Uh-oh. Generative AI is coming to Google Maps. Google is using large-language models to give suggestions on where to go based on its vast horde of reviews, ratings and other contributor data. “Starting in the U.S., this early access experiment launches this week to select Local Guides, who are some of the most active and passionate members of the Maps community. Their insights and valuable feedback will help us shape this feature so we can bring it to everyone over time.” Other LLMs have a tendency to push out magnificently wrong answers; it’ll be interesting to see what results Google will get with this specific set of data. (The chances of spectacularity are not zero.)
Category: Web Mapping
Montreal’s Interactive Construction Site Map
Montreal has launched an interactive map of its many, many construction sites. Per CBC News: “Ahuntsic-Cartierville borough Mayor Émilie Thuillier says the map will help Montrealers see in real time where a construction site is, what the reason for it is and what company is responsible for it. The map also tells users when the work began and when it’s scheduled to end.” Apparently there are problems with illegal construction barriers and abandoned traffic cones: if they’re not on the map, that will be a tell.
Mapping North Korea in OpenStreetMap
Mapping North Korea in OpenStreetMap is, by necessity, an exercise in armchair mapping—i.e., drawing maps from aerial imagery and other data sources—because on-the-ground mapping is, to say the least, impractical. French OSM user Koreller has created a North Korea mapping guide for OSM contributors.1 See Koreller’s diary entry about the guide, plus their entry about mapping the North Korean capital, Pyongyang.
What3Words Confusion Rate Under Scrutiny
The What3Words geocoding service assigns a three-word mapcode to every three-square-metre patch on the planet, the premise being that three words are easier to remember and share than longitude and latitude to the equivalent decimal places. But the main complaint about What3Words (apart from the proprietary nature of its algorithm and database) is that it’s possible to get even those three words confused, especially in contexts where plurals and homophones may not be heard clearly, or where similar combinations of words are close enough to each other that they can be mistaken for each other. There’s actually an entire website dedicated to chronicling errors in W3W.
W3W maintains that their algorithm keeps similar combinations “so far apart that an error is obvious. We also worked hard to remove homophones and near homophones like sale and sail.” They rate the the chance of two confusing combinations appearing close enough to be unclear at about 1 in 2.5 million. But in a new analysis of the algorithm, currently in preprint, computer scientist Rudy Arthur argues that despite W3W’s claims this chance of confusion is far higher, and warns against adopting W3W as critical infrastructure (it’s used by emergency services, particularly in the U.K.) without testing and comparing against available alternatives. [The Register]
Previously: What3Words Hasn’t Had the Greatest Couple of Months: A Roundup.
Engst’s Experience with Mapping Services
Adam Engst of TidBITS: “I’ve been working with mapping services a lot of late and wanted to share some of my experiences in the hope that they’ll help you boost your mapping game beyond simple navigation.” Mostly focuses on fitness-related mapping, but also on how to correct errors on online maps.
What3Words Hasn’t Had the Greatest Couple of Months: A Roundup
The proprietary geocoding system What3words, which assigns a three-word mapcode to each three-square-metre point on the planet, has been getting some grief lately. It’s always been somewhat controversial because it’s a closed system, and because of the steps What3words has taken to protect its proprietary database and algorithms: it’s issued takedown notices relating to the compatible, open-source WhatFreeWords (details here), to the point of threatening a security researcher over his tweets about it last April. Which, you know, got noticed.
It’s also been the subject of several parodies, including what3emojis (emojis instead of words), Four King Maps (four swear words, UK and Ireland only on account of a lack of swear words, which frankly disappoints me) and Maps Mania’s own April Fool’s joke for this year, what2figures, which expresses any point on the globe in just two numbers (I’ll wait).
But more recently it’s come under criticism for having similar sounding word combinations for addresses only a few miles apart: see Andrew Tierney’s blog post (which expands upon this Twitter thread) and What3words’s response. This is especially a problem for first responders trying to locate someone who may have misspoken or mistyped their location, or because of their accent, resulting in rescue teams being sent to the wrong location.
100 Million Edits to OpenStreetMap
The 100 millionth edit to OpenStreetMap was uploaded today, the OpenStreetMap Blog reports. “This milestone represents the collective contribution of nearly 1 billion features globally in the past 16+ years, by a diverse community of over 1.5 million mappers.”
OpenStreetMap’s ‘Unholy Alliance’
OpenStreetMap, says Joe Morrison, “is now at the center of an unholy alliance of the world’s largest and wealthiest technology companies. The most valuable companies in the world are treating OSM as critical infrastructure for some of the most-used software ever written.” Corporate teams, rather local mappers, are now responsible for the majority of edits to the OSM database; Morrison speculates that their participation is about “desperately avoiding the existential conflict of having to pay Google for the privilege of accessing their proprietary map data.” In the end, he argues that we’re in a strange-bedfellows situation where corporate and community interests are aligned. (To which I’d add: for now.) [MetaFilter]
Previously: OpenStreetMap at the Crossroads; OpenStreetMap ‘In Serious Trouble’.
DuckDuckGo Adds Directions to Its Search Results
Privacy-focused search engine DuckDuckGo, which added Apple Maps to its search results in early 2019, has taken the next step and added walking and driving directions to those maps. Like the maps, the directions use Apple’s MapKit JS framework. [Daring Fireball]
Fake Coronavirus Web Maps Are Spreading Malware
Hackers have created fake coronavirus map websites that install malware on users’ computers. According to Reason Security’s analysis, the websites resemble the coronavirus map dashboards produced by legitimate organizations, but prompt users to download an application: the application activates a known malicious piece of malware called AZORult, which collects browser information (cookies, browser histories, IDs and passwords). Not terribly surprising that bad actors are trying to exploit a crisis, but depressing all the same. More at Business Insider, The Hacker News and TechRadar.
Google Maps Data and the Google Maps Platform
Google Maps product director Ethan Russell has a post about their map data: how (and how often) it’s updated, how to submit updates, how it’s managed and checked for accuracy, that sort of thing. It’s one of a series of posts on the Google Maps Platform, which is (now) their maps API for businesses.
Previously: Google Maps Changes API Pricing, Competitors Respond.
Complaints about Facebook’s Automated Edits in Thailand
Facebook’s AI tool has added some 480,000 kilometres of previously unmapped roads in Thailand to OpenStreetMap, BBC News reports, but some local mappers have been complaining about the quality of those edits, and the overwriting of existing edits by Facebook’s editors: see OSM Forum threads here and here. In particular, see OSM contributor Russ McD’s rant on the Thai Visa Forum:
What Facebook fail to state is the inaccurate manner in which their AI mapping worked. The OSM community in Thailand had for years, been working slowly on mapping the Country, with the aim of producing a free to use and accurate map for any user. Information was added backed by a strong local knowledge, which resulted in a usable GPS navigation system based on OSM data. Main road were main roads, and jungle tracks were tracks.
Then along came Facebook with its unlimited resources and steamrollered a project in Thailand with scant regard for contributors … sure they paid lip service to us, with offers of collaboration, and contact emails … but in reality, all our comments went unanswered, or simply ignored.
Sure, their imagery identified roads we had not plotted, but along with that came the irrigation ditches, the tracks though rice paddies, driveways to private houses, and in once case, an airport runway! All went on the map as “residential roads”, leaving any GPS system free to route the user on a physical challenge to make it to their destination.
Local users commented, but the geeky humans who were checking the AI, living thousands of miles away, having never visited Thailand, just ignored our comments. They would soon move onto bigger and better things, while sticking this “success” down on their resume.
Sounds like another case of local mapping vs. armchair mapping and automated edits, where local mappers are swamped and discouraged by edits from elsewhere. [Florian Ledermann]
Previously: OpenStreetMap at the Crossroads.
Satellite Mode, Aerial Mode, Bird Mode
Here’s a silly Google Maps origin story about how “Satellite” was almost named “Bird Mode” pic.twitter.com/wj7CRJUEyx
— Bret Taylor (@btaylor) February 23, 2019
A lot of what we refer to on online maps as “satellite imagery” actually isn’t: the high-resolution stuff is usually taken from airplanes. This can be a point of confusion for some—and, according to this Twitter thread from Google Maps co-creator Bret Taylor, also a point of contention for the Google Maps team before it launched. Some engineers felt that calling the layer “Satellite” was factually incorrect—because of that aerial imagery—and therefore shouldn’t be used; others argued for “Satellite” based on label size and usability studies. It nearly got called “Bird Mode” as a compromise. [Boing Boing]
The Washington Post Maps the U.S.-Mexico Border
The print edition of today’s Washington Post maps the fences and walls along the U.S.-Mexico border. The online version, which I seem to have missed when it was posted in October, offers a much more detailed look: it’s an interactive, scrollable map that offers a flyover view of the border, fenced and unfenced, as it passes through farms, ranches, towns and impossibly rugged terrain between the Pacific and the Gulf of Mexico.
DuckDuckGo Now Uses Apple Maps in Search Results
Google integrates its maps into its search results: synergy! What, then, is scrappy upstart search engine DuckDuckGo, which makes a point of not tracking its users,2 to do in response? Answer: use Apple Maps. “We’re excited to announce that map and address-related searches on DuckDuckGo for mobile and desktop are now powered by Apple’s MapKit JS framework, giving you a valuable combination of mapping and privacy.”