Fifteen years of Google Maps, From Slashdot to Ground Truth

Fifteen years ago today, Google Maps launched to the world.

When it was born, it was a paper atlas in living form, with no pages to turn. Instead of online mapping leader MapQuest’s printable list of directions, navigation routes were overlaid on top of the map itself. And Google Maps loaded map tiles in a Web browser without any special software so you could explore the world without refreshing, a technical feat that had never been seen before.

In 2005, nobody really knew what would come of online maps, or how they would become such a crucial aspect of daily lives in the Internet-connected world.

How Google would partner with Apple to bring online maps to their true home, smartphones, but the alliance would fall apart.

How Google Maps would have more than a billion users and become Google’s second-largest property after its search engine.

Nobody had any idea, least of all Google.

And this was only a decade ago.

On the occasion of this 10th anniversary, Re/code spoke with the people who were there at the beginning, and brought back their stories of how something that now seems so fundamental came to be.

Slashdotted

Google Maps wasn’t supposed to be unveiled until the next day, and even then it was going to be in beta — but an avid Google watcher on the Internet guessed the URL correctly and posted it to the popular discussion board Slashdot.

The Slashdot comments were glowing: “You can actually drag the map with your mouse to move the part that’s being displayed. Way cool!” “This may be the most impressive Web application I have ever seen.”

But Slashdotters had one — very justified — complaint: The map was really only drawn in for the U.S., with Canada and Mexico sketched above and below. Everything else was … ocean.

That sea of blue stood in for the brazen ridiculousness of Google’s ambition to map the world, which would only broaden as the company added satellite imagery and street-side photos. And then, it would largely divorced itself from incumbent data providers by cutting direct deals with countries, states and even cities around the world.

Practically useless

From the beginning, Google Maps pushed forward the notion of “place” on the Internet. You may quibble with how Google delineates some geopolitically contentious area, or dislike one of its interface redesigns — but modern maps are the way they are because of the scale of Google’s investment and ambition.

That said, Google wasn’t first-to-market with an online map. In fact, the company came late to mapping, after noticing that its competitors were getting ahead.

Bret Taylor was the product manager of a Google product called “Search by Location,” which was launched as part of an experimental umbrella called Google Labs in September 2003. It was perhaps Google’s first effort related to mapping.

Basically, you could put in a keyword, as well as an address or ZIP code, and Google would find Web pages that matched both.

“It was a practically useless project,” says Taylor.

The grand example for Search by Location was you were supposed to be able to search for coffee shops near Palo Alto. But Taylor remembers that Sun Microsystems put its address at the bottom of every page of its website, and it named its products after coffee (most famously, Java). So that broke the entire example.

“It had zero users per day,” says Taylor, who is now CEO of productivity startup Quip, after a stint as CTO of Facebook.

That original product was made much more accurate by licensing Yellow Pages information, but it wasn’t the dramatic leap forward that people at Google — particularly now-CEO Larry Page — were hoping to make.

So Google sought inspiration and talent from outside. Just before it went public, it made three relatively small acquisitions in 2004: Keyhole, Where2 and Zipdash. The three deals were led by Page and Megan Smith, who is now CTO of the United States (disclosure: Smith is married to but separated from Re/code co-CEO Kara Swisher).

Do me! Do me!

Keyhole was by far the biggest of the three deals, though the price was never disclosed. At the time, the three-year-old company had 30 employees and was selling an enterprise satellite map Windows application for $69.95. The key technology was a way to stitch satellite images together into a big composite map of the world, and then divide it into millions of tiles, so you could start with a higher-up view and then dive down to a specific location on earth.

There’s a funny story about the Keyhole acquisition that Smith and others recall, from a Monday meeting of then-CEO Eric Schmidt’s key management team.

Here’s how Chris Sacca — who at the time worked on Smith’s team, acquiring companies, and is now a prominent technology investor — describes it:

“I’ll never forget, we were in a meeting discussing the acquisition of Picasa, and this young guy Adrian Graham, who sort of looks like Morrissey, was going through the slides and pitching how we integrate Picasa, and [Google co-founder] Sergey [Brin] was totally distracted. And this was in building 42, in a conference room that had a stairclimber in it, because Sergey wanted to use his time better in meetings.

“He’d shown his laptop to a couple people, and people said, ‘Oh shit, do me, do me.’ And this guy doing the presentation was really starting to sweat, and Sergey eventually gets up and unplugs the projector, says ‘This thing’s cool and we should buy it,’ and he plugs his laptop into the projector and shows us Keyhole. And, literally, these executives are shouting out their addresses because they want to zoom in on their houses from space.”

Smith said she remembers Brin hijacking the meeting. But, for her, the point of the deal was expressed by then-SVP of engineering Wayne Rosing. Smith, who later helped form the Google X experimental product division, recalls Rosing saying, “If our mission is to make all the world’s information useful and accessible — then this is the real world.”

But then there was the matter of convincing Keyhole to be bought. The company was weighing the Google acquisition versus multiple sizable venture capital funding offers to stay independent.

Former Keyhole CEO John Hanke recalls being attracted to a broader vision of making maps freely accessible so they could be used for all sorts of purposes. That would be a big deal for scientists who usually paid thousands of dollars for high-resolution satellite photos — archeologists, rainforest conservationists and the like.

Also, this was April 2003. The U.S. was invading Baghdad right at the time the acquisition was being discussed.

“What if you could surf and go to Baghdad and think about humans in a much more local way? You could see how small the world was,” Smith remembers saying. “The VCs were saying, ‘Come build this huge company.’ And we were saying, it matters for all these different reasons — including peace.”

The Google board - find here - unanimously approved making an offer, and Keyhole took the deal.

Near-death experience

Where2 Technologies, the tiny startup that would be most responsible for creating Google Maps, almost died before being acquired. At the time, the online map standard was MapQuest, which was basically a way to display a list of directions with tiny squares about each turn. Jens and Lars Rasmussen, the two Danish brothers who started Where2, had an idea about making the actual map the center of the display, and letting people scan around and zoom in and out. But nobody was buying it.

“I remember early on, a lot of our detractors — and there were many — telling us, ‘This is not a good area to get into,'” Lars Rasmussen recalls. “They would talk to us about how a person only needed maps, at best, a few times a week.”

Storied venture capital firm Sequoia Capital dropped funding discussions with Where2 when Yahoo Maps launched an update that added Yellow Page entries on a map.

“They pulled out overnight completely from the deal, which felt like quite a blow, as you can imagine,” remembers Lars Rasmussen. “All the other VCs who were circled around us heard about this and also pulled out, so no one would even talk to us anymore.”

This was post-dot-com bust. Lars and his brother were totally broke. Due to visa issues, they were building Where2 in Australia with two engineers named Noel Gordon and Stephen Ma. Without funding, the startup was going under.

“I’m not saying that Sequoia made a wrong decision here,” Lars Rasmussen recalls. “Their view was that our window was going to close before we were done with our development, and so they pulled out.”

But maybe there was another option. Google didn’t have a mapping product at the time, so Sequoia helped hook up the Where2 team with Google co-founder Larry Page.

“Three days later, we were talking to Larry Page,” Lars Rasmussen says.

The little team in Australia had made a desktop app, but Page thought the future was on the Web. So, in the three weeks between meetings with him, the four engineers essentially created the modern idea of a Web application — where data was fetched in the background rather than having to be refreshed to get new data.

(A team inside Google was actually doing something very similar for the first version of Gmail at the same time, but unknown to each other, they created a pretty fundamental Web technology that was later called AJAX.)

“Being in a very heightened state of motivation — as I’m completely broke and without any other options — we scrambled and we took three weeks, worked day and night, and actually built a website specifically to impress Larry and his crew over at Google,” recalls Lars Rasmussen, who now works at Facebook. “It even had the Google logo on it. And we marked up a local search where we used lava lamps as our markers, because Google liked lava lamps.”

The third mushketeer

Way back in 2004, a little-three person startup called Zipdash — actually not even a startup, they hadn’t formally incorporated — was working on a mobile traffic application.

They were very, very early on this — too early. Zipdash was only for Nextel phones. Founder Mark Crady coughed up thousands of dollars of his own money to license traffic data from taxi fleets in the San Francisco Bay Area. From that sample, he was able to estimate traffic delays in real time, and once people started using the app, Zipdash incorporated their user activity to improve the estimates.

It was actually a lot like Waze, the peer-to-peer traffic map app Google acquired in 2013 for a billion dollars. Zipdash founder Mark Crady says that’s still a bit of a sore spot — he sold Zipdash for only two million dollars. But he did get to join Google before the IPO, which wasn’t too shabby, financially.

Crady tells a funny story from his first meeting with Megan Smith. “At one point, Megan asked us how many users we have. I said, ‘200 or 300.’ She said, ‘Thousand?’ We were definitely not Google-size at the time.”

Zipdash landed at Google, and Crady and his tiny team got to work on building what would become the mobile version of Google Maps.

By the way, those humble beginnings didn’t foretell the future. Between Where2’s Google Maps on the Web, Keyhole’s Google Earth on the desktop, little Zipdash’s Google Maps for mobile would become the biggest product of all. Today it has more than one billion users. Crady left Google in 2009, and is now working on a local event listing project.

Life at Google

           At Google, the Keyhole, Where2 and Zipdash teams were basically plopped down in Mountain View and told to make a Google version of what they’d built on their own. They were able to recruit people internally, but they had to win them over on their own. The Danish Rasmussen brothers used the enticing schtick of handing out delicious danishes.

And then they got to work. The Where2 crew shoved some Windows computers in a closet to crank out map tiles. Jens Rasmussen, who is acknowledged as the idea guy in the duo, came up with the idea of the Google Maps pin, rather than Yahoo Maps’ red star. The appeal of it was that the tiny point of a pin could show a place on a map without overlapping with it and obscuring it. (The kiss-ass lava lamps were left on the cutting room floor.)

Jens is very detail-oriented, even 10 years on. He says he tried to approximate the sense of a 3-D map with pins and drop shadows. But if you look closely, the original shadows on the pins crossed over each other and got darker. Kind of like in the middle of a shaded Venn diagram, where the overlapping section is darker than the other portions.

In real life, Jens says with a smile, “You can’t block out the same light twice.” He is keeping quiet about his current projects.

Meanwhile over in the Keyhole cluster, John Hanke recalls working on an elaborate staggered plan for acquiring more high-resolution satellite imagery, because it was expensive. He wanted to start with the cities first. He prepared a big presentation of the cost structure for Larry Page, Sergey Brin and Eric Schmidt.

At the end of it, Brin said to Hanke: “Why don’t we just do all of it?”

Hanke says he was flabbergasted, but that’s exactly what they did. It cost many millions of dollars, and Google set up a special high-bandwidth interconnect to get data from the supplier in Colorado.

But for a while, that scale of ambition wasn’t matched by the scale of interest in Google’s mapping products.

Going big

Initially, both Google Maps and Google Maps Mobile had disappointing traffic. For all the success they’ve had today, people didn’t seem to notice them at first.

When Google Maps was Slashdotted before the team intended to launch it, it got huge traffic. On that first day, it got just over 10 million map views, according to Bret Taylor, who by then had joined the Where2 team, and was Google Maps’ first product manager.

It took months — nearly a year — for the site to equal that launch-day traffic. The Google Maps team thought they had a better product than the competition, and many people agreed with them, but that didn’t mean they showed up and used the thing.

Two things led to the product taking off on its eventual ever-upward traffic trajectory: Google Maps added satellite data from Keyhole, and that trick of looking up your own home from space brought in a big new crowd. And Taylor led an effort over the next Christmas break to completely rewrite everything to make it faster. It also didn’t hurt that Google Maps released developer tools early on, so people outside the company started building on top of it and evangelizing it.

From then on, the growth didn’t stop. By the end of 2006, less than two years after launch, Google Maps was the largest maps provider in the world. Soon it was Google’s second-most trafficked site, after Google.com.

Something similar happened for Google Maps for Mobile. The first version was only available for a few phones, and it didn’t include the traffic data that had initially set Zipdash ahead. It didn’t get a lot of downloads, and even people who downloaded it didn’t use it much. The app only really took off for real when they added BlackBerry support, Crady recalls.

But even then, some people — including then-CEO Eric Schmidt — didn’t quite see the full potential, Crady says. Internally at Google, more powerful figures like Nikesh Arora got more resources for efforts to create a mobile marketplace for payments and transactions like ringtones. That no longer exists, obviously.

In January 2006, Yahoo launched a product called Yahoo Go that brought a lot of its products together: Search, news, mail, weather, traffic. Crady recalls Schmidt asking the mobile team to come up with a killer app to respond.

“From the early days, there was all this talk about, ‘You need to find the killer app.’ And here we were making all these cool features, and as far as we were concerned, this was the killer app,” Crady says. “We were working on the killer app.”

The Apple deal

Today, it’s absolutely obvious that maps is the killer app for mobile. The fact that your phone knows where it’s located means you can drive yourself somewhere new, find a nearby coffee shop, route around traffic, and hail an Uber.

And it’s not just Google Maps. For most of the last decade, there have been comparable products from Baidu, Microsoft and Yahoo, and more recently Apple. But mobile apps didn’t really exist 10 years ago.

It was a huge coup for Google Maps to be installed as the default on the iPhone, but the Apple-Google hookup was a troubled relationship from way earlier than you might think — way, way before Apple launched its own iPhone mapping app in 2012.

Before it debuted the iPhone in 2007, Apple let Google in on the secret. Apple wanted the phone to come preloaded with a mobile mapping application, so it needed Google’s help. But it didn’t trust Google to design the user interface, only to contribute data and smarts. So, under strictest code of secrecy, maps for the iPhone were built in collaboration between a group from Apple and the former Zipdash team.

The two cultures didn’t have a lot in common from the start, Mark Crady remembers, but then Apple became uncomfortable with Google’s work on Android.

Apple started making all sorts of demands, Crady says, including blocking Google from using double-tap to zoom on maps. “It really irked me,” he says. And Apple wouldn’t share iPhone user activity with Google, meaning that Google had to make its traffic estimates without including a huge swath of users.

The list of sticking points got longer. The original deal made Google the exclusive maps provider for the iPhone. When Apple decided to allow third-party apps, that had to be renegotiated. Google asked again for user activity, but Apple wouldn’t give it up.

In hindsight, Mark says, the Google-Apple mapping relationship showed that even Steve Jobs, for all his storied foresight, had no idea how big maps or apps were going to be.

(Then again — given the relationship had been so frayed for so long — Google should maybe have had a Plan B ready to go when Apple made its own Apple Maps app and pulled the plug on installing Google Maps on iPhones by default in 2012. Instead, Google was caught on its heels for a good three months.)

The view from the street

For its next big mapping act, launched in May 2007, Google again brought in outside talent. It recruited a research team from Stanford that had made a 3-D scan of Michelangelo’s David, and acquired Stanford professor Sebastian Thrun’s startup, called VuTool, which was working on imaging using a fleet of cars and off-the-shelf cameras. The outsiders were combined with an internal team that volunteered their experimental “20 percent time” to the project, which would end up being one of Google Maps’ most distinctive features: Street View.

Luc Vincent, a Google engineering director who has been at the company for more than a decade, was working on Google Books at the time and Street View when he could. Vincent recalls that in the early days, the team set up shop buying used Chevy Astrovans for something like $5,000 each.

The Street View crew filled the vans with equipment and drove around Mountain View and Palo Alto taking pictures. They drove very slowly to minimize blur. To make things worse, Vincent says, the vans had lasers mounted on them that were very visibly branded from the company that made them, which was named SICK.

It wasn’t creepy. Not at all.

Kidding aside, by this point Google Maps was attracting all sorts of privacy controversies. Years later, it would be discovered that a Google engineer had rigged Street View to suck up people’s Wi-Fi transmissions as it was driving by and send them back to Google servers. Google was internationally cited and fined over the incident.

Even in the earliest days, people were freaked out by satellite imagery that showed their own home. Now there were street-level pictures. Ultimately, Google agreed to blur license plates and faces. And Street View currently doesn’t capture images in places like Germany, over longstanding privacy concerns including the Wi-Fi incident.

But that doesn’t mean Street View has been tamped down as a project. It’s now available in 65 of Google Maps’ 200-some countries. And these days, photos taken by Street View cars actually help create and validate the underlying data behind Google Maps. Using machine-learning techniques, Google can now fairly accurately look at pictures of buildings and signs and extract street numbers and driving rules.

Ground truth

The last episode of the Google Maps origin story was a tightly held secret for many years.

In a pair of multibillion-dollar deals in 2007, TomTom bought Tele Atlas and Nokia bought Navteq. The two largest map providers were suddenly in new, and not necessarily friendly, hands. Google realized it needed to do a better job of controlling its own destiny.

A skunkworks team at Google started exploring what types of data it would need to build its own maps, who owned the data and perhaps most importantly, who would sell it. In many cases, Google needed to go to down to the city level to get the details it wanted. They called it Ground Truth.

Along the way, says Megan Quinn — who led the data acquisition project, and later the product itself — the team realized that Ground Truth would not only free it from archaic licensing agreements built for CD-ROMs and in-car navigation systems, but would also allow Google to do new things, like create biking and walking directions, that the incumbents weren’t doing.

So Google decided to go for it. “It was very deliberate,” Quinn says now. “The challenge of deciding you’re going to map the world is that you can’t ever stop. The world is ever-changing. And in some ways, this was a real departure for the company, because this is not something you’re going to test and toss.”

A team of 20 people across the world worked full-time on acquiring map data. Sebastian Thrun led an effort to build a sort of holodeck of tools and services to integrate all the datasets. A large operations group in India helped pull everything together.

When the first push was done, for American map data, Quinn sent out an email to every Google employee. She asked them to test the maps in their hometowns, their college towns, anywhere they had lived. For every bug they found, she would bake and send a chocolate-chip cookie.

“I just spent a whole weekend making cookies,” recalls Quinn, now a partner at Kleiner Perkins. “I home-made 7,000 chocolate-chip cookies.”

While the homegrown maps may not have been bug-free at the end of the effort, Quinn recalls the incident as “a rallying moment for the company.”

Epilogue

The early history of Google Maps ends there. Most of the seminal Google Maps team members have moved on, but to a person they recall working on Maps as the most fulfilling and successful project of their careers. They still take it personally when they hear of bugs in the product or complaints about misguided redesigns.

At Google, Luc Vincent remains focused on imagery, but now he’s working on things like the two satellites Google now owns through its $500 million Skybox acquisition of June 2014.

Keyhole people have outlasted just about everybody else from this era. Co-founders Hanke and Brian McClendon are still key Google executives working on Geo products, and co-founder Chikai Ohazama is an entrepreneur-in-residence at Google Ventures. Late last year the mapping team got a new chief honcho, longtime Google executive Jen Fitzpatrick, amid a larger management reorganization.

Today, Geo is one of Google’s main product divisions. Ground Truth remains an ongoing project, and Google developed tools to keep its maps updated through direct user contributions. The division continues to be acquisitive, buying Zagat and Waze and Skybox in recent years. Street View has mapped the Grand Canyon and the canals of Venice. And Google’s maps have laid the groundwork for its most ambitious project yet — self-driving cars.