intentionally miss green lights, just to try to keep up with (keep down to?) an over-padded schedule.

    The worst case I’ve come across was driving behind a southbound Dawes Road bus that did no higher than 20 km/h all the way from Victoria Park to Danforth. But as a less extreme (and more routine) example, on the 24 Vic Park coming home from work, operators usually seem to take their time and the trip takes about 34 to 35 minutes. But occasionally I’ll have an operator that drives what seems to be a more reasonable pace, keeping up with traffic and not lagging behind to intentionally hit red lights, and does the trip in a glorious 30 to 31 minutes.

    There must be some sort of a mechanism to cut back on the overpadded schedules while allowing at least a little buffer space. Is it as simple as just transferring a few minutes from driving time to recovery time at a station?

    Steve: Overly generous running times show up as long layovers at terminals unless operators manage to burn it all up enroute. If individual drivers are running off schedule and causing bunching, this also shows up when you look at the line as a whole. A graphical analysis of a day’s operations shows this sort of thing extremely easily and much more quickly than pages full of numbers.


  7. Leaving it all up to the drivers is not the full answer. The west end of the Queen line is kind of like that right now–the good operators are conscientous, and the ones who like to run late run late (including long breaks at the loop).

    This morning, I waited from about 9 AM to 9:23 AM for an eastbound streetcar at Brown’s Line. (Service is supposed to be “FS”, then 9:14 and 9:25).

    A streetcar went by westbound about ten past nine, and sat in the loop for a good ten minutes before emerging. By this point the following streetcar was already in the loop, and a third westbound car was in sight.

    CIS on the car said -17 as far as I could see when I boarded. Why did he sit in the loop for 10 minutes, letting a 20+ minute gap open up ahead of him? Beats me, but I’m not sure this is the operator I want determining service reliability. Obviously had he just gone around and headed back, he wouldn’t have been so far behind that he had to load people at every stop between Long Branch and Humber.

    Of course the following streetcar–probably the 9:25 AM–followed us all along Lakeshore, in other words sticking to its schedule. So instead of a streetcar about every 10 minutes, we got two streetcars about every 20 minutes.


  8. An addition to my comment–the problem this morning could have happened just the same on a revived LONG BRANCH route, so in this case it isn’t the unreliability of a long route (unless it was a very long bathroom break after a cross-town trip).


  9. Sometimes I wish the TTC would do away with schedules altogether for the frequent routes, and instead try and keep buses/streetcars certain legnths apart, like each streetcar leaves Long Branch every 10 minutes apart (for a round number). This eliminates bunching, overloading, etc. If driver A pulls out 9 minutes behind schedule, driver B will pull out 1 minute later and create bunching. If driver B could only wait for 10 minutes (or less), that would keep the service frequency about even (which is what most riders care about most).


  10. The most surprising thing for me was the number of buses I saw doing this while I was in Montreal a few weeks ago. The second most surprising thing was the longest I had to wait for a bus ever was 45 seconds.


  11. Andrew C has the right idea. On routes where frequency is identified as “frequent service”, throw the fixed schedule out the window and go with a headway structure. The easiest way would be to release vehicles on a fixed headway from the terminal loop. Where there are multiple loops in regular service, things would need to be a little smarter so the bus is released into a gap. CIS would probably need to be modified to track the vehicle progress based on the departure time from the loop.
    In addition an operator tracking program would be required that would manage driver’s time by exchanging vehicle with another driver.


  12. For those times when a route is identified as “frequent service”, would it not be possible to reprogram CIS so that the display the driver gets (+2, or -3, or whatever) is calculated not from how far the driver is off with respect to the schedule, but how far they are off from being halfway between the bus ahead of and behind them.

    That is, if the bus in front is 2 minutes ahead and the following bus is 4 minutes behind, then the display should read “+1”.

    I know the big picture sounds complicated as each bus will have a display that is updating based on its spacing relative to the one leading and the one following, but this is a heck of a lot simpler than some of the paces we put modern computer equipment through.

    Steve: Alas, CIS is rather aged technology and the concept of managing headways rather than schedules is totally foreign to its design. It tracks individual vehicles and plots them on a quite rudimentary diagram of each route. A similar problem exists on the subway where the signal system does a great job of holding trains for their time if they are early, but does nothing to moderate service when it is late.

    With luck, the TTC may see a new vehicle monitoring system sometime in the next decade. The big question is this: will they simply replace CIS with newer hardware, or will they really think about what it takes to manage a large transit system with frequent service?


  13. Read the comments and the thing that was not mentioned is the lack of cooperation between drivers on that route. In the pre-CIS days, if the first bus was running late and had a full load, the operator would only service stops that passengers on the bus requested. The operator would by-pass stops that had people waiting at them and let the bus behind pick them up. This allowed the first bus to make up some of the lost time and allow the second bus to stay more on schedule. This only works if the two buses are going to the same destination.

    Steve: And is so eminently sensible that of course the TTC did away with the practice.


  14. What was the stated reason for doing away with something so sensible???? Reading threads like this is vaguely maddening. Seeing Montreal do just fine with surface route headways is even more so.

    You mentioned that CIS is due to be replaced in the next decade. Have we no hope of widespread improvements in service management before then???

    Steve: Improved service management is really a question of the internal culture at the TTC. There is certainly a change at political level, but we need senior management whose orientation is to customers first.

    To be fair, the last few decades have not been kind with constant budget cutbacks and a political atmosphere that promised support for transit but rarely delivered. Just keeping the lights on becomes a top priority under those circumstances. Now that we have hope for real political support, we need realistic changes to improve service, not impossible requests for reserved lanes on every street in the city.


  15. Hi Steve:-

    This thread reminded me of a time when a friend and I chose a day to go to Hamilton to ride the pre Ti Cat painted Brill Trolley Coaches. It turned out the day we had chosen was quite a snowstorm. This created its own challenges getting there but non the less photographically turned out to be very worthwhile. We too had the opportunity to see radio and CISless operators choose to make a positive difference for their passengers.

    While riding inbound on Barton, we caught up to the coach ahead. Because of being a trolley, we couldn’t play leap frog, but the driver of the coach in front of us played at skip stop, only stopping when passengers wanted to alight and missing stops where he didn’t have any one wanting off. Our driver happily went along with him, thereby giving the late coach the chance to get ahead a bit by the time we reached downtown. A handful of passengers were inconvenienced by a few seconds because the first bus passed them up, but they all took it in stride understanding fully the reasoning behind the decision these two drivers made.

    The attitude expressed by the driver of the coach we were on sounded like a true practicioner of Union Solidarity by Local Brothers; ‘we help each other when we can’, he stated. At the same time, they helped their passengers.

    Steve: I have been told by operators through emails here that skip-stopping is considered a disciplinary offence at TTC. If anyone wants to confirm or deny that, please leave a comment. If this is true, it shows how an organization can lose track of its real purpose of providing good service.


Comments are closed.