How Google designed its wildfire feature for Maps | Hacker News Reader

How Google designed its wildfire feature for Maps

about.google

103 pointsagomez31460 comments

All for extra tools to help in emergency situations, that said I would also like it if I could see street names without zooming to a microscopic level which could be helpful in these situations especially.

Lol. I was coming to the comments section for this. There must be some challenge here that isn't immediately obvious because I feel that no one has solved this well.

I’m confident it’s two reasons:

- too many street names make for a noisy/busy/ugly map, and we all know maps must be beautiful and nothing else

- streets don’t pay for ads, businesses do, so between the two they prefer showing the latter.

Technically now you can tap anywhere on the map once to place a pin and get more info; In practice tapping in Google Maps has become horrendous, it never does what you expect. One such example is tapping on a business near a “walkable area”: you can’t. The area will always focus instead, even if painted behind the business.

Apple Maps does this really well (at least for NYC). It's the only reason I keep it installed, to look at street names while using Google Maps

> There must be some challenge here that isn't immediately obvious

the challenge is profit maximization, this to emphasize it is not a technical challenge (earlier versions of maps were good at this)

And adding STOP and road SIGNAL signs to the maps please too. Apple Maps does it for a while.

I would be happy with a brief persistence of street names at a larger scale when zooming in. IOW, when zooming, street names to persist at the larger, zoomed in size for 400..800 ms or so, then shrinking to the typical size. Don’t know if that is possible with whatever map layers are now in use.

No thanks. Make the names visible at all times. Apple Maps is infuriating to use while driving because there is no clear indication of scale and it constantly zooms in and out. Do I need to turn now or in a few seconds or a minute? Impossible to tell at a glance. Changing scale should be done rarely. I don’t look at my cell phone every 800ms while driving.

Both Google and Apple Maps are poorly named. They are not maps at all. They are navigation services at best. Location sensitive marketing platforms at worst.

If you're willing to sacrifice other features in favor of configurable display* and better hiking trail coverage, OpenStreetMap has the kind of design that tries to show everything at once that most people find awful to look at. I've gotten used to it and to me Google Maps is just empty, I can't find anything, basically the problem you describe.

* with OsmAnd at least, the most popular app for it. The official website at osm.org is more of a demo to show what's in the database than really meant to be a gmaps replacement, though there are still a few rendering styles to choose from on the right.

I've been using OSMAnd almost exclusively for close to ten years. I would describe the app itself as rough but adequate. There are a lot of settings you can poke at, and the way you navigate through the menus of the application to poke at those settings has never been particularly streamlined, however it gets the job done for me.

The map data itself is superb though, with the single exception of having inferior (relative to gmaps) data about businesses. Many businesses aren't mapped, or don't have their open/close hours listed or up-to-date. For this sort of information, I use a web browser and search engine to look up the business's website. For gas stations or restaurants while traveling, I do it the old fashioned way. I look for signs along the side of the road using my eyes. The same way I used to do it when I first learned how to navigate using AAA road atlases. It may seem old fashioned but it's just fine for me.

In other respects, the OSM map data is very detailed and I love the aesthetics of the OSM tiles. In particular, OSM maps are far superior to gmaps in places like parks. In places where gmaps simply says "Whatever Park", OSM will have park benches, water fountains, picnic tables, etc all accurately mapped to within a meter. It's earnestly impressive and very useful. I think gmaps excels at mapping businesses because that's where the money flows, but doesn't give a single damn about useful things like water fountains because water fountains don't buy ads.

> I think gmaps excels at mapping businesses because that's where the money flows, but doesn't give a single damn about useful things like water fountains because water fountains don't buy ads.

I think it's rather the other way around: people that operate those businesses want their information to show up correctly on Google Search (or as many non-tech people call it: the internet).

It's not that Google can sell ads about those businesses so much as those businesses having a financial interest in being accurate on Google search and, which then became big due to being tied in, google maps. It seems like the very definition of using your monopoly in one market to also dominate another, but TomTom (and whatever other commercial services came before Google's map) don't seem to agree or I imagine they'd have taken action.

Meanwhile OSM sucks at this compared to benches/fountains because it's so ephemeral. People do map businesses, but it's out of date so fast, it's easy to stop bothering. Half the stores I mapped in a town near me went out of business in the past few years. Wooden unmaintained benches in a forest last comparatively longer.

#1 struggle!

So frustrating when trying to find out what a road is called and your panning around waiting for the algorithm to show you the street name.

> Yossi Matias and others from his team immediately took action and within hours launched the first emergency OneBox on Google Search. It included emergency information and a hand-drawn map of the fire’s location, which was manually updated in collaboration with a local TV station throughout the fire’s 77-hour duration.

I love this. Do the thing, and worry about doing it the right way later.

You would not believe the amount of multimillion dollar products or features started out as a Excel spreadsheets being mailed back and forth and manually input to something.

Is this really the best policy with time and safety critical situations? Wildfire maps already exist, why not link to that? Who cares if you get it right eventually if you get it wrong when it matters?

The article heavily implies there was no such map for this incident available at the time.

I found this to be one of the most annoying features ever implemented in Google Maps. I was traveling across Colorado last summer and the persistent, annoying, inaccurate, irrelevant, and useless wildfire alerts made us switch to Apple maps, which has a whole slew of legit complaints and missing features. There wasn't a single scenario where this provided any value and only stood to make get vital information harder. To make things worse, like most things from Silicon Valley, there was no way to opt out, as ego reigns supreme.

They might have been useless to you, but I'd personally err on the side of enabling it by default in the affected areas. I've relied on it in the past for updated info about available escape routes during wildfire season.

There's merit in adding a toggle, but map applications (including g maps) are already very settings-heavy. I'm sure some team is constantly trying to balance against configuration overload.

“Functions at the most basic level” seems like a low bar for making a feature opt-out.

My bar is that safety-relevant information should probably be enabled by default. Wildfire spread certainly qualifies.

Here's two examples:

During the 2020 fire season Google maps was helpful in tracking the fire spread until we eventually received evac orders via notification. Very useful.

Contrast that with the experience before they had a well-developed crisis program. I had the misfortune to be in Nice for Bastille Day 2016. Google didn't promptly warn about the terrorist attack and I had to find out from FB / friends. There was a long period of utter confusion because I had personally disabled FB notifications and no one understood the situation or extent of danger.

> I was traveling across Colorado last summer and the persistent, annoying, inaccurate, irrelevant, and useless wildfire alerts made us switch to Apple maps

In a safety-critical situation if the alerts are "persistent, annoying, inaccurate, irrelevant, and useless" are they really beneficial?

As I clarified in the original response, while they might have been "persistent, annoying, inaccurate, irrelevant and useless" to the OP, I'd personally err on the side of annoying people vs not providing important information to people who need it.

How is providing inaccurate information acceptable in a safety-critical situation? What happens if Google says one valley is impassable due to fire and the one next to it is not but their data is a few hours old and the only way out is the valley they say is on fire?

Task saturation is a thing. If I am trying to navigate an emergency situation I do not need to be inundated with useless information. I need to focus on what matters, like navigating potentially hazardous conditions. Conditions that Google may well have gotten me in to!

As discussed in the original link, they're looking at the data straight from the satellite and continually monitoring their accuracy compared to other data sources. Mitigating incorrect information was explicitly one of the design considerations and a launch requirement. Can you elaborate on specifically what you're critiquing here besides the unreasonable expectation to provide perfect information?

The claim upthread was that this data was inaccurate in Colorado to the point that it was not useful.

> if the alerts are "persistent, annoying, inaccurate, irrelevant, and useless" are they really beneficial?

Some people say the same thing about Amber Alerts, they are "annoying and useless" because it isn't their child missing.

How is sending out an inaccurate amber alert beneficial to any actual missing child or their parents? How is doing so frequently any better?

The top level comment provided a single unsubstantiated view of the wildfire data being inaccurate.

As I mentioned up-thread, I was tracking a wildfire in Washington State and was able to determine by monitoring fire radios and watching cameras that the data was pretty accurate.

Your claim is no more substantial. The claim upthread is about Colorado, not Washington.

First: As I stated above, the data was not accurate. At all. These fires had moved miles before Google caught up days later.

Second: Google maps gave delayed, conflicting, or insufficient data on the closures of roads. The ONLY correct, complete, and accurate source should be the local government and counties managing the operations. By presenting itself as an authoritative source of information, they likely put more people in danger, or hampered efforts by local governments to manage road traffic for firefighter crews.

Google repeatedly tried to send us down roads that we knew were closed or could put is in danger if a traffic jam formed. It then taunted you with an incessant, repeated, useless warning that "Cameron Peaks wildfire may delay your travel" that could not be disabled.

I wouldn't exactly be touting the poorly polished turd pile that this project was. The "willingness to help" was certainly there, but explicit consent and information feeds from local governments managing the emergencies, they only stand as a distraction and it would be best if they did nothing.

Similar experience here with the recent-ish flooding overlay. Icons were more in the way than helpful or accurate. I guess it was helpful for someone who's not from here to get an overview, similar to how I am vaguely interested in where the worst air quality is on a worldwide scale, but this overlay was shown at way more local zoom levels.

But the wildfire layer is just that, a layer. You can turn it off. It's not even on by default.

When my hometown suffered a massive wildfire a few months ago, I found Google Maps wildfire layer to be the most accurate data for where the actual fire line was. From two states away I was able to keep my father aware of the situation by phone and get him prepared to evacuate. Fortunately they were able to stop the fire a few miles away from his property.

Meanwhile, Google shut down its crisis map earlier this year[1].

So now, instead of being able to go there to see if there was any information about a developing crisis, you have to first know what's going on so you know what to search, and only then will it show you some information.

[1] https://9to5google.com/2021/02/17/google-crisis-map/

https://www.crisiscleanup.org/map is a kind-of alternative, based on private data. It lets you start from a list of recent natural disasters (in the US).

Tangentially related: there is also a loosely organised team helping to map crisis areas for getting aid and charting where people live(d) that might need help. https://wiki.openstreetmap.org/wiki/Humanitarian_OSM_Team

I highly doubt they shut it down, they probably just closed off public access

I remember going to a talk given by the developer of something similar (can't remember the name of the site, sorry, but was given at a university, so probably a charity or public sector project of some sort) and apparently there are lots of issues with making data about this type of thing publicly available because there are malevolent parties who will try and take advantage of information about such situations in certain ways. Think they said a lot more - the talk was a while ago, but it was an angle that I hadn't thought about and I recall they had to have a screening process for whom they made information available to.

(Edit: this is the talk: http://stadium.open.ac.uk/stadia/preview.php?s=29&whichevent...)

it seems like the only useful information conveyed on that view is somebody wondering where in the world there is a crisis that they can spectate on. a colour-coded map of where you can go in the world to find a problem seems like it's feeding an unhealthy way of consuming information.

google maps does an excellent job of showing crisis information where it's relevant. if you don't know enough about a crisis to google it or find it on a map, it's probably not relavant to you.

There does need to be a global crisis "weather map." I proposed such a solution back in the 2011-2012 timeframe. It needs to be a data-centric system similar to Wikipedia, and, in fact, could and should integrate with other open source, public and crowdsourced information systems like Wikipedia.

Whether the maps are provided by Google Maps or Open Street Map or any other project should matter less than there should be a publicly discoverable way to find out how current and emergent crises will affect their lives.

That's just called "the news". I don't see how having a dedicated, detailed crisis map adds to existing journalism for outside observers.

No, "the news" is not the same as a sort of "digital twin" global crisis data model. There are attempts at trying to map the news. The Ukrainian crisis fostered the creation of the Live UA Maps project:

https://liveuamap.com/

Its coverage has since expanded to include Syria, Afghanistan, Lebanon and beyond.

But mapping "the news" is insufficient. You also have the Humanitarian Open Street Map Team:

https://www.hotosm.org/

Because they sometimes need to create maps of places that have literally never been mapped before. For example, zoom in to Nigeria:

https://www.hotosm.org/where-we-work/nigeria/

There is a hell of a lot to make a global data model and map system work. And we're missing a hell of a lot pretending that "the news" will do it for us.

Again, I foresee that this is the kind of work that needs a Wikipedia-level of involvement on a global scale. There are emergent elements of it, but we're not there yet.

I’ve been getting some sort of advocacy ads using wildfire maps on Google as an example of something that pending US legislation would take away. The details eluded (the ads may need work.) But I suspect this article is tied to that. campaign.

Anyone know exactly what legislation Google is worried about?

They're talking about the "Ending Platform Monopolies Act" popularly known (in my circles anyway) as the "Yelp Welfare Act". The bill would make it effectively impossible for Google to have any features that overlap with any conceivable competitors, so if for example there was some wildfire maps service, anywhere, no matter how obscure or useless, then Google Maps would have to give it equal billing over its own wildfire maps. This is because the Act makes it an illegal conflict of interest to own a platform and "exclude from, or disadvantage, the products, services, or lines of business on the covered platform of a competing business or a business that constitutes nascent or potential competition to the covered platform operator".

Okay. But https://inciweb.nwcg.gov/ has been around for years and gets its data directly from the firefighting teams working the fires.

Yeah, as someone who watches the California fires live, Google was very late to the game with this one and their maps aren't very detailed. There are dozens of good providers of MODIS and VIIRS maps with loads of additional features (like histories) that update immediately. Google's is the most bland and slow of them all.

The feature was launched because of a wildfire in Israel. The US government does not track international fires.

But Google still seem to have implement this feature for the US first probably because the satellite data was readily available. It may take more work to find similar data for the rest of the world.

This is nice, but can Google do a better job covering the basics? Their weather widget used to display air quality (AQI) a few years ago (2018), then it suddenly disappeared one day: https://9to5google.com/2018/11/11/google-weather-missing-air...

It seems they care enough about wildfire data to make this big update for Maps, but can't do something this basic? Their headquarters is in one of the most affected regions in Caifornia by wildfires (they had an orange sky last year), and no one at Google headquarters thought that maybe showing AQI in the weather widget should be restored? (it's not even a new feature, just bring back the old AQI panel). What's the problem here, too simple for someone's promo-packet?

Looks like this is returning at least on Google Home Hub/Max: https://9to5google.com/2021/09/06/google-nest-hub-aqi/

I've switched to use PurpleAir instead. It's likely the AQI on Google Home Hub/Max is also sourcing the data from PurpleAir.

Disclaimer: I work for Google but not on anything this AQI related.

On iOS and iPadOS the weather app displays the information (you will have to scroll.down from the location).

Also shows on Maps

It’s nice I’m sure but when I saw this appear while standing in the centre of town in the rain I wondered whether this is really the most useful layer option to add? (On a more constructive note in this part of the world flood warnings would generally be more useful. )

Even better once activated it glitches out if you’re not in California and asks you to zoom out until it’s on the map.

Would be nice if I could zoom out on the entire state of California and see at least a label showing where the fire is and not force me to zoom in to see it.

This is great but why is there not a weather radar layer in Google Maps? Is it because they don't want to compete with the plethora of weather apps?

Always wondered the same. This feature would be amazing on Android Auto.

Can you describe why it would be useful in Android Auto?

Google Maps seems quite car-centric to me, and most weather doesn't affect people who are traveling via automobile. Rain would not change my plans; but freeway closures would, and the latter is already present.

Fog, rain, sleet, ice, snow, tornadoes can all impact travel plans. Not often, admittedly, but I’ve changed my route based on most of those.

Is there an alerts feature for this, similar to Ring neighbors?

FWIW the peakbagger app (android, ios) includes excellent fire and smoke layers. andrew has added these features over the past few years as fires have complicated the quest for peaks.

caltopo.com is another great tool with fire layers, useful for planning hikes in affected areas.

Reply on news.ycombinator.com