Where the Digital Sidewalk Ends | Hacker News Reader

Where the Digital Sidewalk Ends

smolways.com

128 pointsjmward0173 comments

I don't know how solvable a problem this is. There are two roads that would be labeled nearly identically that are an option for my commute. Let's call them Road A and Road B.

Road A: 45MPH, with a dedicated bike lane with no barrier.

Road B: 40MPH with a dedicated bike lane with no barrier.

The similarities end there though:

Road A: Many turns with trees providing poor visibility

Road B: Relatively straight, good visibility

Road A: Typically has obstructions in the bike lane

Road B: Typically no obstructions in the bike lane

Road A: People speed rather massively

Road B: Many traffic signals; people who speed tend to take parallel Road A instead to avoid the signals.

Road A: The bike lane is sometimes only 30cm wide

Road B: The bike lane is wide enough to ride 2 abreast, and there is usually an additional 30cm painted area between you and traffic.

Road A: Car v. bike accidents are fairly regular, with multiple fatalities in the time I've lived here.

Road B: Every car v. bike accident I'm aware of was at night and involved an intoxicated driver.

I have 3 parallel options for going in this direction, in order of increasing total distance for traveled: Road A, Road B, and a MUP (i.e. combination bike/pedestrian path).

For obvious reasons, I prefer Road B, but it's hard to label a map in such a way that it won't pick the two best choices as Road A (shortest) the MUP (avoid highways).

The regional planning organization in my area uses a metric called Level of Traffic Stress: https://www.dvrpc.org/webmaps/bike-lts/

And they made a routing tool based on this: https://www.dvrpc.org/ruti/

Using the metric from the documentation[1] they would both be LTS-3 with Road B varying from 42% to 60% and Road A at 50%:

Both are over 35MPH; Road A is 2 lanes with an unbuffered bike lane and Road B is 4 lanes with most (but not all) stretches having a buffered bike lane.

1: https://www.dvrpc.org/webmaps/bike-lts/pdf/bikestressdocumen...

Have you tried OpenStreetMap? You can add many details like the width and quality of the bike lanes, and many routers take that into account.

What does "have you tried OSM" mean? If I understand the thread, they're talking about the feasibility of "safe/annoyance-free" bike routing. Are you suggesting apps like OSMAnd? Or are you suggesting that they simply complete the map for their areas of interest?

OsmAnd is very good, but also fitness apps like strava offen use OSM data. By adding more detailed data, you could influence their routing.

I think part of the solution is to allow users to add personal weights to different roads. Obviously the shortcoming with this is it only works if you already know the area, when one of the major features of navigation systems is navigating an area you don’t know. So it doesn’t solve the general problem, but it would lets individual users massively improve their own experience with the navigation system.

If you already know the area, why would you need your nav to direct you to the alternative route instead of just driving it?

Because traffic can be very variable and there are always reckless drivers crashing their cars and blocking roads. And the geography in my area is such that once I’m on a route, I can’t really go another way without going all the way back.

Often I’ll prefer a slightly longer route over another, but if it’s much longer time wise because of bad traffic then I’ll go with my second choice. But navigation apps will almost always suggest the fastest way, even if it’s only an estimated minute faster.

Apple Maps will suggest multiple routes if they’re close in estimated time, but I’d like to weight some of them more heavily such that they might still be the preferred route even if it’s 10 minutes slower since it’s simply more pleasant safer.

Road closures or traffic jams?

My navigation apps keep up with those better than I can.

Is there a dataset of traffic accidents? If there is I'd definitely prefer navigation to take that into account!

some cities/counties have (and publish) bike friendly commuting maps.

Sort of like maps with the road B's highlighted.

don't see why this data can't make it into maps, like A B C roads for cars.

My city has this, and Road A and Road B are listed in the same category :/

I wonder if “ghost bikes” belong in a bike map. Someone died on a bike here. If you’re on a bike maybe be extra careful, maybe avoid this street next time.

Deaths are the most clear-cut data, but close-calls will be more numerous and thus less noisy. Either way it would need to be adjusted for cyclist volume.

I think this is why Google preferentially wanted to take me down a 45 MPH state route that runs downhill and has curves and no separation between the bike lane and vehicle traffic instead of the nice, calm, traffic-less series of roads in residential neighborhoods that was accessible with a quarter-mile detour. Those roads didn't have sidewalks or bike lanes marked on the roads, so they were totally unsafe for a cyclist. As opposed to the 45 MPH road, which is perfectly safe because it has a bike lane.

This is a great project.

No Google has definitely gotten progressively worse at navigating. The original satnav system in my 2004 Prius was very good at the time (probably still mostly is) - "quick" takes the presumed speed of a "major" road, and calculates with that. It led to mostly sensible routes following main roads. These were routes I would've picked myself from a map book.

Google Maps these days...just doesn't seem to do that. They've overweighted their traffic management algorithm or something, but it seems to have no notion of the idea that you can't negotiate 50kmh residential roads with a turn every block at anything close to 50kmh, and such a drive is incredibly stressful. And then will do other things like conclude that a side-road against traffic turn into a main road (so right for me, equiv to a left in the US) is something that will easily be done any time of day rather then consist of waiting for a gap in 4 lanes of traffic.

I suspect it's more than active traffic management, some of it is bound to be probing streets for which they don't have sufficient data.

I've noticed weird little detours in areas where I'm familiar with but engaged navigation merely for an ETA. To me those little detours that don't save time or avoid construction suggest I'm being used to probe those side streets. It's like being a human ICMP packet.

This was a “feature” in Waze in which you would be rewarded (with “candy”) for accepting non-standard routes. It helped them find alternative paths to recommend.

Google has since purchased Waze and now recently merged the teams.

https://www.cnn.com/2022/12/08/tech/google-waze-maps-merge/i...

It might be sacrilege but I find Apple Maps is better now.

In my city the bike lanes are almost all death traps because they only exist to rake in federal funding with no care for making them useful or safe. I avoid them whenever possible. They routed one "lane" onto a narrow sidewalk under a wide railroad bridge and had to change a city ordinance prohibiting bikes on sidewalks in that region of the city.

Do you, too, live in San Jose?

My favorite feature of bike lanes there is how highway access ramps straight up cross them -

- creating a flow of cars accelerating to highway speeds coming from behind and sideways from you, while you are in the perfect spot to be obscured by the car's A-frame.

Or perhaps how I had to move away a sign saying "SHARE THE ROAD" (with bicycles, presumably) from the bicycle path. Multiple times.

No, perhaps my favorite is how the Mayor decided to show off our bicycling infrastructure by taking a bicycle ride the first day of the year, and. . .

. . . was promptly hit by an SUV, which broke his bones[1].

If you don't live in San Jose, that's all you need to know about cycling in San Jose.

[1] https://abc7news.com/san-jose-mayor-hit-by-car-bicyclist/500...

I live in San Jose, and bike a lot, and all three of my kids bike to school, too (I accompany the 2nd grader). Everything you say is true and correct. But, San Jose is still far more bike friendly than the vast majority of American cities, both in terms of driver awareness and infrastructure. An increasing number of marked bike lanes on secondary streets every year, a formal program for road diets on dangerous (for everyone) primary surface streets, sharrows on lots of residential streets, and enough North-South MUTs to facilitate many commuters.

I live in Willow Glen and used to commute to Google in Sunnyvale. About 2mi to get to the GRT, then connect to the Bay Trail in Alviso, and about a half mile of streets to get to the campus. It's not just SJ, but the existence of the Bay Trail, the GRT, the San Tomas Aquino Trail, the Stevens Creek Trail, and even the Los Gatos Creek Trail all make a big difference for cyclists around here.

Compare to European cities San Jose definitely is bad on bike infrastructure. But if you compare to neighbor cities like Santa Clara (low bar, I know), it's pretty ok. Sam Liccardo at least actually cares about cycling around the city, and made bike infrastructure improvements, like 3rd around downtown (yes it took some time after the change to educate our bright drivers how to make right turns there correctly, but it mostly works now).

And San Jose's Department of Transportation traffic signal unit at least has an email address I can use to tell them which traffic signal failed to detect bicycle and turn green, and they actually fix them.

> the Mayor decided to show off our bicycling infrastructure by taking a bicycle ride the first day of the year, and. . . . . . was promptly hit by an SUV, which broke his bones

Your description may conflate the 2017 archival video with Mayor Sam Liccardo’s 2019 New Year’s Day bike ride.

The archival footage comes from a 2017 event touting Ford biking infrastructure, but the reporting does not say that the New Year’s Day bike ride was part of a promotional event.

Are we talking about the time the car hit the mayor on his bike in 2017, or the time the car hit a different mayor in 2019, or the more recent one in 2023?

Either way, when you ram your car into a bicycle it is known as "el san jose hola" or "The San Jose Hello" around here.

Here’s the Belgian version with a good part of the population commuting by bike and the politics by taxi:

"Health minister and deputy prime minister open car doors into busy bicycle lane."

https://www.politico.eu/article/belgian-ministers-send-cycli...

This is one area where autonomous vehicle companies can and should be contributing more to the public sphere. Much of this data already exists at a very high quality within mapping datasets to adjust priors for unseen pedestrian risk.

There simply aren't institutional mechanisms to make mapping data public though.

Are there that many (or even any) autonomous vehicles designed to drive on bike paths and/or sidewalks exclusively ?

No, but presumably there are huge datasets from Tesla, Waymo, et al. that are tracking the boundaries of the road, crosswalks, the existence, or lack, of sidewalks, pedestrian density, desire paths, and all sorts of things that aren't on the map. If they aren't actively being tracked, the raw data could be used as training data against areas with known non-car infrastructure to create a model that maps areas with no data.

Why do you think a private company attempting to beat the competition would make any of their data publicly available for other companies to benefit?

I didn't say they would be eager to do that, but I would hope they would understand why it might be a long term loss to hoard data that makes roads safer.

One answer is that they might have to in order to be allowed to operate. Since operating a private service on public infrastructure is not a right, it is perfectly reasonable to require that companies share their data especially when that data relates to safety.

Getting run over by a Tesla in an unmarked, but legal, crosswalk that Waymo knows is frequently used, and should be crossed with caution is not desirable under any reasonable model of capitalism and competition.

You can also make the case that if you want ANY autonomous driving to be commercially successful, you need to make sure that ALL autonomous driving is safe. Up to date maps of where the squishy humans are likely to be is a great way for all players to ensure that a single incident doesn't set back the entire industry. A Tesla running over a pedestrian isn't a win for Waymo.

Basically my feeling is that it is perfectly reasonable to require companies to share their data about the location of sidewalks and other infrastructure in exchange for getting to use public property. It benefits everyone, and is likely to have little to no impact on their competitiveness.

You might well ask the same about code, ML models, or testing datasets, all of which have established concepts of open access. The Waymo open dataset in particular contains some of the data being discussed.

Can you clarify the difference you think is significant here?

There are actually quite a few, they show up in the news now and again

https://futurism.com/the-byte/delivery-bot-crime-scene

https://www.bbc.com/news/uk-england-cambridgeshire-63821535

https://www.npr.org/2022/05/20/1100290734/delivery-robot-woo...

Starship delivery robots are the most notable these days since most of the others are bankrupt.

Either way, very few pedestrian spaces in SF are out of sight of driven mapping vehicles and most of those are parks that are easily mapped or interior spaces.

There is quite some research being done on autonomous wheelchairs.

For biking and walking, could accepting data from Strava users (or other places that let you download GPS tracks), let you infer where there are sidewalks and good bike routes?

Eg if you have 20 GPS traces in an area and they all turn at one point, that’s a good place to turn. Or you can assume something has a sidewalk if many people have walked there?

> Or you can assume something has a sidewalk if many people have walked there?

There's still a quality difference between a well-worn path and sidewalk. It can be a great way to find places to build new sidewalks. (There's the classic story of the University that didn't pave sidewalks in its quad until well worn paths in the grass were visible, using essentially crowd judgement/"ant hill optimization".)

https://en.wikipedia.org/wiki/Desire_path

This is where Strava is interesting because of how fine grained the data is.

You can reasonably assume that a place frequented by road bikers is an acceptable surface for electric scooters, or wheelchairs (the surface only, not necessarily an appropriate setting).

A well worn path will show many walkers and runners, but almost no road bikers.

I'm sure there's all sorts of accurate inferences that can be made from the data.

I refer to Strava and RideWithGPS heatmaps whenever planning a new cycling or running route, and they are very useful, but they still need vetting (satellite and street view mostly) since 1) people have different tolerances for safe/comfortable interactions with traffic, and 2) road race data is often mixed in (despite users being able to tag races), which frequently are on (closed) roads you wouldn’t want to be on otherwise.

Strava already sells heatmap data to municipalities for urban planning purposes, and AFAIK anyone with a free account can see the public heatmap data (though it is probably against the ToS to incorporate it in another map without paying). It’s even tagged by modality and surface type.

Mapbox presumably does this now. The issue with the real world is that you do need some high quality truth data to do any decent routing. Different users/vehicles are more or less sensitive to road conditions. If you want to provide an effective micro-mobility routing solution you have to know with certainty where the sidewalks and bike lanes are, how wide they are, how f'ed-up they are and provide directions accordingly. Waiting until 10,000 different transportation agencies decide to fix sidewalks and bike lanes on different schedules and different standards isn't gonna solve issues any time soon.

The prototype device they have near the end - looks like a raspberry pi + GPS receiver? The only possible reason I can think of for that is to get higher accuracy GPS than a smartphone, but is it that much more accurate? And is that accuracy that useful? I can think of a bunch of Bicycle User Groups here in Melbourne (AU) that would help contribute to initiatives like this (while AU is generally better than US for cycling, there's still a _long_ way to go), but I would think an app would make it much more scalable than hardware (also, are there not external GPS receivers available for phones)?

My other thought is, once there's a decent set of data like this, how is it updated? While road alignments don't change that often (so car routes don't need to change often), the small-scale details like this change much more frequently

I do not understand the mentality of blogging sites that do this, why lose visitors over simple text display?

"Your Browser Is No Longer Supported"

"To view this website and enjoy a better online experience, update your browser for free."

It appears to be made with Wix.com website builder.

https://support.wix.com/en/article/supported-browsers

Yup, I used Wix. It's the best I can do with the resources we have.

Out of curiosity, what browser are you using?

Apple only uses OSM for areas for which they have limited data. Just an FYI. The article implies that all these different map apps are using the same underlying data when they aren’t.

Fwiw Strava probably has the best data for this- you can tell how good a cycling path is from average cycling speed.

This feels like more of an "America sucks for cycling" problem than a mapping problem. I have never experienced this issue using Google Maps for cycling in the UK because roads like that don't generally exist. (And in fairness they may use high quality mapping data from OS.)

The entire world sucks for cycling.

A few urban areas, mostly in Europe, are an exception to this. I'm sure that the entire UK isn't good for cycling, are you talking about a relatively small area in a city?

Of course data matters but the routing engine is also very important, with the calibration of the weights on the edges of the graph. For cycling it feels like Organic Maps is so much better than Google Maps and Apple Maps, consistently offering safer routes which are also faster in practice because you don’t have to slow down for dangers. Google Maps is notoriously bad in rural/mountain areas, except if you are looking for mountain bikes routes…

I know this article isnt about the book but I haven't thought about it in decades.

Loved it as a child. I should have another look.

https://en.m.wikipedia.org/wiki/Where_the_Sidewalk_Ends

Why not bulk import this from your city/county/municipality GIS system?

Dunno about San Francisco city, but many will track sidewalks as part of maintenance management. Depending on how well organized they are, they might have additional features such as trees and street lighting.

I think this is a trend for the future. Mico-knowledge (tm) like this will enable things we haven't thought of. Things like this 'fill in the gaps' that make it hard to create new services and take advantage of opportunities. I hope this continues to be followed up on.

I hope to use street view and Mapillary data to see if I can improve this myself, actually, but street view is too expensive and Mapillary too sparse

If you’re editing Openstreetmap, using data from Street View is prohibited by the license, even if you pay for it I believe. However, Microsoft has licensed Bing Streetside for this use for free; the dataset isn’t as good as Street View, but is much better than Mapillary.

Thanks for that (I haven't tried streetside) - I'm not editing openstreetmap at this point but trying to find homes near decent infra.

Hi CalRobert, Your use case is indeed one we are very keen to work on. What tools have you been using to find homes near decent infra? I don't find https://www.walkscore.com/ to of much use -- but maybe I don't understand how to use it.

Hi! I sent you a message on your site’s contact form

I think the Street Complete approach is much more practical than anything voice based: slightly gamified "quest" based entry which lets you tackle an area or a data type in detail.

It's pretty much impossible to do any meaningful and accurate data entry while riding a bike, whether the interface is touch, voice or whatever. You need to be able to verify that the phone has accurately captured what you were trying to tell it, not just take it on faith that it's worked, and that means looking at the screen carefully. Good luck to them making this work reliably.

Not sure why you want a fake email / zip code in your database but okay

Especially where it's a numeric field with a spinner widget letting you go to -1 with one click off the starting page. And assuming no-one outside the US would want to have a look at this.

Even after reading the entire article, I don't understand. Mind clarifying?

If you click through to the app it requires an email and zip code to "register" and view the map. But it doesn't require actually receiving an email.

I wanted to see the map but I got hit with a registration form asking for my zip code, but I don't live in the US.

I suppose the tl;dr would go something like this: 1) we think mapping/navigation/routing should be as good for bicyclists, pedestrians, people in wheelchairs, on scooters, and micromobility generally as it is for those driving cars. 2) the reason mapping/navigation/routing is _not_ as good for micromobility as for cars is scarcity and incompleteness of map data necessary to enable it (we're using OSM to make this determination). 3) we're working on some tools to address the data problem, with an eye to building high-quality micromobility mapping/navigation/routing on top of it.

I was looking for an explanation of the comment above me, not of the article, but thanks anyway! Excellent work.

The splash page showing a GIS image of San Francisco is always a sign that the lights are on and nobody is home.

On one hand people feel compelled to put up images of San Francisco to make people think that they're with it. On the other hand if you've been to San Francisco you'll know it is where GIS applications go to die. It's where I'll walk past endless open restaurants because my coworker insists on finding one with Yelp and five of the restaurants listed on Yelp won't close. It's where the only place my handheld GPS can get a fix is on the roof of the Moscone Center. It's where my handheld GPS struggles to route, even for a car, because there is just too much geometry in too small of an area.

Go to some normal city, say St. Louis, MO or Billings, MT and you will see GIS applications "just work" in comparison.

I've done GNSS localization in SF and I can't relate to what you're saying here. It's not notably more challenging than any other dense urban environment and the public commercial maps (e.g. Google, Apple) have better data than they do in most cities. I saw more issues with Pittsburgh and Phoenix because they hit weird edge cases no one had foreseen than SF, which had the typical issues you expect to see in urban canyons. Plus, there are local GNSS companies testing in the bay area, so stuff tends to just work out of the box.

We chose SF Bay area because we live here. I just did some surveying in the greater Detroit metro so I hope to look at the data there too. However, mapping and surveying is a supremely local thing, so we started locally.

I don't understand what you are saying. Are you saying GIS, GNSS and RTK are worthless in SF?

Reply on news.ycombinator.com