How Bad Is TTC Service: A First Quarter Report

Among the TTC’s many promises under its Customer Charter is the provision of quarterly stats on the reliability of each of its surface routes.  This information recently went online on the TTC’s website, although you have to dig to find it.

The path is from Customer Service on the top navbar, then to Customer Charter on the side bar, then to Quarterly reports, and finally scroll down.  Or you can just click here.

This table covers the first three months of 2013, and lists the reliability of every surface route.  “Reliability” is defined roughly as:

  • If the distance between a vehicle “B” and the one preceding it “A” is within three minutes of the scheduled headway, then the vehicle is within the acceptable window of reliability.
  • The measure is taken at various points along a route (we don’t know the locations or number for any route), and summed across an entire quarter’s operation.  This will smooth out everything but very large scale, long-lasting disruptions, and will tend to give an index that tracks the overall behaviour of the route.

The system-wide target for streetcar routes is 70% punctuality (within the headway window), and for buses it is only 65%.  Looking at individual routes, there are huge discrepancies.

No route gets over the 90% line, although several are in the mid to upper 80s.

  • 8 Broadview
  • 31 Greenwood
  • 44 Kipling South
  • 78 St. Andrew’s
  • 510 Spadina

Of these routes, four are relatively short bus routes where congestion is not an issue, and with only a modicum of effort, operators should be able to stay on time.  The Spadina car is a special case because it runs with a very short scheduled headway for much of the day, every day of the week.  It is physically difficult for cars go get more than (H + 3) minutes apart, and impossible for (H – 3) because this would be a negative number.  Service that meets the target is very easy to achieve even if the line appears chaotic at times simply because there are so many vehicles close together.

Life is rather more desperate on some major routes that don’t come close to achieving the system-wide targets.

Routes below 60%:

  • 7 Bathurst
  • 35 Jane
  • 51 Leslie
  • 56 Leaside
  • 58 Malton
  • 60 Steeles West
  • 90 Vaughan
  • 95 York Mills
  • 96 Wilson
  • 102 Markham Road
  • 141 Mt. Pleasant Express
  • 160 Bathurst North
  • 161 Rogers Road
  • 165 Weston Road North
  • 171 Mount Dennis
  • 196 York University Rocket
  • 224 Victoria Park North
  • 303 Don Mills Night Bus
  • 306 Carlton Night Car
  • 353 Steeles East Night Bus
  • 385 Sheppard East Night Bus
  • 501 Queen
  • 505 Dundas

We will spare a special thought for riders on routes that don’t even make 50%:

  • 97 Yonge
  • 99 Arrow Road
  • 105 Dufferin North
  • 107 Keele North
  • 117 Alness
  • 142 Avenue Road Express
  • 145 Humber Bay Express
  • 301 Queen Night Car
  • 502 Downtowner
  • 503 Kingston Road

The two Kingston Road services are notoriously off schedule/headway, and it is not unusual to see pairs of cars on headways that should be over 10 minutes.  Downtowner cars short-turn at locations guaranteed to avoid actually serving any riders while achieving some sort of on-time goal.

And a truly special mention for the basement selection of routes:

  • 144 Don Valley Express (35.90%)
  • 309 Finch West Night Bus (38.81%)
  • 508 Lake Shore (25.08%)

(Lake Shore is probably a special case because it is not uncommon for cars to be missing on this route, or for them to enter service off-schedule.)

These stats are not broken down by time of day nor by day of the week.  From analyses I have published (and many more that sit in my files), there is a common problem that evenings and weekends see service on routes, some of which may have tolerable “average” values like Dufferin (76.03%), completely fall apart.  Nobody appears to be managing the line, the schedules may or may not provide adequate running time.  Buses and streetcars run in packs and short turn with no discernable purpose related to providing regularly spaced service.

For years, the TTC has trotted out stock excuses for service quality and made claims that things aren’t all that bad.  Well now we see their own numbers, and even with the limitations of their methodology, it is clear that the TTC’s claims were hogwash.

Unreliable service is a particular problem on routes with wide scheduled headways, especially the all-night services.  A ±3 minute measure may not be entirely appropriate for a 30 minute headway, but we have no way of knowing whether the night buses and streetcars are only slightly off schedule, or completely unpredictable.  At least with Nextbus, one can find where a vehicle actually might be regardless of its scheduled arrival, but this plays havoc with any attempt to make transfer connections.

In a masterful understatement, the TTC observes:

We continue to seek opportunities to improve performance on our routes.

The first step in recovery is to admit that you have a problem, and with the publication of these indices, the TTC has at least taken that first step.  Whether they continue to blame external factors for every ill remains to be seen, but they have been addicted to that behaviour for decades.

For all the talk of a Customer Charter, the quality of service is a non-negotiable part of what transit is all about.  When people complain about not being able to get around the city by TTC, reliability and capacity are two major issues.  These are linked because, on some routes, capacity is lost to irregular service.  On others there simply are not enough vehicles especially on streetcar lines where a car shortage has been a standard excuse for over a decade.

I cannot avoid a special mention of all those bus routes in the list of poor performers.  For years, detractors of streetcars prated that only with buses could we have reliable service.  Rider experiences on many routes, and now the statistics, prove this to be wrong with bus routes running as bad or worse service reliability than many streetcar lines.  The “flexibility” of buses only allows for even more exotic ways to run erratic services.

In a week when the TTC rolls out its new Group Station Managers, they need to turn a blinding light on surface operations, the weakest part of their network.

33 thoughts on “How Bad Is TTC Service: A First Quarter Report

  1. Maybe if main stream media can pick up on this … they like many, generally love to criticize the TTC for the silliest reasons (sleeping TTC fare collector story tomorrow anyone?).

    How about for a change, we expose the true reason for poor service on majority of surface routes … route management / planing …

    This is the heart of all problems the TTC [has] (not mommy hating bus drivers), its something more buses / drivers/ dedicated surface routes / streetcars / signal priority / … cannot fix on their own … something that does not necessarily take a lot of money to fix either, rather a mentality shift throughout the organization.


  2. I have to ask how the TTC can justify an absolute value like 3 minutes as the standard? Since many routes run with headways that are either below, at or just above 3 minutes it is possible to be one or two full headways off, but still have an acceptable level of service reliability? I do understand the difficulties involved … but still I think a metric that accounts for the specific headway if it is scheduled at 10 minutes or better would be more informative. But on a positive note, who would have thought the TTC would publish something (however imperfect) about themselves that shows them in a negative light. I’m still not sure how I feel about Andy Byford, but good for him for driving this through.


  3. It is very hard to accept “as normal” or acceptable 60.85% on the 506 Carlton streetcar line, esp when you wait 20 minutes in the rain. Even so, I agree that streecars are nororiously better than buses.


  4. WOW… I actually am extremely shocked at how low the performance is. I figured we’d have issues on particularly busy bus routes, or ones that are on the main congested avenues without bus lanes or priority signals or dedicated RoW but this is abysmal.

    As a casual TTC rider (evening/weekend), and one that uses iNextBus, TTCWatch & Rocket Man apps for the iPhone, I am not regularly affected by these details and the app allows me to mitigate (and to be honest, my closest services are the 83, 506 and BD subway)… but my wife riders the BD line east to VicPark and then heads north every day for work… I think I understand her better now 🙂

    What was interesting is that even the streetcar RoW routes like Spadina & St Clair still have lots of ground to make up. I also thought it was interesting that the 504 King is around 70%, when it’s the busiest surface route in the city and I think has more high density along it, whereas the 501 does so much worse.

    The fact that the TTC is trying to be more customer service centric, and that Andy Byford really is a smart man with great experience, and that we have a number of newer/bigger vehicles on order, means that I have a lot of hope going forward… but wow, we are starting from a poor place.

    Would spacing stops farther apart help this? It always surprises me how frequently the streetcars stop. You can usually see the next stop from the one you are at. As it is we have obesity & poor health epidemics in North America, so would it really hurt anyone to add another 100m between stops? I know there is a safety issue during the night time hours, but there is also less traffic then too. Could the TTC operate the “safe-stop” policy at night times to mitigate, but have the standard stops farther apart during the day?

    Steve: Actually, the number of very closely-spaced stops is fairly small, and their elimination wouldn’t do much for service quality (there are problems with close spacings on bus routes too). On streetcar routes, the TTC will not do “request stops” because there is no way to guarantee that passing motorists (who are problems even at official stops) will not run down alighting passengers. 100m is a fair distance for some people, and dropping an existing stop may not work out as simply as it looks. Obesity may be a problem, but it’s not the TTC’s job (or that of the many riders for whom longer walks are a burden) to sort that out. We could just as easily suggest turning off all of the elevators and escalators in the name of “health” while ignoring the fact that many people have problems with stairs. Let’s go whole hog and close half of the subway stations! People can just walk, and look at all the money we will save.

    Service reliability, as I have written many times, is primarily affected by a lack of proper vehicle spacing starting right from the terminals, and this occurs even on days when there is no traffic, no inclement weather, no surges in passenger demand. Service management is haphazard on many routes, and at times non-existent. Stop spacing is the least of their problems.


  5. Unless one knows where on a route these ‘observations’ are made one cannot really see if the figures mean anything. Example: If one is observing the westbound 504 at Victoria Street one will see 504 cars that have come all along King Steeet, if one observes at Jarvis one will miss the (many) that are short-turned and re-enter service at Church. Having figures of any sort is a good start but one does need a context but, as you say, admitting to a problem is the first stage of a solution. However, I think we are still a L-O-N-G way from having reliable surface transit – one hopes based on headways rather than schedules.


  6. I’ve wondered about inter-vehicle communication when short-turns & bunching happens. Seems like every day there is a horror story about a promised “another vehicle coming soon” and people are left stranded for a LONG time, or that the following vehicle is rammed.

    Do drivers along routes talk? Is this relayed through central control vs. direct commz? If my iPhone apps can track vehicles, then surely that info reaching drivers should be easier within the system? Isn’t in common sense if there are 2 vehicles close together with capacity, to hold the leading one, request passengers to shift/aggregate into the one behind, send the one ahead and that would make the whole route run better? Assuming a more empty bus runs more on time as less people getting on/off. For streetcars my model would apply, cause they can’t overtake, for buses it’s more flexible.

    For streetcars, does the driver himself know if travelling from point A to B took X minutes longer or shorter than average? Not sure this would help but curious. In the age of big data, I feel that more dynamic transmission of information could help?

    In terms of drivers not stopping for streetcars, I wonder if there needs to be something more obvious to drivers if a stop is approaching or if the stop request has been pressed. We all get stuck behind streetcars and get frustrated because street parking makes it hard to pass, so I feel drivers are even more aggressive and try to squeeze by quickly as the vehicle slows. Would that help?

    Steve: There was a plan to put displays on board that gave more detailed info about vehicle locations including small maps so that ops could see what was happening around them and position themselves on the line better. That project was part of a replacement of the decades old first-generation systems on the TTC as well as the back end systems. However, it was all a victim of budget cuts.

    Ops are not allowed to use their cell phones while on duty and cannot use, for example, Nextbus to see the condition of the route around them, never mind actually talking to other operators to co-ordinate service.

    As for motorists passing streetcars at stops, they are expected to know the rules of the road. Do you drive through red lights just to save time or because you are frustrated? Of course not. There are times I think that streetcars should have small artillery to deal with the miscreants, and have absolutely no sympathy for motorists (and cyclists) who ignore the fact that they must stop for streetcar passengers.


  7. 65% is the goal?

    That is face palmingly low.

    “Our goal is to only have 1 out of 3 of you ticked off at us.”

    Steve: Considering that in even the simplest of days, you probably ride two vehicles to make a round trip, the “reliability” compounds so that the odds are only 4/9 that you will not encounter a vehicle outside of the “punctuality” window. The more you ride, the lower the chance that you will get good service every time. It’s rather like flipping a coin and hoping for heads every time.


  8. Steve:

    Considering that in even the simplest of days, you probably ride two vehicles to make a round trip, the “reliability” compounds so that the odds are only 4/9 that you will not encounter a vehicle outside of the “punctuality” window. The more you ride, the lower the chance that you will get good service every time. It’s rather like flipping a coin and hoping for heads every time.

    For a system that is built on (and indeed, takes pride in) direct transfers … that is astonishing. It seems clearer and clearer why people agree with the ‘subways x3’ mantra … that seems to be the only way for service to be reliable.

    It’s interesting to note that Chicago is constructing a BRT on Ashland to move people across the city. The expected average 16mph (25km/h) speed that is slower than the parallel Red line ‘El’ (21mph / 33km/h) but it will be far more reliable and consistent than the existing bus service.

    Also interesting is the left-side doors on the buses, meaning that BRT stations can share a central platform.

    It’s clear that TTC service standards have got to go up, and while I’m not picturing centre-running BRT in Toronto’s future, it’s long past the time for introducing bus lanes and better communication and fleet management systems.

    Cheers, Moaz


  9. This assessment criteria is completely wrong for the 144, it’s not even supposed to *try* to get a good number in this area — the 144 riders are an exceptionally picky clique that use very different criteria to assess “good” service.

    But at least it’s a start.

    Steve: Yup — the criteria need to be different depending on the type of route, the headway and the time of day / day of week. The TTC’s one size fits all scheme does not produce meaningful results in many ways. If anything, the +/-3 standard is fairly generous, and the fact the TTC can’t even hit that on most routes says a lot about the quality of service and management.


  10. One wonders how a route like 196 manages to stay under 60%. Perhaps insufficient running time?

    The 502/503 have a good reason to be that low on the charts and that’s because there’s no evening or weekend service to pad out the numbers.

    Now the next step is to break the numbers down by time of day, day of week, timing point, and branch.

    Steve: A low rating for the 196 probably indicates that most trips run bunched. This can happen completely independently of the running time, but a detailed analysis would be required to see just what is going on. Also, the problem may be different at various periods rather than one consistent issue all day.


  11. The astounding inability to get buses and streetcars to depart their terminals on schedule (for low-frequency routes) or headway (for high-frequency routes)… which you have documented very well… remains astonishing. What is to be done about it?

    Steve: For starters, Andy Byford has to start caring about more than how clean the buses and stations are. We have picked the low-hanging fruit, and now must climb to the higher branches. The TTC will be coming out with a five year plan near the end of May, and it will be interesting to see whether and how it will address service quality problems.


  12. You are right Steve there is much to be done and it is all taking far too long. Five Year Plan? How about a “Now Plan?” Look how long it has taken to get just 6 station managers appointed. Clearly there needs to be more supervision ON THE STREET. There should be a “Starter” at every subway station to see to it vehicles leave On Time, not ahead of time, not late but On Time. Novel approach, I know. Prohibition against multiple buses going through an intersection transfer point is another practice that needs to be cracked down on. This creates wider gaps in service and longer waits.


  13. If anyone’s going to be pointing out the MOOSE* in the room, it’ll be you, Steve.
    I’m interested to see if things will finally get better.

    *Major Obstacle Overlooked by Substantially Everyone


  14. I am quite surprised that five Night routes ended up on the below 60% list. Those are 301, 303, 306, 353 and 385. Congestion must be particularly bad at 4 am in the morning.

    Steve: What the night services have lacked for two decades since the suburban expansion is published timepoints for all vehicles with orders not to leave these points until the scheduled time and any planned connections have taken place. Now it’s pot luck like so much other TTC service.


  15. Ok, I have to chime in. Do I understand correctly that the punctuality measure essentially looks at how many vehicles show up the planned amount of time (+/- 3) after the previous vehicle? It’s essentially a trip-per-trip measure?

    If so, that is an entirely unacceptable measure. Never mind the extremely low standard expected or the fact that the negative effects of short turns will not appear in the measure (by which I mean the effect on passengers unexpectedly forced to wait for a second vehicle). The fact is that it simply has nothing to do with customer experience which is supposed to be the target.

    There are lots of possible ways of measuring reliability, but here is one idea that is similar in intent but actually bears some connection with reality: for the “test points” where reliability is measured, note all the times vehicles actually depart that location. Then compute what fraction of the time a passenger arriving at the stop will be able to leave on a vehicle within the amount of time that they would expect given the posted schedule information. Now average over whatever period of time you want.

    Simple. Now, I don’t really propose that as the way to do things, since I’ve only spent 5 minutes thinking it up, but I think it is indisputable that measuring it that way would actually have something to do with the real passenger experience. In particular, for the common case of bunched service, it would read out on-time service during and immediately before the bunch, while reading as not on-time between bunches. In the extreme case where vehicles are supposed to show up every 3 minutes but really show up in parades every 20, it could report on-time less than a quarter of the time (3 minutes before the lead car shows up until the last car of the parade departs).

    Note that the fact that vehicles are close together does not count as a negative in my measure; but that’s fine because what is wrong with vehicles close together is not that they are close together (which is actually good — remember “Always a car in sight!”), but that if some vehicles are close together over here, they must be far apart over there.

    I really want to give TTC management (especially the new TTC management) the benefit of the doubt, but I don’t see how a competent person can create, use, promote or report the existing reliability measure with a straight face.

    Steve: The TTC is looking at a measure used in London that penalizes gaps wider than the scheduled headway (with an increasingly large penalty as the gap widens), but gives no “credits” for short headways. This effectively measures the degree to which passengers are inconvenienced by having to wait longer than the advertised schedule says they should. One problem with this scheme is that above a certain threshold (say 10 minutes), being “on time” becomes at least as important as being “on headway” because riders will tend to plan their arrival at stops to fit the schedule. A 20-minute service that runs 10 minutes off schedule isn’t a lot of good to many because they will spend a lot of time waiting for buses that are not where they should be. Some balance is needed in measurement tools that is sensitive to the type of service offered at different times and locations on routes.

    The TTC has been using the three minute rule relative to scheduled times for years, and effectively hoping that the headways will sort themselves out if vehicles are close to schedule. However, this is hard to manage in some circumstances, and efforts to get “on time” actually interfere with service regularity. It’s like chasing a mirage that always recedes in the distance. The three minute rule was carried over to headways simply because that’s what was already in place, but I don’t think the TTC has thought through the implications of how the formula works under various circumstances. It has severe problems on routes with frequent service because a parade of buses or streetcars can have one big gap (off headway) and many following vehicles that are within 3 minutes of the planned service, even if they are not carrying many riders. As a measure of useful service, this does not work for obvious reasons.


  16. I can’t help but notice that a substantial chunk of the routes missing the 60%, 50% and 40% marks are blue/night services. And while that lines up with my experience using those services, it doesn’t line up with any of the reasons the TTC gives for bad service–there is no congestion problem at 3am on a Thursday night.

    Steve: As I said to an earlier comment, a big problem with the night services is that there are no published and enforced time points or scheduled connections, something that was lost with the suburban expansion two decades ago. On a 30 minute headway, with no demand to stay very close to schedule, it’s easy to get 3 minutes early with light traffic, or to drop back behind schedule after a longer-than-normal break enroute.


  17. L. Wall writes

    “The 502/503 have a good reason to be that low on the charts and that’s because there’s no evening or weekend service to pad out the numbers.”

    With the 301 numbers worse than the 501, and the 306 numbers worse than the 506, I don’t think evening/weekend service is padding out the numbers for the other routes.

    It would be interesting to drill down into this data further. By time-of-day. By different sections of the route. Sure, the 506 gets 60% – but you can be assured it’s worse at Main Station, where much of the service has been short-turned at Broadview, Parliament, and Coxwell (it’s not unheard of being short-turned twice, and even 3 times occasionally, trying to travel from Yonge to east of Coxwell).

    And sure the 501 get’s 50%. But I bet you it’s far lower east of Woodbine, and particularly west of Humber.

    Steve: This type of breakdown shows up in some of the analyses I do which show that average headways and the unpredictability of them are worse on the outer parts of the line. This happens on suburban bus routes too.

    Demands for point-to-point premium fare express buses arise from this abdication of responsibility for service on the outer parts of routes.


  18. Steve:

    The TTC is looking at a measure used in London that penalizes gaps wider than the scheduled headway (with an increasingly large penalty as the gap widens), but gives no “credits” for short headways. This effectively measures the degree to which passengers are inconvenienced by having to wait longer than the advertised schedule says they should.

    Thanks. Maybe there is some hope.

    I agree that a different standard is needed for longer headways. Probably the real solution is to stop scheduling short-headway service and start managing it by headway; then evaluate schedule-based service according to a schedule-based metric and headway-based service according to a headway-based metric.

    Incidentally I took the 506 to High Park on the weekend with the kids, starting from downtown. They saw all sorts of things out the window and had a grand time. It’s a really interesting trip for them, going all the way from downtown towers to an almost rural loop in the park. No short turns or service problems for us this time.


  19. The night routes have different issues-running time on some routes, connections, Thursday/Friday/Saturday nights, and the early morning runs. Running time is either too short (eg 306) or too long (eg 385). Too short on a 30 minute headway leaves little options for adjustments. Problems such as disablements, unsanitary, traffic, fights, etc with few vehicles on the line can greatly cause gaps. Passenger volume can also be quite high on Thursday, Friday, Saturday nights as well as the “worker runs” early morning. Sunday mornings go to 9:30 am. Traffic is very heavy on bar nights. Connections are also an issue-only the 320 and 300 are required to connect for ANY surface route. A lot of operators hold back or advance to create connections, which can temporarily create gaps. One final problem is that evening and night operators do not get breaks, so they do try to get to the ends early to eat, etc. The night and weekend service is primarily ignored by service planning.


  20. The 508 is really a bad example – there are only three or four streetcars that make the run in either rush hour and people can use a 501/504 connection instead. I have to wonder why the TTC really operates the 508 if it is going to have lousy service to begin with.

    As for the 502 and 503, could part of the problem be that they essentially operate downtown on along other routes (the 501 and 504)? Again, except for running between Bingham Loop and Queen St. E., is there really any need for them? Or drop on of the two and move all the streetcars onto the other route (i.e. if they drop the 503, then all the streetcars currently operating on the 503 would operate on the 502, thus increasing service on the 502 run.)

    Steve: The 508 and 503 are both peak period trippers, and their actual operation seems to bear only passing resemblance to the schedules. As for the 502 with its 20-minute midday headway (even though the replacement 22A Coxwell runs much more frequently evenings and weekends), the TTC seems to be incapable of keeping this on time. It will be a moot point this summer as Kingston Road is to be torn up for track replacement, and the streetcars will be replaced with buses. Whether there will be enough streetcars to operate the system in the fall, when Spadina/Harbourfront also will be back to full operation, remains to be seen. Kingston Road and Lake Shore are good examples of forgotten services on the TTC because both are on the outer ends of a major line.


  21. I am surprised the TTC does not consider a do-over of the 13 Neville Park for the duration of construction with buses going from VP station looping around via Kingston Road, Queen and Victoria Park. It would increase service and make for a worthy connection without duplicating service too much.

    Steve: They have not announced the replacement service design yet.


  22. Oh ok. Whatever they do it cannot be any worse than the way they are handling the situation with the Queens Quay track work and its buses. There is NO route management and buses turn themselves, layover and do whatever they please wherever they please on that line.

    If they crunched the numbers for the 509 replacement bus they would find that while frequent drivers are taking it upon themselves to dictate how the route operates (to the detriment of customers). I have seen a number of buses layover for 10 minutes at a time, just this morning the operator of the bus I was on parked on Wellington for 10 minutes for unknown reasons WITH PEOPLE ON THE BUS.. why she could not layover at Exhibition Loop O/S is beyond me.

    Steve: I have watched the 509 on Nextbus, and some of the behaviour is just astounding. At times, I have found many of the vehicles operating in one direction (all westbound or all eastbound) and many having a siesta in Exhibition Loop. The attitude seems to be that it’s a construction zone, and nobody bothers to manage the service. To be fair, as I write this at 10:15 on a Saturday evening, the service is reasonably well-spaced along the route. Maybe someone is watching.


  23. The issue of headway-measuring versus schedule-measuring is a crucial one, because picking the wrong measure can leave the ‘reliability’ figures disconnected form passengers’ experience.

    “On-time” should mean “close to the expected time”. Generally, “close to” means ~5 minutes (although I’ve seen anything from 3 to 10, depending on the system and mode). However, “expected time” is harder to pin down. If the frequency is so high that no-one uses the schedules, then people don’t expect to wait longer than the headway. For less frequent service, people expect to turn up at the scheduled time.

    Personally, I think “on-time” should mean within 5 minutes of scheduled time for headways over 10 minutes, and a maximum gap of headway plus 5 minutes for headways of 10 minutes or less.


  24. Steve wrote:

    “I have watched the 509 on Nextbus, and some of the behaviour is just astounding. At times, I have found many of the vehicles operating in one direction (all westbound or all eastbound) and many having a siesta in Exhibition Loop. The attitude seems to be that it’s a construction zone, and nobody bothers to manage the service. To be fair, as I write this at 10:15 on a Saturday evening, the service is reasonably well-spaced along the route. Maybe someone is watching.”

    And maybe that’s partly why you need a supervisor actually on the streets. It’s great to have a supervisor in a room somewhere tell them over the intercom system/phone system to get a move on, but it’s different if the supervisor is on the street at Exhibition Loop personally watching what is occurring.


  25. Torontostreetcars. I totally disagree. There should be no supervisors on any regular service route either in a room or on the steet. Only in exceptional situation should a supervisor be involved (accidents, construction or busses going out of service). The rest can be managed by computers and is a very simple problem with the correct on board notification systems and on street priority systems. Especially for routes with no branching or short turning.

    This should be phased in on routes that perform badly. Operators should be docked if they don’t follow the on board notification.

    When was the last time you went to McDonald’s and got burnt French fries? Never. Once the machine starts beeping even if people are very busy they go and dump the fries. Buses need a similar system. If they haven’t started to move when break is over they start beeping and flashing lights. Problem solved.

    Steve: It’s not quite that simple because the variety of situations that can come up would require quite a repertoire of computer responses. That said, the first thing that’s needed is a recognition that the problem exists and work on strategies that can deal with this. Some of them will no doubt involve improved computer systems, and some are plain old human management.

    The situation on Dufferin is considerably worse than on some other routes about which I will write when I have time, and I suspect there is a long standing “culture” around how this route operates.


  26. Steve wrote:

    “That said, the first thing that’s needed is a recognition that the problem exists and work on strategies that can deal with this. Some of them will no doubt involve improved computer systems, and some are plain old human management.”

    That’s the thing, you always need a person to provide the computer with information to assist with the information. For example, how does a computer know that the 509 is currently operated by a bus, or that a bus route is on diversion because of an emergency (i.e. a fire) or due to construction? Only a person will know this and can put this into the computer. Supervisors will always be required. And sometimes they do need to ‘see’ the problems for themselves. Even a video camera will not give you the same sense of what is going on as actually being there in person.

    That’s why sometimes I believe that supervisors need to actually go out to where problems are. So, for example, if buses are continuing to back up at a terminal point, sometimes a supervisor needs to go to the area and see what’s going on. And sometimes a superior showing up will do more good than some devices beeping at you.

    Steve: Any robot can follow a recipe, but if the “lore” that knows certain dishes cook better in specific pots, or that the freshness of the ingredients are will affect the taste, or that no matter what the recipe says, the dish will burst into flame if left in the oven, if this knowledge and the ability to detect those conditions isn’t included in the “recipe”, then the robot’s work is for nothing.


  27. Steve:

    Any robot can follow a recipe, but if the “lore” that knows certain dishes cook better in specific pots, or that the freshness of the ingredients are will affect the taste, or that no matter what the recipe says, the dish will burst into flame if left in the oven, if this knowledge and the ability to detect those conditions isn’t included in the “recipe”, then the robot’s work is for nothing.

    All true … but I’d be happy if TTC could deliver McDonald’s style consistency. I’ll take the robots! 🙂

    BTW, 502 is a 16-minute scheduled frequency now, not 20. I was about to make a sarcastic comment that doesn’t make difference to it’s actual arrival times – but a quick look at NextBus at 12:30 pm, shows it’s actually running okay … for once.


  28. The large majority of issues TTC faces on a daily basis are already entered into computers – in near realtime – construction, accident and emergency information is already all standardized. Current traffic conditions are available on most major roads, and expected travel times per time of day can already be determined from the GPS historical record. Some vehicles already have load monitoring (rough I know) which could be expanded (empty vehicles should behave differently than full ones). Vehicle scheduling (bus or streetcar) as well as when staff are supposed to go on/off schedule are also already entered into a db somewhere (and even if not, it’s fairly easy to determine by the vehicle numbers reported by the GPS system what type of vehicle).

    Steve – I agree that a computer is only as good as what it knows … but there is a lot more information available than what any one human could know, especially one sitting on a street corner with a clipboard.

    The other issue of course is that the on-street supervisors actually have little control – they can only stop vehicles at a specific point … this of course doesn’t work on long routes, and for short routes the supervisor is likely not even assigned. The vehicle driver with a light or beep or whatever has the ability to improve performance of the route on a minute by minute basis, for the entire route (multiplied by the number of drivers).

    Let me be clear though, I don’t think we need to build ATC into every route 🙂 I’m just saying that on routes that perform badly on a regular basis, let’s get creative and implement some of these obvious technologies. At the beginning we can keep it simple and build a repertoire of tricks … I’m sure most route supervisors would rather be dealing with the complicated issues rather than stupid things like when vehicles are leaving stations.


  29. To be honest, I’m surprised that the “reliability” of the 196 is so high. If it was only bunching, then the route would be tolerable. Instead, it is well known that operators will “hide” at Downsview Station and York U to have other operators behind pick up their loads, or worse, miss entire trips altogether.

    There are other routes where this is known to happen on occasion, but none as common, obvious or frequent as the 196.

    Dan Garcia
    Toronto, Ont.

    Steve: The way the TTC calculates the stats, this sort of behaviour won’t be picked up on a frequent route. The reason is that if one bus is missing, the resulting headway “gap” may still fall within the three-minute window especially if the bus behind the gap tries to close things up a bit. The most extreme example of how the math fails to report things is on the subway. It would technically be possible for half of the trains to be missing, but those still in service would be within three minutes of the target headway. Failing to count the number of vehicles as well as the space between them obscures this sort of problem, but looks really good in the performance stats.


  30. TorontoStreetcars wrote:

    That’s why sometimes I believe that supervisors need to actually go out to where problems are.

    That is absolutely true, and I’ll repeat an example I have mentioned before.

    A few years back, the TTC removed most (all?) supervisors from the streets and placed them at a desk with a computer with “tools” such as CIS information and Google Maps (or was it Mapquest?). Traditionally, supervisor positions were held by people with seniority and experience, but this new environment dissuaded more experienced employees from taking the position and the TTC had to resort to filling the positions with people having less “street” experience. The move wasn’t all bad, since younger supervisors tended to be more computer savvy, but even that had its problems.

    An operator I know told me of a time when he was driving the Lawrence East route. For some reason, there was a problem that required him to short turn at the east end of the route. The supervisor, with his computer maps ordered him to turn south at Manse Road and go down to Coronation, then turn right to go over to Homestead and come back up to Lawrence. He asked the supervisor if those instructions should be the other way around, and the supervisor was upset (to put it lightly) that an operator would correct him. He explained that there was no traffic light at Homestead and Lawrence and making a left turn there would be difficult and time-consuming, while the intersection with Manse provided the benefit of the traffic light.

    The supervisor approved this correction to the instructions, but it goes to show that people who know the ground conditions, or are at least there to make decisions, are a better way to operate.


  31. Is anyone else amused by how bad the blue-night reliability is?

    Since TTC’s primary excuse for poor service times is “traffic”, how can they possibly explain why the 300 Bloor-Danforth BARELY meets their already pathetic 65% reliability?

    As a semi-regular rider on the aforementioned route, I find it’s not uncommon to have buses “pile-up” with large gaps followed by 3 or more buses grouped together on a virtually empty road.

    By my count, 7/23 of blue night routes aren’t meeting TTC standards and 4 more are barely passing (less than 67%).

    So nearly half (47.8%) of blue night routes are failing or in danger of failing to meet standards, during the lowest traffic-flow period of the any given day.

    Steve: What this reveals is that nobody is managing the service, but moreover, that nobody cares about the service. Don’t forget TTC management have at times tried to cut back night service because, except for a few routes, it is “uneconomic”. The idea that it is something to be nurtured, with reliability and improvements such as protected connections, simply isn’t part of “TTC culture”.


  32. I’m surprised so many “premium” express routes perform so poorly. Considering the extra fare and that these routes are designed to meet the needs of Toronto’s elite who don’t live far enough out to take the GO train, these would have been a top priority.

    As for the 196, if there is any route which should be run based on headways rather than schedules, it is that one. The line benefits from not only a high frequency, but its own lane and right of way as well. getting a bus to hold because it is too far ahead of the one behind/too close to the one in front should be relatively easy. Perhaps the TTC could implement headways on surface routes as a pilot project on this route to see how it affects performance.

    Steve: I think that the problem is generic — headways are not well managed, especially on routes with infrequent service. Under those conditions, it is easy to stray more than 3 minutes off headway (especially when as an operator you have no way of knowing where other buses are relative to you), and the reliability numbers fall through the floor. On frequent routes, some of the service will fall within the three minute rule simply as a matter of statistics because there are so many vehicles on the route.

    As I have written before, when the scheduled service has a short headway, it is possible for many vehicles to not even operate and the service would still score as 100% “punctual”. For example, a five minute headway allows for buses to be up to 8 minutes apart. Thus a route that should see 12 buses per hour could get by with 7.5, or to put it another way, with bunching of four buses that would not count against the route.

    Conversely, a route with a 12 minute headway has a window from 9 to 15 minutes, and so a 5 bus/hour service could actually have only 4 and still be “punctual”. However there would be less leeway than on the more frequent service.

    Basically, the methodology used by the TTC does not produce comparable results between routes or time periods. In some systems, a bus running early (any form of bunching) counts against a route, but in Toronto it does not. Also, because no account is taken of whether, thanks to short turns, buses passing a point are actually of any use to people who might be waiting there, the quality of service can be overstated.


  33. Has there been any discussion of how the Toronto surface routes compare to other systems of similar size/scale and that use similar metrics? Even if the metrics don’t match up, it would be interesting to see what the different systems in North America state their on-time stats are … if for no other reason than to understand what all of their goals are.

    Steve: I wrote about this in a previous article, to the extent I could find published info.


Comments are closed.