Robust, equitable and efficient all-night transit for Boston

Do you think that the T should implement real, useful 7-day-a-week late night service? Make your voice heard! Email latenightservice@mbta.com by April 4. More details here. 

A more condensed version of this proposal can be found at Commonwealth Magazine.

The recent post regarding the T's early morning routes has been one of the top three most popular ever posted to this page, surpassing 5000 views (and much more quickly than any previous post). But if you thought that I'd just discovered the early morning routes, you'd be wrong, I've known about them for some time (yet never had need to ride one). However, what piqued my interest was the fact that these routes could be used for something much larger: actual all-night service for the MBTA service area.

In the aftermath of the MBTA’s decision to cancel its recent late night service program, it might be useful to consider some facts that are not well known, and that may provide the pathway toward establishing a robust late night transit service that is regional in scope, that responds to clear needs, and that does so affordably. Of the top 15 transit agencies in the country, only three—Boston, Houston and Atlanta—fail to provide some overnight service. The plan laid out in this proposal is built upon the T’s current early morning service, but rather than serving only Friday and Saturday nights, it is geared primarily toward getting people to their late night and early morning jobs.

The MBTA currently runs approximately a dozen early-morning trips, originally geared towards fare collectors and now oriented more towards early-morning workers (they were not shown on public schedules until 1999). These trips are shown on published schedules—often with just a small schedule notes—but otherwise not publicized (although this page described them in some detail). These trips arrive at Haymarket around 5:00am, with connecting service via the 117 Bus to Logan Airport

A study of early morning service conducted by CTPS (MassDOT’s Central Transportation Planning Staff) in 2013 found these services to be well used. Indeed, there was extreme overcrowding on one route: the single 117 trip (Wonderland-Haymarket) carried 89 riders. In response, the MBTA added two additional trips as well as earlier trips on Bus routes 22, 23, 28 and 109.

This map shows ½ and 1 mile buffers of the proposed late night

network superimposed on the T's current route map.

See a full-size map here.

This proposal would use these trips (with some minor changes) as a baseline for a new, more robust “All-Nighter” service. This would allow the use of current MBTA bus stops and routes, and be mostly an extension of current service, not an entirely new service. It would provide service to most of the area covered by MBTA rail and key bus routes. The changes include:

• The primary connection point would move from Haymarket to Copley. This significantly shortens many of the routes and avoids time-consuming travel through downtown Boston to Haymarket, allowing a single route to operate with one vehicle instead of two, thus keeping costs down. In addition, Copley is somewhat more central to late night activity centers.
• The current early-AM routes provide good coverage near most rail and “key bus” corridors with the exception of the Red Line in Cambridge and the Orange Line north of Downtown. (This plan does not address the longer branches to Braintree and Newton which serve lower-density areas which would have lower ridership and higher operation costs.) To fill these gaps the Clarendon Hill route would be amended north of Sullivan Square to follow the route of the 101 bus serving Somerville, Medford and Malden. A new route would be added following Mass Ave along the Red Line/Mass Ave corridor to serve Cambridge, then run through Davis Square and terminate at the Clarendon Hill busway.
• A separate service would be run from Copley to Logan Airport. It would follow surface streets from Copley to South Station and the Seaport making local stops, use the Ted Williams Tunnel to the airport, and then terminate at the Airport Station, where it would allow connections to the 117 bus, which would terminate there rather than Copley. This bus could be operated or funded by MassPort in partnership with the MBTA, much like the Silver Line, since it would directly benefit the airport. This service could be through-routed with the 117 bus to Wonderland via the airport, which wouldn’t require additional buses and would eliminate a transfer.
• Hourly service would operate on all routes, with a “pulse” connection at Copley. (What's a pulse? Here's the answer.) All buses would be scheduled to arrive at approximately :25-:28 past the hour and depart at :32-:35 past, allowing customers to transfer between the various lines at this time. A dispatcher could hold buses to make sure passengers could connect between lines. With hourly headways, a timed and guaranteed connection is required to provide any network effect and allow access between routes. 
• Cities served by these routes could set traffic lights to “flashing yellow” for the routes between midnight and 5 a.m. to best accommodate schedules (this is already the case on many of these corridors).
• Buses to the airport would allow employees to arrive a few minutes before the hour, in time for shift start times, and would then make a second loop through the airport to pick up employees finishing shifts a few minutes past the hour.
• Airport buses would also allow overnight travelers to make their way to downtown by foot, bicycle, Hubway, taxicab or TNC (Transport Network Companies like Uber and Lyft), and make the “last mile” to Logan on a bus. This is especially important for late-arriving flights to the airport at times when there are often few cabs available. The MBTA could explore public-private partnerships with TNCs or other providers to bring customers to Copley Square to access all-night service.
• The :30-past pulse time would allow workers finishing shifts on the hour to access buses to Copley, or walk to Copley itself, for connections to their final destination.

This service, based on current late-night and early-morning published schedules, would require 10 vehicles for four hours (approximately 1 a.m. to 5 a.m.), or 40 hours of service per day (with an extra hour on Sundays). At this time, the MBTA operates approximately 10 hours of service covering the early-AM routes, so the net hours of service would be 30. In addition, these trips could be added to existing shifts, so rather than a deadhead trip between a terminal and garage at the beginning or end of service, they would utilize a bus already in service, saving an additional 6 hours (approximately) of service, so the net hours per day would be 24.

Assuming a marginal cost per hour of service of $125 (since this service would require no new capital equipment or vehicle storage, because most of the bus fleet lies idle overnight, the full cost should not be used for these calculations), this would cost approximately $1,095,000 per year; assuming ridership of 843 per night (based on existing counts), the net cost would be $757,000, with a subsidy of $2.46 per rider, in line with existing bus subsidies—the cost might be slightly higher if the T needed to assign an inspector to the overnight service and extra police personnel, but they may already be on duty at those hours and could be shifted from overnight layover facilities.

Further, if Massport provided the link between Copley and the Airport on an in-kind basis (as they do for SL1 airport fares), it would reduce the cost to the MBTA by approximately 10%; if Massport through-routed such services along the route of the 117 it would reduce the MBTA’s expenditure by 20%. Thus the range of cost to the T would be somewhere between $600,000 and $1.25 million, between 7% and 13% of the cost of the most recent discontinued late-night service. This service would serve approximately 308,000 riders annually.

While “Night Owl” bus service was run from 2001-2005, it was perceived as serving very different population and purpose than this proposal, focusing on the “drunk college kid” demographic on Friday and Saturday nights only (the most recent late night iteration had the same issue, although the T's equity analysis showed otherwise). While that population would certainly benefit from overnight service, this service would be aimed directly at providing better access to overnight jobs—in addition to the airport, most routes would pass nearby major hospital clusters—especially from low-income areas.

These routes would (unlike the prior late night services) follow existing bus routes and stops, provide coverage to much of the region’s core neighborhoods—but not necessarily to each rail station’s front door. For example, the Green Line in Brookline would be served by the 57 bus along Commonwealth Ave and the 39 bus on Huntington Ave, within a mile of the B, C and D branch stations in the town, thus providing a similar level of service more efficiently (and obviating the need to create nighttime-only bus stops along the rail lines). Most of the densely populated portions of Boston, Brookline, Cambridge, Chelsea, Everett, Revere, Malden, Somerville and Medford would be within a mile of service, with additional service to parts of Newton and Watertown. 

In addition, by following normal bus lines, buses would use existing, known stops along major streets (rather than requiring passengers to search for nighttime-only stops adjacent to or nearby rail stations), and bus numbers could even match daytime routes (for instance routes could be named: N15/9, N28/SL5, N32/39, N57, N1/88, N93/101, N117) to provide continuity. The goal is to make the system both useful and easy to understand both for regular users and customers with less-frequent overnight needs. (Using existing routes would also reduce the start-up costs for such a service.)

The T’s current plans to mitigate the removal of late-night service are anemic, targeting a single line or a couple of trips on a single day. This proposal, on the other hand, would bring overnight service to much of the area which hasn’t had such service in more than 50 years. It would be a win-win solution. It would benefit the Fiscal Management Control Board by focusing on low income areas and job access routes while costing a small fraction of the recent late-night rail service, and by showing that its goal was to provide better service, not just cut existing trips. But more importantly, it would benefit the traveling public, by allowing passengers to make trips by transit to major job sites at all hours of the day.

It would be important, as well, to run this plan with discrete goals in mind; while the late night service was painted as a failure by MassDOT, by comparing the ridership to the previous iteration of late night service, it was an unmitigated success. The T's mitigation plans would add buses piecemeal to its early morning system with no specific performance metrics. Instead, it should look in to creating a better network with specific goals, and measure the efficacy of the system in providing better connections to people traveling at odd hours.

This plan is designed to be affordable and robust, serving real needs across the region, responding to social and mobility equity, and doing so without the need to turn to the private sector, which cannot and will not offer similar service at such affordable costs. Should it work, it would enable the MBTA to set a standard for quality 24/7 service—service which is provided in Philadelphia, Seattle, Cleveland and Baltimore, not to mention peer cities like New York, Chicago and San Francisco—and the kind of service a city and region like ours both needs and deserves.

*****

Here are sample schedules, assuming a :30-past-the-hour pulse at Copley. Schedules are based on current early-AM service. These times would be repeated hourly at 1 a.m., 2 a.m., 3 a.m. and 4 a.m. daily and 5 a.m. Sunday. Each route would require one vehicle unless otherwise noted.

Ashmont-Andrew-Copley (15 Bus, Red Line Ashmont Branch)

Dep Ashmont Station 1:02

Andrew Station 1:17

Arr Copley 1:28

Dep Copley 1:35

Andrew Station 1:46

Arr Ashmont Station 2:00

Mattapan-Dudley-Copley (28 Bus, Silver Line Washington)

Dep Mattapan Station 1:03

Dudley Square 1:14

Arr Copley 1:25

Dep Copley 1:35

Dudley Square 1:46

Arr Mattapan Station 1:57

Hyde Park-Roslindale-Forest Hills-Longwood-Copley (32 Bus, 34 Bus, 39 Bus, Orange Line, 2 vehicles)

Dep Hyde Park 12:50

Forest Hills 1:04

Longwood Medical Area 1:16

Arr Copley 1:25

Dep Copley 1:35

Longwood Medical Area 1:44

Forest Hills 1:56

Hyde Park 2:10

Watertown-Brighton-Kenmore-Copley (57 Bus, Green Line)

Dep Watertown Square 1:02

Kenmore 1:19

Arr Copley 1:25

Dep Copley 1:35

Kenmore 1:41

Arr Watertown Square 1:58

Clarendon Hill-Davis-Harvard-Copley (Red Line Alewife, 87/88/89 Bus, 1 Bus)

Dep Clarendon Hill 1:03

Davis 1:06

Harvard 1:12

Arr Copley 1:25

Dep Copley 1:35

Harvard 1:48

Davis 1:54

Arr Clarendon Hill 1:57

Malden-Medford-Sullivan Square-Haymarket-Copley (Orange Line North, 101 Bus, 93 Bus, 2 vehicles)

Dep Malden 12:49

Medford 12:59

Sullivan Square 1:07

Haymarket 1:17

Arr Copley 1:25

Dep Copley 1:35

Haymarket 1:43

Sullivan Square 1:53 

Medford 2:01

Arr Malden 2:11

Broad & Ferry-Sullivan Square-Haymarket-Copley

Broad & Ferry 1:00

Sullivan Square 1:10

Haymarket 1:17 (express via Rutherford)

Arr Copley 1:25

Dep Copley 1:33

Haymarket 1:41

Sullivan Square 1:48 (express via Rutherford)

Arr Broad and Ferry 1:58

Wonderland-Chelsea-Airport (Blue Line, 111 bus, 117 bus)

Dep Wonderland 1:31

Chelsea 1:44

Arr Airport 1:55

Dep Airport 2:00

Chelsea 2:11

Arr Wonderland 2:26

Copley-South Station-Airport

Dep Copley 1:32

Arlington via Boylston 1:34

Washington via Boylston 1:35

South Station via Essex 1:38

Seaport 1:41

Terminal A 1:45

Terminal B 1:47

Terminal C 1:49

Terminal E 1:51

Arr Airport Station 1:55

Dep Airport Station 2:04

Terminal A 2:04

Terminal B 2:06

Terminal C 2:08

Terminal E 2:10

Seaport 2:14

South Station 2:18

Washington via Kneeland 2:21

St James via Charles 2:24

Arr Copley 2:27

Alternate Copley-Airport-Wonderland through service (2 vehicles; this would provide better connections downtown but may not serve airport shifts as well from Chelsea and Revere):

Dep Wonderland 12:38

Chelsea 12:53

Terminal A 1:06

Terminal E 1:12

South Station 1:19

Arr Copley 1:25

Dep Copley 1:35

South Station 1:41

Terminal A 1:48

Terminal E 1:54

Chelsea 2:07

Arr Wonderland 2:22

Does your transit system have a pulse?

In this post I'll describe something which is probably unfamiliar to big city transit system users, but which is very familiar for those who use smaller systems: a pulse. A pulse is a timed transfer between multiple routes in one location (or, in some cases, multiple locations) where buses wait for each other in order to allow passengers to transfer between them. Large systems with complex networks generally don't use pulses both because of the complexity of scheduling and bus frequency: a transfer will often only mean a few minutes' wait. But with 30- or 60-minute headways on many smaller systems, a pulse is an efficient means to create a usable network.

There are four requirements for a pulse system to be feasible:

• The system must be small enough. With more than 15 or 20 routes, the complexity of scheduling every bus to one central point will overwhelm the pulse savings, and may also result in inefficient and overlapping routings to reach the pulse location. Some large systems may have pulse features in peripheral locations, or certain times of day.
• A convergence of routes. Trying to schedule multiple pulse meets at multiple locations is quite complex, and mid-route meets are operationally inefficient since buses usually need a few minutes of schedule padding to allow for variances in travel times. Most pulses take place at a central location where multiple routes lay over. So a network needs to be focused on a single location.
• Minimal traffic. Pulse networks are based on buses keeping their schedules. In cities with heavy traffic, unless there are busways, a bus that is delayed for five or ten minutes may result in the rest of the pulse being delayed, or passengers missing transfers.
• Not too much crowding. Crowding on transit is generally a good thing, since it means that people are using the system. Too much crowding, in addition to passenger discomfort, leads to slower run times which, much like traffic can cause a pulse network to break down. In addition, crowding shows that more frequent buses are needed, and pulse networks provide coverage and predictability, but are not easy to scale, because to change the headways on one route, you need to change the headways on all the others.



Burlington, Vermont's 30-minute pulse system shows how routes of
different lengths operate with different numbers of buses; this is
more difficult with an hour headway pulse.

• Routes have to be similar in length. Pulses work best when a single bus can make a roundtrip in an hour, including schedule padding. Issue arise when, say, a destination is 28 minutes of schedule time from the pulse location. You can't feasibly run it with one bus, since if that bus is at all delayed it will either delay other pulse buses or cause missed transfers. But putting two buses on the route is a poor use of resources, since each bus will now lay over more than half the time. So an hourly pulse network works only with routes where most round trips can be completed in under 50 minutes, and where others are long enough so that resources don't sit idle. This is less of an issue with 30 minute pulses, as routes can be shorter and still align with the pulse.

San Francisco's late night transit services involve a series of timed transfers.

Where are pulse networks run? Pretty much everywhere. For instance, many of the regional transit authorities in Massachusetts run pulse networks, even if they don't advertise them as such. For instance, if you look at the schedule for nearly every bus in Brockton's BAT network, it will leave the "BAT Centre" (or the BAT Cave, and after it receives the BAT signal; the Centre has received high praise from Miles on the MBTA; another feature of a pulse system is that it makes it worthwhile to invest in central infrastructure since all routes serve it) at exactly the same time. Buses pull in, passengers transfer, buses pull out. Simple.

Once headways drop below 20 minutes, transfers become very, very difficult if they're untimed, which is why pulse systems make sense in lower frequency networks. Most of the time, transferring between subways in New York means waiting only a couple of minutes for a train. But after midnight it is often an exercise in futility if you have to change lines, since you may spend as much time waiting in a station for up to 20 minutes as riding the train. Without information on departures or guaranteed transfers, even the country's only full-service 24-hour subway loses much of its utility.

Jarrett Walker has a good if somewhat wonky description of how a pulse system works here, as does this presentation from the Chittenden Country Transportation Authority (in Burlington, Vermont, where I stole the graphic above). Even San Francisco gets in on the pulse system, for it's late night service most buses start in a single location, and a few other timed transfers are accommodated as well. In Boston, the transfers to the 117 downtown during early morning service are a proto-pulse, although with a more robust overnight service, a pulse would make more sense.

The T's "secret" early AM service … unmasked

In 1960, when the MTA cut overnight service (for the first time), some trips were retained. At the time the purpose of these trips was to allow MTA (and later MBTA) fare collectors to get to subway stations. This shadow system was not made "public" until 1999, but by "public" it means that the trips have different numbers combining multiple routes and are shown only on some online timetables and on printed timetables as just a note in very, very small print.

But they're incredibly useful. Say you have a 6:30 departure from Logan Airport. Without this special knowledge, your only option is to drive and pay to park or take a taxicab or TNC vehicle. Everyone loves paying $30 to get to the airport, right? The T is useless for flights that depart before 7 (the earliest outbound Blue and Silver line services get to the airport around 6 a.m.). Even though the airport runs at full capacity at 6 a.m., many flights depart earlier, and most airport staff have to arrive by 4 or 5 in the morning. Once you learn the secret of the early AM buses, you can get to the airport, or downtown, quite a bit earlier.

An outdated map of early-AM T services; the 109
service was added in 2014 after a study showed
demand for additional early services.
The network actually serves most of the region!

Here are the routes covered by the buses. There are two sort-of-separate services, the ones which operate to Dudley to connect to the 171 bus at 3:50 and 4:20 and have a later trip downtown. The others have a single trip downtown to meet the 117 for a connection to Logan (as far as I know, the T does not guarantee this transfer by having the 117 hold until connecting buses have arrived). They are as follows (I'll mention internal route numbers in the 191-197 series since those are sometimes referenced in schedules or online trip planning):

• The 15 bus operates trips to the airport via Dudley and Andrew from Ashmont, as well as to Haymarket. The later trip follows the Silver Line's route, the first use the 171; the later trip does not have connecting service to Logan. These trips are shown on the 15 bus schedule; the later trip is officially known as the 191 (see, more confusing than it needs to be).
• The 28 bus operates from Mattapan to Dudley and meets the 171 and 15 as shown above for transfers. These trips are shown on the 28 bus schedule.
• The 32, 24 and 39 operate as one continual trip from Hyde Park to Roslindale to Jamaica Plain, Copley and Haymarket, and connect to the 117 to the airport. This trip is shown on the 39 bus schedule, although the route is officially the 192. This route does not operate on Sunday.
• The 57 bus operates from Watertown to Kenmore, Copley and Haymarket, officially as the 193 although the trip is shown on the 57 bus schedule. This route does not operate on Sunday.
• The 89 and 93 buses operate from Clarendon Hill to Sullivan Square and on to Haymarket as the 194. This is shown on the 89 bus schedule.
• The 109 and 92 operate from Broadway and Ferry in Everett to Sullivan Square and on to Haymarket. The portion of the trip to Sullivan is shown in the 109 schedule and the 92 schedule. This should allow a transfer to be made at Haymarket to the 117.
• The 117 operates several early morning trips inbound to Haymarket in addition to the connection outbound to the airport. 

Is there any information on the MBTA's website about these services? No! I can't explain this. The only map I could find was from a 2013 study of these services from CTPS; the T can't be bothered.

These buses run, they have plenty of capacity (well, most do) and they are, for all intents and purposes, kept secret from the traveling public. The schedules are buried, there's no information about connecting services to Logan, and no effort has been made to create an "early AM" page with information about which buses run, where the run, and when they run. Most of these buses have been running these routes for close to 60 years—and close to 20 years on public schedules—yet no one knows about them. And the MBTA's website does its darndest to keep customers in the dark.

Yeah, real helpful

For instance: if you load the 171 bus schedule, you get an error message that there are no trips, because it automatically loads the inbound schedule, which indeed doesn't have any. You need to load the outbound schedule to see the trips. And the 171 is good for Hubwayers; there's a station to drop your bike right in Dudley.

Or check out the 57 bus schedule. It shows a bus leaving Watertown Yard at 4:33 and arriving at Kenmore at 4:50 (a trip which, during rush hour, is scheduled for more than twice as long). Yet there are no times given for any intermediate stops. So does the bus make these stops? Probably. But who's to say it doesn't run express? If you want to take the bus from Brighton (Washington Street at Chestnut Hill Avenue) not only are you not given a time, but really no guarantee that the bus would actually run.

Despite this, these routes provide a good base for a discussion about how late-night MBTA service could actually be provided, not just on Friday and Saturday nights, but every day, for allowing low-income workers to get to jobs at the airport and elsewhere. With the T required to mitigate cutting late night service, and currently proposing a very weak mitigation plan, that's an additional discussion we need to have. But for now, the agency at least ought to tell people about the service they already provide!

Siting Rail Stations: New Bedford

The South Coast Rail plan between Boston and New Bedford and Fall River—should it ever be built—raises a lot of hackles because it costs a lot of money and its benefits are hard to quantify.

(What I'd add is that in addition to current commuters getting a significantly faster trip to Boston than the current highway system, Fall River and New Bedford could be very attractive "gateway cities" with good "bones"—old, attractive housing stock, walkable downtowns—and natural amenities—namely, the Atlantic Ocean—but are currently just a bit too far from the major employment center in New England to take advantage of that. Thus, they can't provide affordable housing for people with jobs in Boston. One hour trip-time rail service would change that equation dramatically. New Bedford is a lovely town with a lot of vacant land. But traffic renders it isolated from Boston. We need to connect it to the rest of the state.)

But what I want to address today is the siting of the rail station in New Bedford, which—if it is built where the plan currently shows it—would be a major planning failure.

Here's a raw screenshot of New Bedford: where would you put a train station?

Just looking at the layout of the city, it's pretty easy to spot the downtown, and the surrounding densely-populated neighborhoods. The rail corridor runs to the right of the roadway east of downtown, so it would make sense to site it somewhere east of downtown, just based on this. Let's take it a logical step forwards: here is that same map with some annotation of major traffic generators, attractions and transportation nodes:

Oh, that makes it even easier. There's direct access to the ferry terminal with service to Martha's Vineyard, so instead of having to drive to Woods's Hole across a clogged Canal bridge, people going to the vineyard could take a train from Boston (or the 128 Station, or even the park-and-ride stations further north along the line) to New Bedford, walk on to a boat, and have a city-to-island trip of 2 hours even. The Whaling Museum would be a stone's throw away, so it could catch tourists from Boston taking the train to see the attractions (this happens, if there's service). And it turns out that Downtown New Bedford is really quite nice, with very pleasant and walkable narrow, cobbled streets extending several blocks inland (the city was developed 3/4 miles inland by 1893, although some has succumbed to some pretty dreadful urban renewal), much like Portland, Maine's Old Port.

Now, where is the proposed station? In about the stupidest place possible.

Should we build a station near downtown, a quick walk across a city street, and adjacent to the ferry terminal? Or half a mile north of downtown, where you have to cross a highway to get there, and nowhere near the ferry? What does MassDOT think? They think that the second option is better. To channel John McEnroe: you cannot be serious. And much like the umpire's call which led to his outburst, this is a terrible decision.

No. Wrong.

The state's idea is that the Whale's Tooth Station—as it's called–will spur economic development in the currently-industrial area nearby. That may be true. In that case, build a station there, and then have trains terminate downtown. But the zoomed-in image of a site plan for a station in a city like New Bedford should never have an arrow with the words "to Downtown".

It gets worse: part of the reason they've selected the site is that it allows development of a parking structure to serve both ferry passengers and rail passengers. Of course, ferry passenger would have to walk half a mile, or probably take a bus shuttle, and many rail passengers would be driving because the station is really only easy to get to by car. Build it in the right place and you obviate the need for the parking entirely: people driving from further afield can use the King's Highway station just off the highway a few miles north, and passengers from New Bedford can walk, or take a taxi, or a bus, to the station downtown. Those ferry passengers, instead of driving downtown, can park in King's Highway, or Taunton, or Westwood (or take a train from Downtown Boston), and take the train to to the ferry.

Why two downtown stations a mile apart? Because New Bedford's population is concentrated along the coast, and two stations allow easy access without a car. Census tracts immediately adjacent to the coast in New Bedford have population densities in the 5,000–8,000 range, but much of the area is commercial and industrial. Just inland, population densities range from 12,000–18,000 with triple-decker lined streets much like other New England cities—as dense as Boston, Cambridge and Somerville. A city like this should not be served solely by a single car-centric station.

And the multi-modal ferry connection is icing on the cake: current plans call for 75 minute train times, but some documents suggest that sub-60 minute times would be attainable. Given that the plans are to electrify the corridor and that it is, for the most part, arrow-straight—if built to 110 mph standards, the 16 mile tangent section between Taunton and King's Highway could run in 10 minutes, station-to-station—this should be easy for an limited-stop train; the 1:16 time includes eight stops between 128 and New Bedford, so a summer Friday evening train could easily make the run in an hour. The ferry time from New Bedford to the Vineyard is an hour, although a faster vessel could probably cut that to 40 minutes. With a few minutes to transfer, a two-hour trip to Vineyard Haven would be attainable. This is faster than the current driving-plus-ferry time, which doesn't include a traffic buffer, and half an hour to park or line up for the ferry. It's probably just as fast as flying, when you factor in getting to the airport and security. You could get on a train at South Station at 5 o'clock and be on the Vineyard by 7. Try doing that today on a Sunday morning in March, let alone a Friday in July.

All of this would work … if you build the right connection (and, no, I'm not the first to make this argument).

In the 1990s, the T made massive siting errors with the stations in Plymouth, Newburyport and others, which suppresses both ridership and economic development. The as-proposed station location in New Bedford wouldn't be quite that bad. But it could be much, much better; and New Bedford is a much bigger city than Newburyport or Plymouth. Luckily, we have time to fix it, and do it right.

Let's use numbers to examine the T's late night debacle

Surprising no one, the T cut late night service. Why? Because it cost money to operate (like any transit, but still). Because it wasn't as efficient as rush hour passengers. While we don't have all of the T's data, we have some, so let's look at it, shall we?

1. Passengers per hour on a partial system 

According to the T via BostInno, the late night service carried 16,000 passengers when it ran until 2:30 (two hours of service) and 13,000 when it ran until 2:00 (90 minutes of service). Per hour, this means that the system carried 8000 and then 8700 passengers per hour. From the same post, there are 72,000 passengers during a peak commuting hour, 33,000 during a weekend afternoon, and 14,000 during the first hour of service in the morning. 

Of course, these numbers are misleading, because while all routes run at rush hour, and most run early in the morning and during the weekend, fewer than a dozen bus routes run during the late nights, accounting for 300,000 daily passengers. So the universe of people who could use late night service is only about 70% of the total passengers (and this doesn't account for reduced train boardings because passengers can't transfer to a bus that doesn't run). If only the late night routes ran between 5 and 6 a.m., we'd expect ridership to be just 9800 per hour, or about what late night ridership is. (And if you were to, say, look at a Saturday morning, it's probably significantly lower.)

So by that logic, should we cut service at any time it is less than, say, 10,000 per hour? Maybe we shouldn't run service before 6:30 on weekdays? Or before 10 on Sundays?

2. Cost effectiveness compared to rush hours

The T has claimed time and again that late night service isn't cost effective, because it costs more than average. I'm not making this up: the T's chief administrator said exactly this. But that's not how averages work, especially with a peak-demand service like transit. It is always going to cost less per passenger to provide transit at high use times, and more at low use times. The idea is where you want to draw the line. The T paints the picture of this money-hemorrhaging service during late night hours, but only compares it against the overall average, which includes times when the T makes money (yes, during rush hour, the T makes money). A much fairer comparison would be to compare it to, say, all weekend service, or 10 p.m. to end of service on weeknights. Those times probably have similar cost effectiveness. If you compare it to 8:15 a.m., it's extraordinarily inefficient. But if we had the numbers to compare it to, say, 10:30 p.m. on a Tuesday, or 9:00 a.m. on a Saturday, I bet it would hold its own.

It should be notable that the stated cost for late night service went up from $7.68 in 2014 to $13.39 in 2015, a 74% increase, even as, when measured by passengers per hour, it became more efficient. This might be some more dubious—I mean, innovative—accounting from the FMCB, and it wouldn't be the first time I've accused them of that. 

3. Defining success (moving the goalposts)

When the Night Owl service ended in 2005, it was a failure: at the end, only 655 passengers were using it nightly over two hours: 327 per hour. (It should be noted that in 2004, the subsidy for the service was higher than the 2013-2015 service.) The current iteration of the service carried 13,000 passengers in just 90 minutes, averaging 8667 per hour, 26.5 times as many. The service didn't make money, but it was never going to. The beancounters saw the numbers, and declared it a failure. It's too bad, because based on its ridership, it was a success.

I think that a lot of the work that the fiscal and management control board is doing is worthwhile, and there are certainly places in the MBTA budget where efficiencies can be found to reduce costs. (I've, uh, strongly suggested some on this page.) Yet when they want to cut service, they have a knack for massaging numbers to fit their narrative, even if the bigger picture doesn't support it. This makes the whole operation much less credible: decisions made with poor data should be anathema to their process.

Unfortunately, they're not.

Penciling out cost savings from a working D Street

Last week, I was quoted in the Globe (*) about the travesty that is the D Street light. The city has said they've fixed it, but they haven't, because oh noes, cars might have to wait! So buses still wait at a 100-second-long traffic light for a green signal. Apparently they tried transit priority and it didn't work (which is a pitiful excuse; it's proven technology) and gave up. So we're back to square one.

Now, there's a lot of discussion about the T's huge deficits and the need for fare hikes, but this sort of low hanging fruit is apparently anathema. I've calculated it out before, but here is a quick sketch of how much money could be saved in operational costs by having a working transit signal preemption at D Street. One caveat is that at rush hour there might actually be enough bus traffic that signals might not be able to let every bus through at speed: there are 30 buses per direction per hour, so one bus per minute, and the cross traffic needs at least, say, 20 seconds for a pedestrian signal. But given that the current average delay, with deceleration and acceleration, is one minute, even 20 seconds is a heck of a lot better.

Let's say that the average bus experiences a 45 second delay (this is conservative, especially since schedule padding is required based on the longest time possible to wait, more than three minutes per round trip, or nearly a quarter of a Silver Line Waterfront round trip). For the Silver Line Waterfront shuttle from Silver Line Way to South Station, this accounts for 10% of the total run time, for the SL2 it's 7% and for the SL1 4%. How many trips are affected? Based on recent schedules:

SL1: 128 trips weekdays, 99 Saturdays, 126 Sundays.
SL2: 142 trips weekdays, 75 Saturdays, 70 Sundays.
SLW: 67 trips weekdays.

Total: 337 trips weekdays, 174 trips Saturdays, 196 trips Sundays. Or 2055 trips per week. And these are one-way trips, so actually 4110 trips per week (this will go up once Silver Line Gateway service begins running). Assuming that each of those trips loses 45 seconds, we're talking about 3082.5 minutes of operating time per week wasted sitting at the D Street light, or 51.375 hours per week. 52 weeks in a year gives you 2671.5 hours of savings, and at $163 per hour (the cost in 2013 to run a T bus) the cost savings amount to $435,455 per year. It's probably higher by now.

Let me repeat that: by fixing D Street, the T could operate the exact same amount of service on the Silver Line, but cut costs by nearly half a million dollars. The argument that it would just be built in to schedule padding is spurious, since a more predictable light would save considerably more time, and the run times are short enough that the same amount of service could be operated with fewer vehicles, or the same vehicles could run more frequently. (For instance, if the Silver Line Waterfront service time went from 15 minutes to 13.5, a bus could make 9 round trips in two hours instead of 8.)

If the FMCB is serious about cutting costs, they should be banging down the doors at BTD for this sort of common sense solution to save money. There's no need to cut service; in fact this would increase service quality. Yet everyone is content to pass the buck, and years later, service never improves.

(* Kind of funny story, but Nicole reached me at the Birkie Expo in Hayward Wisconsin, and I spent ten minutes going on a rant about how dumb everyone at the T and BTD is and basically said "yeah, all of this is on the record." Because it's all true. And if you think this blog is excitable about transit, I may get just as excited about ski marathons.)

Suburban car sharing at Commuter Rail park-and-ride lots

Suburban park-and-ride lots are a fact of life. While rail stations near housing result in better land use and higher ridership (Windsor Gardens in Norwood, for example, has 600 riders per day with no parking. Of course, driving downtown from Windsor Gardens is a 1 to 2 hour endeavor at rush hour, on the train it's 35 minutes.), there are some areas where sprawling populations are too dispersed for walkable stations and park-and-rides make some sense. (And, no, I'm not referring to somewhere like Ashland, Westborough or Kingston, where 1990s-era stations were built away from downtowns but next to big parking lots; despite this these stations have ridership no higher than Windsor Gardens.)

Take Littleton, for example. Its old train station is little more than a crossroads, not even the center of town, so the park-and-ride nearer routes 2 and 495 makes sense (the town itself is quite dispersed, a combination of sprawl and conservation land). The same could be said of South Acton, where the always-full parking lot combines with a small density node, although a walking-based station at West Acton would help to relieve some of the parking capacity issues at South Acton. Littleton's parking is also at capacity, and will likely continue to increase when new schedules come in to effect this spring with 30-minute travel times to Porter Square, speeds unattainable by car without traffic, let alone on Route 2 at rush hour.

Littleton's observed catchment extends to several nearby towns, and grassy areas adjacent to the lot are frequently used by motorists to park when the lot is full. Expansion is possible but not cheap: the MBTA-owned parcel is on a slope and construction costs would exceed the $5000 to $10000 per space typical of construction on level ground. But there is a need for more parking: the Fitchburg corridor serves a dispersed population with poor highway access to Downtown Boston, and providing parking allows many more people to use the train rather than drive downtown.

What makes Littleton somewhat different from other park-and-rides is the presence of several large nearby corporate campuses, most notably Cisco three miles to the south and IBM three miles to the north. Each campus houses more than 1000 employees, and each has abysmal non-driving access. (IBM is served by the end of the LRTA Route 15, which takes only 45 minutes to get to Lowell's train station.) For all intents and purposes, access to either of these campuses requires a car.

Until the double track was completed to Littleton and the station there rebuilt, there were no reasonable reverse-peak transit options to get past South Acton, a longer drive from either of these campuses. But with the new schedules—and the recent promise of an even earlier train to Fitchburg—there are now hourly-or-better reverse-peak trains to Littleton, allowing city dwellers to get to Littleton's train station without driving (even against peak traffic, getting to and from Littleton at rush hour is no picnic). Of course, this station is still miles from major employers, hence the proverbial last mile issue.

There are some possible solutions. Bicycles are allowed on reverse-commute trains, and the roads in Littleton and Boxborough are relatively bikeable (cyclists can use a closer back entrance to Cisco closed to vehicular traffic to keep cars from cutting through local neighborhoods). But cycling isn't for everyone, especially given the narrow area roads and heavy rush hour traffic. Uber operates in the area, but the density of cars is relatively low (meaning potentially long wait times) and fares high: $7 to $10 from the train station to employer, as much as the train ticket from Boston. Corporate shuttles are another option: RedHat and Juniper run a shuttle from Alewife to their Westford campuses and the new CrossTown Connect TMA is exploring shuttle services from Littleton. But shuttles are expensive; a single route can cost $50,000 to $100,000 annually.

If only there was a parking lot full of cars sitting unused all day …

Imagine the following:

Jack lives in Boxborough and works in Boston. He leaves home at 7:30 to catch the 7:45 express train at Littleton. He arrives at the parking lot and parks in a reserved parking space at the Littleton station and takes the train to work. In the evening, he takes the 5:35 express from North Station which arrives at 6:25, gets in his car and drives home. 

In the meantime, Jill, who lives in Cambridge, gets on the train at Porter at 7:53 and arrives in Littleton at 8:38. She then gets in Jack's car—parked in that reserved space nearest to the platform—and drives 10 minutes to IBM, where she works. In the evening, she returns the car to the same spot in the station and gets on the 5:15 train back to Boston, and the car is awaiting Jack's return.

What incentive would Jack have for letting Jill borrow his car like this? Cars aren't cheap, gas isn't free, and most people won't lend their car to a stranger. But Jack can be incentivized to lend Jill his car. Littleton's parking availability is 2%, meaning that Jack isn't guaranteed a space, especially taking this later express train. And the cost to park is $4 per day, or $70 per month, meaning that the annual cost for Jack to park is $840. In this scenario, we can offer Jack  free parking with a guaranteed space; money is nice, but the extra 10 minutes at home every morning may be worth a lot more to Jack than $840. This idea might work at park-and-rides which aren't at capacity, but the guaranteed space is a kicker.

But with free parking, isn't the T just subsidizing Jack and Jill's arrangement? Not at all. If there wasn't a vehicle available for Jill's last-mile commute, she wouldn't take the train, which costs her nearly $3700 per year. So the net increase in revenue for the T is nearly $3000 annually.

Jack is still letting a stranger borrow is car. But maybe Jill isn't such a stranger after all. There is some precedent: car sharing company called RelayRides (now Turo) pioneered peer-to-peer car sharing in the past decade, although they have now relaunched based on longer rentals. FlightCar has a very similar model to the plan proposed here, except for longer periods of time at airports. Instead of paying $20 to $30 per day to park at the airport, your parking is free—and you get paid if someone rents your car. More to the point, Jill is an employee of a major corporation, not a random person getting off of a train. Jack would be assured that she wouldn't go off joyriding during the day, but rather drive the car a few miles to the company parking lot (where, perhaps, Jill would have preferred parking). And because employee schedules are very set, it would be the same person driving Jack's car every day, not whoever swiped their Zipcard.

Once money changes hands, car sharing does get more complicated, although this model could be used in the longer term to help further reimburse Jack for fuel and maintenance expenses beyond the parking savings. Insurance would also be an issue, and an umbrella policy would have to be set up, perhaps through a TMA or a larger organization overseeing this program if it were to grow. A safety net would need to be provided, but one already exists: both Jack and Jill could participate in an employer-based guaranteed ride home program in case Jack needed to come home early and there was no way for Jill to get to work otherwise, or if Jack called in sick and there was no car for Jill to drive to work, or if Jill stayed late at the office and Jack needed to fetch his car.

Technologically, the car could be set up in a number of ways. The easiest (and cheapest) would be to give Jill a spare key; this would work for a test project. A Zipcar-esque swipe-in system could be set up, but these cost hundreds of dollars to install, and might not be appropriate for a pilot. Installing GPS doesn't come free, but Jack might have more peace of mind with tracking of Jill's movements (when she's driving his car) and even text alerts to let him know when she had returned the car to the station, so he would be assured of having a car waiting for him to get home at the end of the day.

There's also the question of vacation schedules: if Jack goes on holiday, there is no car waiting for Jill. At those times, back-up cars could be used. Jill would have separate keys for back-up cars of other participants in the program. When Jack wasn't working, she could use some of those cars, and those participants would also benefit from the guaranteed and free parking. (This would be even easier to set up with car sharing technology.) And since Jack and Jill would have each other's emails, they would know vacation schedules ahead of time, so there wouldn't be surprises. (It's only an issue when Jack is on vacation anyway; when Jill is out of the office, Jack's car just sits in Littleton all day.) If there were five outbound participants in this program, perhaps two back-up cars would be necessary.

In addition to cost, this program would be superior to shuttles because of the flexibility involved. Because Jack's commute downtown from Littleton is nearly an hour each way and Jill's commute from Littleton is only a few minutes, she has the ability to take an earlier or later train and still get in a full work day without affecting Jack's commute. With a shuttle, it would require multiple trips at a higher cost. With shared cars, Jill could shift her schedule by an hour with no ill effect for Jack. Shuttles also lose efficiency when they have to make multiple stops at multiple employers, which are often not in a straight line (violating Jarrett Walker's "be on the way" principle) while car sharing would allow commuters to more easily disperse to suburban job sites.

Assuming Jill switches from driving the whole way to driving and transit, and assuming 225 work days per year, she would save more than 10,000 vehicle miles traveled annually. Jack might save VMT as well; on days when the lot is full, he'd have a guaranteed parking space and wouldn't have to drive in to work—or to a nearer park-and-ride like Alewife—when the lot was full. In addition, this sort of scheme, at a larger scale, may actually increase the utilization of the parking lot. It's not too far-fetched to imagine one inbound commuter parking at 7:00 and a reverse-commuter leaving at 7:45 and a second inbound commuter parking at 8:30 and a second reverse-peak commuter using that car at 8:45, which doubles the efficiency of a single parking space.

With enough utilization, a shuttle would make sense: if 15 cars were being driven between Littleton and Cisco, for example, a 14-person van meeting each train would be feasible. But shuttles often suffer from a chicken-and-egg conundrum: without riders in place, there's little incentive to invest $100,000 annually in a shuttle. Car sharing could provide a bridge to shuttle service for larger employers, and for smaller ones, provide a link for employees living in the city to work at suburban campuses.

Like park-and-rides, suburban office parks are a feature of the suburban and exurban landscape—like it or —although they are quickly losing allure (GE moving from suburban Connecticut to downtown-ish Boston is one such example). But they won't disappear overnight. By leveraging the existing Commuter Rail network, park-and-ride infrastructure and the idle cars occupying the lots, we may be able to provide better options for commuters trying to get to these offices without trying to retrofit transit on to a landscape where it doesn't work well. This wouldn't be a panacea for the poor design of dispersed office parks. But it would provide transit accessibility for minimal cost.