Human Transit

school starting times can transform transit efficiency

Matt Conway writes:

AC Transit (Oakland, CA and surroundings) recently came to an agreement with schools to adjust the start times at those schools to reduce the cost of providing bus service. I've never heard of schools adjusting [start times] to save agencies money!

http://www.actransit.org/2011/08/26/end-of-summer-adjustments-in-bus-service/

Start and end times of schools can have a huge impact on transit agencies that are expected to serve them.  If three large schools choose to start at the same time, rather than at staggered times, they can dramatically increase both the fleet needs and the operating cost of the transit system.  To serve such a short, sharp peak of school demand, the transit agency needs more drivers to work very short shifts, and must own more buses for them to drive.  Staggered school times can replace four buses with one, and replace two 1-2 hour driver shifts with one 3-hour shift.  Note that drivers must often be paid more to work a short shift; often the rule is that they must be paid for four hours even if they work one.  That's understandable; would you be willing to commute to work for just one or two hours' pay?

When I worked as a network design consultant for small agencies across California and the American West, I encountered this problem constantly.  Schools were under intense budget pressure and were cutting their own school bus service, shifting their transportation needs on the transit agency.  Schools would change start and end times for their own reasons, and falsely assume that the transit agency could deliver a fleet of buses at whatever time they choose.  It's an example of a common problem wherever many government agencies overlap; your agency can appear to "solve" a problem by shifting it to some other agency.

Has this situation improved?  Are many public transit authorities now co-operating with school districts to ensure that expensive transit resources are used efficiently?

beyond grey

San Francisco's Bay Area Rapid Transit system (BART) is, let's be frank, extremely grey.  Most of its above-ground stations feature vast expanses of exposed concrete, true to the prevailing modernism of the age.  (Most of the system was designed in the 1960s.)

At stations like MacArthur, where the grey station infrastructure interacts with the surround grey ramps of the freeways, one can wonder if the original BART planners were so obsessed with competing with freeways that they deliberately chose freeway-like lines and colors, especially where real freeways were nearby.  This, of course, would be competition by resemblence rather than by differentiation.  At one stage, that probably made sense.

And yes, cool grey can be beautiful, but only if there's color to throw it into relief.  Modernism sometimes drew encouragement from the coolness of classical Greek and Roman architecture, but of course the ancient world seems colorless to us only because paints, fabrics, and other vehicles of color don't survive the centuries. 

So it was fun to open my mail this morning and find this painting by Alfred Twu, reimagining the freeway-dominated landscape of MacArthur BART station with a more tropical sense of color.  Why must we go to Germany to see bright colors and strong choices in design?

UPDATE:  I can't resist highlighting a comment from jfruh:

I always think that BART is what someone in 1969 thought the future was going to look like.

If you're too young to remember 1969, I strongly recommend reviewing Stanley Kubrick's great film 2001: A Space Odyssey (1970).  When I rode BART for the first time in 1976, I felt like I had arrived in the world of that film. 

on casual carpools, or "slugging"

Emily Badger has a useful article on casual carpools, though it would be a little more useful if she -- or her editors at Miller McCune -- didn't keep implying that public transit is somehow the enemy.

Casual carpooling -- or "slugging" as some of its partisans like to call it -- is a perfectly rational response to very congested freeways with High Occupancy Vehicle (HOV) lanes.  At informal queues, usually located near an onramp, motorists who want to use the HOV lanes meet up with other commuters who want to ride the lanes as passengers.  These passengers fill the empty seats in the motorist's car so that they can all travel in the HOV lane.  The phenomenon appears to happen where and when an HOV lane offers quite dramatic travel time savings, as it does on certain Washington DC freeways and on the San Francisco Bay Bridge.  It happens only in intensive commute periods, because that's when the HOV lane's advantage is substantial.

For many, it's fun to think of casual carpooling as some sort of revolt against conventional transit.  The term slugging, Badger explains, arose as an insult uttered by "bitter bus drivers" who saw their waiting passengers disappearing into private cars.  Miller McCune's headline describes slugging as "the people's transit," as though conventional transit is something else.

In fact, casual carpooling or "slugging" is largely compatible with conventional transit.  Really, the two are mutually beneficial.  The casual carpool markets in San Francisco and Washington are both parallel to rapid transit lines, but the trains are still full.  As for competition with peak bus services, the long one-way commuter bus run is one of the most expensive services a transit agency can operate.  Often, each bus can be used for only one run during each peak, so all the costs of owning and maintaining the bus must be justified by a single trip.  Drivers for these peak buses are also expensive, because there are costs associated with the short shifts that peak-only service requires, and because drivers must usually be paid to get back to where the shift began before clocking out.     

Long commuter bus runs can still make sense, but they are very expensive compared to conventional two-way, all-day transit.  If casual carpooling reduces the demand for them, the effect on transit is to flatten the overall peak that transit has to serve, increasing its potential cost-effectiveness and improving the utilization of fleet.  It's especially helpful on the AM peak, which is usually the sharper of the two.

So slug away, if you need to feel that you're attacking something.  I prefer to call it a casual carpool, because that term describes what it really is.  And I see no reason not to welcome them.  In fact, when new HOV lanes are developed, the casual carpool phenomenon should be planned for, both by ensuring that there are safe and logical pickup points and also by counting casual carpool trips in the mobility benefits of the lane.

Of course, such planning would contradict the libertarian fantasy -- heavily stressed in the Miller McCune piece -- that casual carpooling is a "government-free" form of spontaneous social organization, a kind of Tahrir Square for the cul-de-sac set.  In fact, "slugging" is a freely chosen response to the design of the government-funded transport infrastructure -- just like everybody else's commute.   

the horrors of "transferring" in 1974, and a happier future

Connections, or transfers as North Americans depressingly call them, are the foundation of a simple, frequent transit network that's there whenever you need it.  I laid out the basic argument here, but in brief, a transit system that tries to run direct service from everywhere else (so that nobody has to make a connection) ends up as a confusing tangle of hundreds of overlapping lines, few of which are frequent enough to rely on or simple enough to remember. 

But once you decide to offer free connections, you face challenges.  Rail rapid transit systems usually let you change trains behind the faregates without paying another fare, but surface transit systems (bus, rail or ferry) struggle with how to encourage transfers without encouraging fare evasion.  If you give paying passengers a transfer slip providing a free ride on a connecting service, you get all kinds of abuse: people who don't need the transfer slip sometimes sell it, or give it away, or simply drop it where someone else can find it, and ride free.

The blog of Vancouver's transit agency TransLink, called The Buzzer, recently featured this 1974 explanation of "transfer regulations" in that city, as printed on the route map. It's a nice example of the kind of regulation that was common at that time.

Note how much complexity follows from trying to insist that transfer slips must be used only for connections, and that you must not do anything else while connecting. The authors have thought of everything you could conceivably do that isn't exactly what they intend, and are resolved to prohibit all of it in meticulous detail.

Fortunately, not many transit agencies take this tone with their customers anymore.   

There seem to be three defensible approaches to connection-pricing, at least in the North American and Australasian contexts where I work:

• One is to stop giving transfer slips, but instead sell a reasonably priced day pass.  This is now common in parts of the US, especially in the sunbelt.  With a day pass, you charge a bit more and in return you give up trying to control how the slip is used.   Of course, this is still unfair to people making spontaneous one-time trips, which are, in my view, a pretty foundational element of healthy urban life.  
• An intermediate approach, modeled by AC Transit in Oakland, is to provide a transfer slip for a nominal extra charge, such as 25 cents.  This is supposed to deter people who don't need one from taking them, and thus reducing the number that are discarded where others can pick them up and use them. 
• The third approach, which I have to say I prefer, is to take a deep breath and just not worry about it. Give out free transfer slips because the design of your system is based on encouraging connections.  Treat the transfer slip as a 90-minute or two-hour pass, good for unlimited rides within that short period.  Do what you can to discourage them from being sold or transferred, but don't try to control how they're used by the person to whom they're issued.  If someone can make a quick round trip in that time, or do a stopover, then that's fine.  Portland's TriMet, Vancouver's TransLink and San Francisco Muni all do this, as do the integrated networks in Melbourne and Brisbane, Australia.  In all these places, and many more, a base fare buys you unlimited use for 90-120 minutes with the fare zones specified.

Of course, there are lots of other kinds of connection-pricing policy out there, including a range of partial policies and non-policies, and it all gets much more complicated where there are multiple entangled transit operators watching their own balance sheets, and no citywide agency with the power to co-ordinate them. 

Many of the security problems around transfer slips should disappear in the new era of smartcards.  A good GPS-based smartcard system will know that you got off another transit vehicle nine minutes ago, so it will be able to determine whether you should be charged a new fare when you board the next one.  It will be interesting to see whether agencies that still have fare penalties for connections take the opportunity to elminate them.  It will still be politically hard: if you eliminate your connection penalty without raising the base fare, your revenue will tick downward.  If you raise the base fare to compensate, lazy journalists will do stories about your evil fare increase, ignoring the larger fairness of the change.

But one thing is true everywhere, because it's a matter of geometry: if you want frequency and simplicity in your network, you have to encourage connections.  So any pricing scheme that penalizes connections, or that shrouds them with punitive rules like this 1974 policy, is in conflict with those basic goals.

do complex networks require complex maps?

In his guest post, Aaron Priven explained the design process that he led for the distinctive AC Transit network map, in 2003.  Here are some pieces of that map.  (For the whole thing in its most recent version, see here for PDF, or here in a version that you can pan and zoom online.)

 

Aaron writes:

One basic principle we wanted to follow was that there should be a one-to-one correspondence between the ink lines and the bus lines. Many American maps are like the Seattle map [pictured here] where no matter how many bus lines are on a street, they are shown by a single ink line. Trying to understand where a single line goes is very difficult on these maps. One has to hunt around for the bubbles to figure out where the route is going, and sometimes this can be very obscure.

For example ...  there is a Line 30 bubble on Dexter Ave N., and another one further west, on an unlabeled street near Queen Anne Ave. N. How does Line 30 get between the two places? Does it dip downtown and come back in a U, or does it travel on one of the two sets of east-west streets shown? Or does it do something else entirely?

We wanted to make it easy to figure out how to follow where a bus line goes, which we thought was a basic goal of any bus map.

No argument there.  But here's the AC Transit map's legend:

Note that there is no attempt at frequency mapping here: a line that runs every 10 minutes can look just like a line that runs every hour.  Instead, they've used multiple colors mainly just to help people sort the lines out. 

Then, to help distinguish between levels of service, they use a range of different kinds of "bubble" symbols.  But they are focused on explaining span -- when service starts and ends -- rather than frequency.  Is that the right priority?  Consider this:  If you tallied every person who looked at this map to figure out a particular route, what percentage do you think care that the last trip of the day is midnight or later?  Now, what percentage do you think care if there's service on the line in the next half-hour?

Frequency usually tracks with span:  Lines that have the longest span are usually major lines in other respects, such as frequency.  But as Aaron notes, that's not always the case, and the AC Transit system in particular is full of complexity.  So the design of this map really comes down to a choice:  Are we trying to accurately portray the complexity of the system, or are we trying to help people see, amid this complexity, the portion of the network that they're likely to find useful?

Aaron states his view clearly enough:

Ultimately, the question behind all of this is how to show the complexity of our system. In an ideal world we might simplify the system itself, but this wasn't a choice open to us in the map project. We chose to try to capture the complexity of the system in the map rather than hide it away by creating something simpler, since we felt this would mislead potential riders. It is a disservice to passengers to portray the system as other than it is.

I disagree.  Emphasizing complexity conveys the message that the service takes a lot of work to figure out, so it tends to encourage people to consider transit only for one or two trips that they've learned how to make.  In fact, despite all the complexity on this map, a large share of the agency's service, and ridership, is on a few major lines.  On these lines the next bus is always coming soon, so if you want to travel anywhere on those major lines -- which together serve most of the major destinations and a lot of the residential density -- the system that you need is really simple. 

On the AC Transit map that simplicity is largely concealed.  The map forces everyone to wade through the complexity, even if the information they need is simple, and even though a large share of customers, perhaps even a majority, are after fairly simple information: Which of these lines is likely to be there when I need it?  Span (when the service runs) is an important part of that, but the distinction between 10 minute frequency and 60 minute frequency affects a lot more people.  The one span category that I do recommend highlighting is peak-only, because peak-only routes exist for such a short period that showing them prominently on the map makes it harder to see the all-day structure, which is the network you'll need to care about if you're going to rely on transit for a range of trips.

Don't miss the comment thread on Aaron's post, which debates this issue in some detail.

guest post: aaron priven on the ac transit (oakland-berkeley) transit map

Continuing the recent series on frequent network maps, today's post is by Aaron Priven, who actually managed the redesign of a network map. I don't agree with everything he says, but the resulting map (current version here in PDF, here in a version that you can pan and zoom online) certainly shows a lot of thought, and it's interesting to see the thought process explained.  I'll share my own responses to this map in a near-future post.

 
Jarrett’s post on frequency mapping, and a number of the comments there, referred to the AC Transit system maps. (AC Transit is the bus system for a large portion of the eastern shore of San Francisco Bay, including cities such as Oakland, Berkeley, Richmond, and Fremont.)

I was the project manager for the redesign of the AC Transit system map back in 2002-2003. As such I am primarily responsible for its design, although most of the specifics were worked out by our cartographer, Kristin Bergstrom, who had previously designed the San Francisco Muni map.

One basic principle we wanted to follow was that there should be a one-to-one correspondence between the ink lines and the bus lines. Many American maps are like the Seattle map, pictured here, where no matter how many bus lines are on a street, they are shown by a single ink line. Trying to understand where a single line goes is very difficult on these maps. One has to hunt around for the bubbles to figure out where the route is going, and sometimes this can be very obscure. For example, on the Seattle map shown in Jarrett’s post, pictured here, there is a Line 30 bubble on Dexter Ave N., and another one further west, on an unlabeled street near Queen Anne Ave. N. How does Line 30 get between the two places? Does it dip downtown and come back in a U, or does it travel on one of the two sets of east-west streets shown? Or does it do something else entirely?

We wanted to make it easy to figure out how to follow where a bus line goes, which we thought was a basic goal of any bus map.

We followed the lead of the Paris and New York maps in using colors designed to be as distinct as possible from neighboring routes, so it would be easy to follow the red, blue, or green line from place to place.

At that time, we were introducing new LED headsign, which were capable of showing several different colors. One compromise we made was that we limited ourselves to the colors available on these headsigns (or as close as we could get in print).

The advantage of using colored headsigns is that it becomes clearer much further away that the headsign is blue than what the route number is. If you’re two blocks away and you want to know whether that bus in the distance is the 11 bus you need, or the 62 bus you don’t, it’s very nice to be able to see the purple light and know that you have time to go get a coffee, because the 11 will be blue-green.

Unfortunately, the color gamut of the headsigns is limited. On the 2001 Paris map, they used a range of different colors – red, maroon, blue, purple, yellow, brown, chartreuse… We did not have those choices. Our choices were restrained even further a few years later when the California Highway Patrol decided the red headsigns looked too much like red reverse lights and started giving tickets to buses showing red! We had to reassign the colors to avoid using red, reducing our palette even further. (We tried to get legislation passed to change this, but while we successfully got legislation to allow us to use most colors, we were still not allowed red.)

For line and bubble symbology, we considered a number of different aspects of each line. We did consider frequency, but we ultimately decided that the most important aspects were variance, span of service, and express running.

Variance: Even more than now, we had a lot of bus lines that had several different variants – different weekday and weekend routes, routes shortened off-peak, two branches serving different areas, and so forth. Many of these did not have distinct route numbers, even using suffixes. We used different dash patterns to show different variants, and keyed the dash patterns to the frequency guide on the back of the map (which had a from-x to-y listing, inspired by Chicago).

Span of service: AC Transit had no regularity when it came to span of service , with different lines operating radically different spans, and while usually frequent service had a longer span, there were many exceptions. There was no standard span for “all day” service. We felt that the most urgent question someone could ask was “can I get home on this line?” For that we decided to base bubble symbols on hours of operation. Originally we had bubble shapes for peak service, 24-hour service, several kinds of all-day service depending on what time the last bus left the origin of the line (after 6 p.m., after 7 p.m., after 8 p.m., after 10 p.m., and after midnight), and a catchall “weird” category for weekend-only service, midday-only service, or other strange hours. If there was any service on the line on weekends, the bubble was solid; otherwise, it was hollow. (I’m not sure where the idea came from that we distinguished Saturday from Sunday service – in fact for years all of our Saturday and Sunday schedules have been identical, except for later service on two All Nighter routes because BART starts later on Sunday.)

As time went on we made some changes, most importantly reducing the number of “all day” bubbles to two, and accommodating the change from 24-hour service on some lines to having parallel 800-series All Nighter lines.

Express running: Clearly if a bus is running on a street but doesn’t stop, that’s an important distinction. Much of our express running is on freeways, but not all, and we have a number of limited stop services. I liked the way this was shown on the Muni map, with thin lines with round stops where the limited stops were, and we copied that, more or less, for our map.

The other important feature we wanted to have was a complete street map along with the transit map, so that people wouldn’t need to consult two different maps. We had a complete street index and intended at some point to add block numbers whenever possible, although the block numbers never quite happened. (The current map does not have all this detail, mainly because of printing costs.)

So that’s what we were thinking back in 2003. I’m not sure we were very successful, and I think much of it needs rethinking. I thought at the time that people who didn’t understand the differences between dash patterns or bubble types would just ignore those differences, rather than be frustrated and confused by their lack of understanding. We didn’t (and don’t) have the money to do user tests of any of this, and I think that’s something we could have found out.

Nonetheless, I think a lot of the principles we adopted are still worth thinking about.

It’s clear from Jarrett’s August 7 post that he uses “The Frequent Network” as a name for service that is both frequent and has a long span of service, but at least at AC Transit the two don’t always go together. Which is a “major route,” a 15-minute service that runs from 6 to 7, or a 30-minute service that runs from 6 to midnight? We’ve had both of those in the past. Maybe this is a detail, and it doesn’t matter as long as “major” is defined in some consistent way, but frequency is just one aspect of a bus route. I, at least, would rather wait 30 minutes for an infrequent bus than wait two hours before finding out that this bus line has no more buses scheduled for the day, or doesn’t run at all today.

We have all kinds of routes, and not all of them fall in the simple categories of “Frequent Network,” “Infrequent All-day services” and “Peak-only service.”

I am still very concerned that on most bus maps it’s too hard to figure out where some lines go.  Our service has so many different lines along certain streets (Broadway, MacArthur, Shattuck, the loop around Hayward BART) that on our map individual lines get lost in the elongated pastel rainbow, especially when we’re limited to colors that are so non-distinct. (What do you mean you can’t tell the difference between the three shades of teal??)  But I’m not convinced it’s any better to basically combine them all and hope people can search out the bubbles, especially when in the real world that kind of map is easy to make errors in creating. (It’s not at all obvious which bus line a particular ink line refers to, especially if the bus line it originally referred to was rerouted or eliminated some time ago but this ink line was neglected. Little short stretches of ink line, one or two blocks long, end up staying on the map years after the bus line is gone.)

The highway map analogy, with different types of ink lines for different levels of service, would lead to a better conclusion than a flat all-the-same map like Seattle’s. Better a few types of line than just one. But transit is inherently more complex than the street network in that multiple lines can and often do go along the same street for long stretches, and it is important to know which. If you are planning a trip across town, you need to know precisely where your transfer point is, and where particular bus lines go. It needs to be as easy as possible to figure that out.

Ultimately, the question behind all of this is how to show the complexity of our system. In an ideal world we might simplify the system itself, but this wasn't a choice open to us in the map project. We chose to try to capture the complexity of the system in the map rather than hide it away by creating something simpler, since we felt this would mislead potential riders. It is a disservice to passengers to portray the system as other than it is.

oakland: a new streetcar proposal

Bravo to Chip Johnson of the San Francisco Chronicle for doing a column on Daniel Jacobsen's Oakland Streetcar Plan, which was just released. 

Jacobsen is an undergraduate at Stanford University who did the entire plan as a research project.  Drawing on the well-established literature of the US streetcar revival movement, including trips to Seattle and Portand to observe their streetcars, Jacobsen lays out a plan for a streetcar along Broadway, from Kaiser Medical Center at the north end to 2nd Avenue and the Amtrak station at the south end.  (The north end is very close to MacArthur BART station, and suggests the possibility of a Phase 2 extension west along 40th to this station and potentially on to the high-rise centre of Emeryville, just off this map to the left.)

The report has chapters on technology, alignment, economic development impact, environmental impact, and costing. 

As Johnson emphasises, this study was a very good deal compared to what consultants might charge:

Jacobson accomplished all of this, including travel expenses to Portland and Seattle, with a $1,275 grant from the office of the Stanford Vice-Provost of Undergraduate Education. But he spent only $987.

That's pretty cost-effective, especially when you consider that Oakland city officials paid $300,000 for a streetcar feasibility study in 2005 and applied for an additional $330,000 in feasibility funds this year.

(To be fair, a few consultants are among the acknowledgments, so some of their expertise may have slipped in for free, much to Jacobsen's credit.)

I have not read every word, but skimming it I'm impressed.  (I've also looked at the pictures!  A good consulting report must make sense to people who only look at the pictures, and Jacobsen's does.) 

Jacobsen is careful to position the streetcar in an appropriate place in the hierarchy of modes, rather than imagining that streetcars could someday go all over the city as they did in 1900.  This graphic, for example, is very helpful in making it easy to support streetcars AND Bus Rapid Transit, and to understand how both interact with longer distance modes such as BART and commuter rail.

The extensive Bus Rapid Transit projects already planned for Oakland are described as complements rather than as competing ideas.  

The proposal is not claiming to be a great innovation.  The streetcar boom promoted by the well-marketed Portland Streetcar has already produced nationwide proposals for similar services, and Jacobsen stays within the safe boundaries of that existing work.  He lays out all the standard arguments, and suggests the same Czech vehicle that Portland used.  But applying existing practice to a new city is still a lot of work, and Jacobsen has done a fair bit of that work, thinking through alignment, economic development impacts, and so on.  Much of the meat of the study is devoted to exploring developable land along the corridor to assess the potential for transformation along the scale of what occurred in Portland's Pearl District.

If anyone expects me to attack the proposal out of some ideological aversion to streetcars, they'll be disappointed.  Jacobsen positions the streetcar as an economic development tool, and does not make unreasonable claims about the mobility it will provide or bus services it could replace.  My critiques of streetcar-centric thinking have focused on the need to understand how little mobility improvement they provide when compared to high-quality bus routes.  (Streetcars that are promoted based on explicit denigration of buses can lead to political neglect of the bus system, and this can make them net reducers of mobility.)  Streetcars do clearly have a near-term redevelopment impact; my only caution is that this impact -- like all facts about the market -- is the result of a complex aggregation of beliefs in the culture, and the one thing we know about culture is that it keeps changing.  I am skeptical that the land value differentials that streetcars create today will still be there a few decades from now, as buses continue to improve in a way that increases their resemblence to streetcars, but all that has been amply hashed out here. 

Jacobsen is an effective writer, and as you can see from his seemingly dormant blog, he's been thinking about this stuff for a while.  An earlier sketch of his for an Oakland streetcar network was my jumping-off point for my post "the problem with downtown shuttles," which emphasised the need for very, very high frequency when serving very short trips. 

While Jacobsen's work is certainly an inspiring bargain, it certainly shouldn't be compared too directly with what a consultant would do for the $300,000 pricetag that Chip Johnson cites.  That price would buy much more detailed design and operation plan, down to a level that would flush out any fatal flaw or cost-exploder.  

The biggest fatal-flaw danger zone, to me, is that there will need to be a lot more thinking about how the streetcar will interact with several very major bus corridors that must use parts of the same street, including some that are meant to function as Bus Rapid Transit.  Jacobsen notes the issue but doesn't delve, and it can be quite tricky.

This same segment of Broadway, for example, is already one of the problematic chokepoints on the proposed Telegraph/International BRT project.  Adding a streetcar to this chokepoint could make it worse.  Buses readily go around small obstructions that occur in their lane, but streetcars lack the maneuverability of buses, so a trapped streetcar can easily turn a small obstruction into a large one.  Ideally, the streetcar should become part of a package of transit improvements that collectively argues for an exclusive lane, streetcars and buses, through this sensitive section.  

With all that said, bravo to Daniel Jacobsen for this tenacious and thorough work.  I wish I'd been this focused as an undergraduate!

quick read

San Francisco Bay Area readers absolutely must read this piece by Tom Radulovich of the BART Board of Directors, on the ongoing problems of transit planning in the Bay Area, and the tendency to under-invest in dense inner cities where the ridership potential is greatest.  

the problem with downtown shuttles

A common misconception about downtown areas is that great things can be achieved by "shuttles" or "circulators," short routes that just run around in downtown.  The problem with these lines is that they save time for the customer only if they are very, very frequent.  This is an issue that separates people who need only a symbolic service (such as a line on the map or a photo of the transit vehicle in front of the development they're trying to sell) from people who want actual mobility.

Frequency is really important, but it's also really expensive.  Doubling the frequency of a service (i.e. halving the "headway" or elapsed time between consecutive trips on the line) comes very close to doubling its operating cost.  If you double your peak frequency (i.e. the highest frequency you run) it also doubles your fleet, which doubles your fleet capital cost, your ongoing mainteance cost, and the size that your storage and maintenance facility needs to be.  So it's not surprising that we see a lot of downtown shuttle services that offer a line on a map, and a photographable vehicle, and even some mobility for people who aren't in a hurry, but that don't really compete with walking, at least over part of their length.

For an example, let me pick on a very credible-looking route from a proposed downtown Oakland (California) streetcar network, from the interesting Bay Area blog 21st Century Urban Solutions, by Daniel Jacobson of Stanford.  Daniel has lots of good ideas, and I recommend his blog, so I hope he won't mind me using his idea as an example.

The blue, red, and green lines on this map flow across the downtown and out into surrounding neighborhoods, so they would presumably be useful to people who are travelling 2-3 miles or more.  

But the orange line is just a simple square loop.  (If you know Oakland, the four streets are Broadway, 17th, Madison, and 8th.  That suggests a one-way loop, which is a topic for another time.  For now let's pretend the loop is two-way.)

The long sides of the loop are about 770m and the short sides are about 710m, so the longest distance served by the route (from one corner to the diagonally opposite corner) is only 1.48 km or 0.9 miles.  So if you walk at 3 mph, a walk between opposite corners of the loop would take 0.3 hours or 18 minutes.

If a streetcar or shuttle bus doing the orange line wanted to be faster than walking, how frequent would it have to be?

Let's generously assume our service averages 10 mph.  Well, the travel time onboard to go 0.9 miles at 10 mph is 0.09 hours or 5.4 minutes.  So if you just missed one trip, and the streetcar is still going to be faster than walking, that means the next one needs to come soon enough that your travel time plus wait time doesn't exceed the 18 minutes it would take you to just walk the whole way.  The in-vehicle travel time is 5.4 minutes, so that means the headway (time between consecutive trips on the line) can be no longer than 12.6 minutes (18-5.4).

To get to that, I've assumed that the only market the streetcar is serving is between its opposite corners, the longest possible trip that this route can serve.   But the average trip served by this route is obviously shorter.  If you assume a shorter trip, the frequency needs to be even higher.  Let's say we're just travelling the length of one long side, say from 8th to 17th on Broadway.  Now the walk takes 10 minutes while the ride would take 3 min, so it's faster to walk unless the next vehicle comes in 7 minutes or less!

If every urbanist and architect and downtown booster had been stepped through that calculation, we'd see a lot fewer short downtown shuttle services, and the few we had would be much more frequent.  Such services may be helpful for people who don't want to walk for reasons other than time (such as weather, frailty, disability, or laziness) but they offer very little to the average working person who really needs to get somewhere and whose time is valuable.

Sensitivity to frequency is also the main reason why spines of very frequent bus service, converging from routes all over the city, are better than dedicated shuttles at serving short intra-downtown trips. Downtown boosters want specialized downtown services.  But if you just want to get somewhere fast in a major downtown, head for a street where lots of bus lines are running on top of each other to provide a very high combined frequency; many downtowns have one or two corridors where a bus is coming every minute or two all day.  The key here is that the bus service is there anyway, it's already paid for, and it's extremely frequent. By contrast, a city will need to come up with every dollar to run its specialised shuttle, and is therefore likely to settle for a frequency like 12 min or 15 min.  That sounds good, but it doesn't compete with walking for the short trips that the shuttle claims to serve.