San Francisco bike routes

San Francisco currently has a bunch of numbered LCN routes mapped in OSM. (Example: cycle.travel, OpenCycleMap, Waymarked Trails.)

As far as I can tell, the route numbers are a 1993 system that have since been replaced by destination-based signs.

Some of the routes appear to be outdated too. For example (and this is the particular reason I’m looking at this), Broadway Tunnel is no longer signposted as a bike route and the sharrows and signage direct you over the top instead, though there is apparently a bike-activated warning sign. It does appear to be in the SFMTA ArcGIS map as “Broadway Tunnel Bike Route” without any supporting infrastructure, but if signage/sharrows point over the top, that should be what we map. The route refs aren’t in the SFMTA maps at all.

I suspect I may going to have to once again implement an override to say “OSM bike route relations in this city are problematic, don’t trust them”, but thought I would drop this here in case anyone can advise different or, best of all, fix all the route relations to reflect the current situation.

As with many municipal sign upgrades, some of the numbered route signs have been removed, but others remain. The Wiggle was signposted at Steiner and Waller by number in 2016 but only by name as of 2022:

But elsewhere the oval signs remain:

I haven’t lived there in a few years, but from what I recall from more recent visits, your assessment of the status of the route numbers is about right. The numbering system used to be prominently signed, but they’ve since mostly been replaced by the destination-type signs. I wouldn’t be surprised if there were a few relic numbered shields still out there (in fact, as I was typing this it looks like Minh found some), but the newer signs are predominant and make no reference to the route numbers, nor do any official maps. Based on this, I’d support removing the ref values, or at least moving them to unsigned_ref (poorly_signed_ref?).

Edit: or maybe old_ref?

1 Like

The 1993 system is still extant to the extent that lcn (numbered) signs exist: as Minh confirms, “the oval signs remain.” The newer (destination-based) signs are supplemental: text in Richard’s link to SFMTA calls them “additions to the city’s bike network,” not replacements of it.

While I haven’t OSM-updated the city-countywide lcn routes myself, I was mightily impressed years ago when Contributors did, as OSM, OCM and CyclOSM display one of the most impressive-looking urban cycleway networks in the USA (with a regularized numbering protocol readily understood by Americans familiar with the rules used). SFMTA took the trouble to officialize it by erecting (and maintaining, to this day) signs to reflect it. I wouldn’t use old_ref unless it were certain to be accurate.

Without something authoritative from SFMTA that the network has been deprecated (and it appears Minh provides evidence it hasn’t, I agree), I think it best that the routes remain. Richard, I don’t know how cycle.travel might best address what you characterize as “outdated” routes (and you offer solid evidence of that in at least one case re Broadway Tunnel), as we (the USA’s mysteriously-evolving local bicycle routing protocols) continue to tease out the reality of this.

Of course, it is good we understand that there are distinct taggings for both bicycle infrastructure and bicycle routing (and we do OK or better for both in many places), but as signage for both of these AND the newer destination-based emerge, OSM might find ourselves developing even-yet-newer tagging schemes to accommodate the destination-based signs (too). These have emerged, and so in OSM have become an issue in San Francisco, in Santa Cruz and as a recent topic in the Australian community. It is confusing as it stands today and could use wider community discussion to achieve better or best forward momentum. Though, what seems to exist now is the sort of questioning-as-to-what-is-best we see here and in that Oceania thread…yet, then, resulting stagnation of discussion. I’d like to do better and I think our wider community would, too.

It’s a worldwide-needed solution, really, as these destination-based signs become more prevalent and noticed by OSM mappers. I remain listening, as the volume is only recently getting turned up loud enough to hear. Thanks to all who read and participate.

1 Like

In general, cities haven’t ever done a great job of managing their signs. In Cincinnati, there’s even an Interstate route that probably only exists because the city transportation department never got around to taking down the signs, but as a motorist, you wouldn’t necessarily know that it’s a zombie route. On the other hand, there are county highway routes elsewhere in the Bay Area that had been marked by only a few scattered, neglected signs until all of a sudden the signs got some love again.

In some cases, it might make sense to retag a route as some kind of disused: and map the individual signs if they’re no longer useful for wayfinding. In this case, the proposed override to ignore S.F.’s numbered routes as a network might be warranted.

If it weren’t end-of-business / end-of-week here in our Greater Bay Area, I’d phone up MTA and ask them. Indeed, come Monday, I might just do that.

Two questions might be “do numbered bicycle routes continue to exist?” and “how do the destination-based signs supplement or replace these?”

Additional questions or sharpening of focus is welcome. More next week, I suppose; happy weekend.

Edit: In Santa Cruz, I’ve seen local authorities do some highly confusing (deliberate? likely not, though lazy? well, maybe) things regarding how bicycle route signage does or does not reflect the reality of either the route, any network it might fit into, or anything short of “ignore this sign, even its context confuses”). Though, the recent destination-based signs are a great improvement.

But they don’t remain everywhere: in many places (I think most, but evidently not all), the oval signs have been taken down and newer directional signs put where they used to be. So I don’t think they’d be especially useful for wayfinding, as I doubt you’d be able to reliably follow a numbered route end-to-end anymore.

Additional clarification from SFMTA as to whether the remaining numbered signs are relics/oversights or meant to blaze a still-existing numbered network would be welcome.

For instance, take this intersection, where routes 44 and 49 turn from 23rd to Eureka. In 2016, the oval signs were present directing riders to turn:


By 2022, the oval sign had been taken down and replaced with a directional sign:

So now even if other signs exist along the numbered routes, they can’t be followed through the entire route anymore. I think this is a strong indication that the numbered routes no longer provide meaningful wayfinding and as such don’t qualify for ref anymore.

The existence of (in this case, local cycle network, truly lcns with MUTCD M1-8a) signs do strongly indicate “a route in this network exists.” However, lack of signs, and oh, boy-oh-boy do I know this fairly well about bicycle routes (numbered, in a network…flavors and decades of this…), does not necessarily mean that “a (route or) network does not exist.”

In logic, we say a contrapositive may not be either naturally or non-explicitly existing; there is no assumption that a (or this) contraposition is true. It existence must be proven, really. For example, we don’t dote over speed limit signs missing meaning we do or do not tag a particular way with a maxspeed tag: because of certain data which are both legally extant and locally complex, we very well might. A speed limit sign, sure, the tag is a slam-dunk. But other factors, too. A route can exist without being signed (for example, it can exist in legislation), and that doesn’t automatically mean a network is removed because a sign (or even some) were removed.

I’ll see what I can find out from being a “curious citizen mapper phoning the MTA with some basic questions” I might do early next week. Thank you for the pictures and explanations, it feels like we get closer to this with better discussion and examples.

As I’ve thought about this a few days, the answers to my own questions, given no other data besides speculation are thus:

  1. Yes, numbered bicycle routes continue to exist. Signs of them for some routes continue to exist, so the network itself exists. A route not being signed does not automatically mean it does not exist…well, not without evidence it has been deprecated by the authority.

  2. The destination-based signs supplement the numbered routes. There are numbered routes, OSM has tagging conventions to enter the routes, and we do. To the best of my knowledge, OSM does not map either the destination-based signs (though we could).

This means there are numbered routes (and some are signed), AND there are destination-based wayfinding signs. These are two different semantic things in the real world; one does not replace the other, even as one type of sign might “replace” the other type of sign in at least one instance.

This statement suffers from at least one inherent assumption (limiting its ability to further a useful implication): it was never the case that 100% of the time, you could “reliably follow a numbered route end-to-end.” A given sign that tells you to “turn here to follow 47” is doing nothing more than that in the hyperlocal instance. But even as no, or few if any routes had every single segment of them signed with route ovals, that doesn’t mean they don’t exist (in whole or part along segments where they aren’t signed).

It still is signed, but less prominently. And destination based-signage does not replace these, it supplements these with a wholly different method of “guidance.” One allows a cyclist to “follow routes,” the other aids in “wayfinding to a destination.”

This seems a misunderstanding of either the numbered route system (which isn’t necessarily “wayfinding” in the sense of reaching a destination, but rather “following a route”) and the destination-based signs, which certainly ARE a form of wayfinding. But if you figure out a way to fit together “wayfinding” with ref=* tagging, please do let everybody know here what your thoughts are, as it could be something that gets nods, applause and eventually applied around the world. (Like in Australia; if you haven’t, please see the topic and photos of their not-so-dissimilar destination-based cyclist signage).

Unspoken (or only barely spoken about) are how OSM might tag the destination-based signs, and/or how they might form a network, whether node-based (the location of the sign “to” the destination, on what might be called a guarantee-of-infrastructure-connectivity, implied by the destination-based sign) or simply as a mapped set of nodes representing signs with destinations. I’ve given this some thought and it has some complications, but I very much welcome further discussion about how these signs can be captured in OSM to provide cyclists useful map data, whether rendered, part of how routers might use them, or both.

I did not phone SFMTA, but anybody could and ask those two simple questions.

If the city is no longer willing to tell you how to stay on Route 47 as it traverses The Wiggle, then it’s quite useless as a route. It begins to recall connecting the dots along an abandoned railway based on a few fragments of remaining trackbed…

This is currently documented as destination=* plus destination:symbol=bicycle (which might be qualified by :forward or :backward as needed), but depending on how you get to those destinations, destination:bicycle=* might be more appropriate.

1 Like

If a sign doesn’t exist (because it did, but was superseded by a sign of a different sort of bicycle wayfinding, not bicycle routing) does the route still exist?

I say, unless you can point to it being explicitly deprecated, yes, it does still exist. Am I wrong? “Useless as a route” may not be true with regard to signage, but what if you are in possesion of perfectly valid data in OSM which could route you? (And it exists because those route data once did or actually do “exist” in the real world, as asserted but not signed by an authority). Those data aren’t useless. Though we do seem to (frequently?) call into question whether a route exists solely because it isn’t signed (in a hyperlocal case).

I believe we’ve established that destination-based signs do not replace local routes in a network, but rather supplement them. Yes? It would be good to get wider agreement about this.

Thank you for the destination=* references! However, only what you mention in them (that I could find) says “bicycle” or anything bicycle-specific. That might not stop them from being used for these (destination based signs for cyclists), but it seems we should develop or further this tag (and/or any associated tagging scheme) so that it “better or best” accommodates them. Time for some taginfo research…

Edit: ~1000 uses in various flavors, all in NW Europe. I’ll further familiarize myself with these.

Can you remind me, what’s the source for these numbered routes, other than signs that used to be posted and perhaps online maps that show similar named routes?

Our wiki says the w:San Francisco Bicycle Plan was unanimously approved by the SF Board of Supervisors in 2009. I think it is Chapter 1 that describes the (numbered) routes, though I can’t (currently, quickly) find a link for those. We’re going back quite a few years, whether 2009 or way back to 1993 (which was after I lived and frequently biked in San Francisco, though I didn’t go to SF Bicycle Coalition meetings, though I do attend those in Santa Cruz). So, online references may be few or non-existent.

Full disclosure: I am the author of our wiki page, and I wouldn’t have made it if my sources weren’t valid (at least “at the time”).

Edit: Found it. See Wayback Machine and the resulting document’s Chapter 1, pages 1-7 through 1-10 for tables of the route numbers, which contain Tables 1-2 through 1-4, which “reflect modifications that have occurred since the 1997 Bicycle Plan was published.”

It states, “The signed and numbered bicycle route network developed by the 1997 Bicycle Plan and designated within the San Francisco General Plan’s Transportation Element has been updated over time.”

SFMTA’s 2019 Bike Program Report refers to the Bicycle Plan as its “foundational document”, but how do we know that the numbered route system in chapter 1 of that original plan resembles anything the city still maintains? If we have to rummage through the Wayback Machine or throw our hands up about documents that aren’t publicly accessible, we’re skating quite close to the limits of OSM’s scope. Much as I admire the beautiful oval Golden Gate route shields, I think this obscurity combined with the active replacement of route signs bolsters the argument for at least replacing ref=* with unsigned_ref=*.

1 Like

The Wayback Machine is simply a methodology for maintaining the web-availability of documents “of a certain age.” As the initial publication date of the Plan was 1997, when the WWW was in its infancy, I have no problem with this access method. For example, just because I have to go to the underground stacks (several stories tall / deep) at Moffitt Library (at UC Berkeley) and find a (very) old book doesn’t mean its contents are invalid.

The US Highway System was not “replaced” by the Interstate System (Eisenhower’s National Defense Highway System). Rather, they co-exist. While there may be examples of numbered networks of transportation routes which are “wholesale deprecated,” we know about these if and when they happen: there is at least some public notice, if not fuss and fanfare. We would know about it (and are not presented with any evidence) if SF’s numbered bicycle network (or indeed even a single route of it) were deprecated. True, there was “the Great Renumbering” (of highway routes in California in 1964), a streamlining of legislative routes, the Interstate numbering harmonization and what we know today as “state routes” in our numbered highway system. But that didn’t eliminate, for example, “legislative routes,” (which were always unsigned), it merely clarified them.

Hundreds of people worked for decades to develop this numbered route network and San Francisco erected many (hundreds of?) signs, many remain today. This network did not simply fade out of existence, even if some signs were replaced with a (different) method of wayfinding.

Perhaps I should have phoned SFMTA and gotten answers, that would be definitive. So would proof that SF formally eliminated the network (or specific routes). And I don’t know why you choose the word “replacement” (of route signs with destination signs) when the city’s own description of them (I quote earlier in the thread) calls them additions (to the network), not replacements of it.

Again, you yourself offer that the numbered oval signs remain. I don’t want to get too lost in the weeds of history, like you, and I also don’t think OSM should have things which truly don’t exist. (Though things like proposals, invisible but actual borders and such are exceptions we agree as a community to continue to include, with limits).

Seems like we need some definitive authority on this topic from SFMTA themselves.

Edit: Page 31 of the (2019 BPR) document Minh linked immediately above this post states “Wayfinding, a system of information signs along key bike routes, helps keep people on the right track when bicycling to their destination.” Clearly, wayfinding signs do not “replace” signed, numbered routes. Rather, they are a supplemental aid.

I’m not going to say whether you are wrong or not, but I would venture that you are “outwith standard mapping practice in the rest of the world”. OSM in the US is definitely an outlier in its eagerness to map bike routes without on-the-ground signage.

If that’s what the US community has decided floats its boat, then do that. It’s not what I would do personally but there are lots of things in OSM which aren’t what I would do personally.

But please do tag the fact that they aren’t signed. unsigned=yes is a well-known tag for this, but I wonder if this thread is suggesting a few variants might be needed. signage=superseded, perhaps, or signage=unmaintained, or signage=intermittent, or sometimes signage=proposed.

1 Like

No I do not agree this has been established. For just about every practical on-the-ground purpose in San Francisco, the destination-based signs have replaced the numbered bicycle routes. Most of the time the destination signs have been placed in the physical location where a route number shield used to be. The City no longer lists or promotes the bicycle route numbers anywhere on its websites despite numerous bike-related pages. What more evidence would be needed?

Very few remain today, I’d doubt more than a handful across the whole city. They’ve almost wholesale been replaced, to the point where I don’t think one can really argue the routes are signed in any meaningful sense. For what it’s worth, a city agency (the San Francisco County Transportation Authority, to call back to another fun thread), called the destination-based signs an “upgrade” to the previous (route number-based) design, not a supplement: New Bike Wayfinding Signs Installed Throughout The City | SFCTA.

I don’t find this stance particularly productive for real world applications. If these routes can no longer be verified in any substantial sense by ground truth or active government documents, in what sense do they still exist? At best I think unsigned_ref would be appropriate, though old_ref is probably more true.

1 Like

I never said its contents are invalid, only that they may no longer be relevant to OSM. Your research skills would be quite valuable to OpenHistoricalMap, however.

Funny you should mention this. Legislative routes still exist, technically, but they’re out of scope for OSM. Back in the day, NE2 endeavored to map all the constructed portions of legislative routes as route relations. As far as I know, all these route relations have been deleted as either redundant to sign routes or unverifiable on the ground. The only remnant of this historical mapping is a few scattered unsigned_ref=* tags for constructed but unsigned concurrencies, such as State Route 112.

NE2 didn’t bother to map the significant unbuilt portions of these routes. In the case of SR 112, that highway=proposed would go right through the open waters of San Francisco Bay, because the state never did fill it in. Yet it is, indeed, a designated route that exists on paper.

Incidentally, there is but one sign for SR 112:

Imgur

Richard, I’d like to say I “feel your pain.” Though, by way of explanation, it isn’t quite that the US is OVER-eager to map bike routes (which are signed or “officially” designated), the US reality of both signing practice and mapping practice is that both of these are relatively underdeveloped compared to, say, Europe. San Francisco (at least in the 2010s) was/is a shining example to the contrary: signed, numbered routes in a dense, sensible, cohesive network. There are other US communities which sign local bicycle routes sometimes using M1-8a numbered oval signs, but they are few and far between, with much less density than Europe. So what you may perceive as eagerness might more readily be called an attempt to elevate relatively underdeveloped networks into at least a minimal reality of what they are and how they are represented in the real world. (Sometimes a network “published” by a local authority, sometimes signed — though, again, in the US these are relatively rare compared to Europe).

As SFMTA’s website phone number for their Bicycle Division is non-operative, AND the generic front door (dial 311) phone attendant couldn’t look one up (whew!), he relayed a message to this “city unit” to phone me back with an answer to “Does the City’s signed, numbered bicycle network still exist and/or is it being replaced by signage reflecting destination-based wayfinding?”

And, “we’ll see.”

It may be that unsigned=yes gets tagged on these (that’s one solution, maybe partial, maybe insuffiicent, maybe wrong), it may be that a full, robust data entry of the newer destination-based signage (with a strategy to correctly enter ALL of the destinations and ALL of the signs) as comprehensively as the network=lcn route relations are NOW entered is warranted. Given San Francisco’s status as (in my opinion) the USA’s best shining example of a local cycleway network, I believe that OSM deserves a concomitant “build-up” of the destination-based signs to the same extent that the (deprecated? fading? now-unsigned or less-signed?) existing network is mapped. Especially before it is “torn down” (or widely replaced with unsigned=yes tags). Though, that may be wishful thinking on my part.