Hacker News new | past | comments | ask | show | jobs | submit login
Google Maps Instant (inspired by Youtube Instant) (hartlabs.net)
78 points by vijaydev on Sept 11, 2010 | hide | past | web | favorite | 59 comments

The funny thing about the "instant" search is around the time of AJAX exploding four or five years ago, this type of "real-time" search was all the rage. Over time it disappeared, most likely because it both wastes resources and a lot of users found it annoying. From that, auto-complete suggestions was the next evolution. Now we seem to have gone full circle. As impressive as it is that they're handling way more searches than before with lightning speed, the whole thing seems really naff. The marketing video for Google Instant in particular is hilarious - "I don't even have to press Enter!"

I work for Nokia and we used to have "real-time search" on Ovi Maps and this was removed around a year ago, again due to a confusing and poor user experience. After all this buzz I saw some people on the internal communication tools ask if we have the technology internally to do instance geosearch.

Marketing is everything.

This is a classic engineer response. It's not about who does it first, it's about who does it right. We didn't have the technology to implement this properly 5 years ago. I've never seen the Ovi Maps tool you mention, but I'm willing to bet that it wasn't polished enough to make it natural for users.

Now, I'm not entirely convinced that Google Instant is polished enough either- anecdotally, I've already seen some confusion. But saying "it's all marketing" willfully ignores the lesson we've all been taught by Apple: a revolutionary feature that isn't intuitively usable is no revolutionary feature at all.

Actually, to me it seems like the real lesson most companies are picking up from Apple is to grossly overstate every little thing they do in the name of hype and buzz.

I mean, revolutionary feature? Seriously? My whole point is this revolution happened five years ago and people learnt their lesson and moved on. I'm sure a lot of developers will be asked to put this back in, but on the plus side if you're not dealing with Google scale it's a trivial thing to implement.

I think you guys are talking past each other. Your claim is that, if Google Instant is successful, it will be because of marketing/hype/some-other-non-technical-reason. His claim is that it will be because Google made the feature (which everyone agrees existed in a basic form 5 years ago) accessible to the average user.

I don't think you guys are going to be able to settle this debate without some data about real life users.

There's always a chicken-before-the-egg argument with companies like Apple. Do people love using their products because of their marketing, or is their marketing effective because people love using their products?

I think it's naive to say the former. Apple posted 3 billion dollars in profit last quarter. If all Nokia had to do to post those numbers was grossly overstate every feature in their phones (which should be easy, there are so many), they'd probably be doing it.

The thing that really bugs me about this is that personally, as a DIY techie, I actually like the power-user oriented nature of the Nokia phones a lot more than the iPhone (at least, at first, when the iPhone was hopelessly feature-barren). I probably would've really liked your real time search! But I'm not arguing about what I want from a product. I'm arguing about what the majority of people want.

I'd have to disagree that "marketing" is what accounts for the pattern of Google leading web technologies into ubiquity. I think the very precise engineering they put into making a great product, then the massive user adoption that follows, is what allows for other websites to use similar technologies with confidence. In other words, the fact that Google implements something gives that implementation both massive social proof, and user familiarity.

Granted, it's not all about engineering and a huge user base. If you replaced the word "marketing" with "buzz," I would tend to agree with you. If there's one thing that's become apparent to me recently, it's that nearly any idea, product, person, place, tv show, etc, can be validated by generating a shitstorm of buzz about it. I suspect there's some reptilian instinct that changes the way we view things if we see others accepting it, and media attention, though often a misrepresentation, does the trick. For companies like Google and Apple, buzz is almost too easy.

The concept doesn't translate for maps.

An address as entered from left-to-right goes from the very specific to the general. Therefore all the api calls and screen updating you make are completely irrelevant until I type in my city half -way through.

google maps on my android phone does this type of instant-search recommendation and i always thought it was stupid, too. i just typed in "1218" on it and it's trying to recommend me "1218 grand-saconnex, switzerland" (i live in the united states).

if it's going to do search recommendations, it should at least do a regional geo-ip/gps lookup and offer suggestions i'd probably be driving to.

That is possible, but when I was making this, I didn't want to use server-side technologies (php) because I wanted it to stay fast, which would have been necessary to determine your approximate location and get the GeoID prior to page generation.

It is easily possible, however, it just would have been soooo slow in comparison.

Then don't use php... It's possible to grab user location via javascript, the client just needs a compatible browser. I'd say judging by the people who would stumble upon your project more than half would have a browser capable of grabbing location via javascript.

I only typed in my house number and 2/3 of my road and got the QUERY_LIMIT, but the result was correct with no City or State. I do live on a road with a very unique name though.

You live on Satellite Tracking Station Road?

Well, there goes trying to hide my identity on HN...

Although if you could get people to order their search queries by most important to least important, like regular search, it would probably work fine.

It semi does, I was just typing city and country names and enjoying the instant snap to.

I typed in my home address in Pittsburgh, Pennsylvania, and it put me in Long Beach, California.

I think you're on to something.

Someone looking for a job on the Google Maps team?

He says so right under the input box. :)

I'd love a job on the Google Maps team.

Psh, Google, you need janitors? Yeah? Call me.

"Geocode was not successful for the following reason: OVER_QUERY_LIMIT"

I was half way through typing my city's name...

Adding a quarter second gap between new searches would probably avoid this.

And not making it a modal popup would prevent it locking down my browser.

I sorda forgot that in there when I made it, then it got copied over to the CDN. So changing it was a little difficult. But I did ultimately change it; it now redirects you to a different version of the page with a fresh API key and a different IP address.

While it hasn't fixed the API limit issues completely, it has helped significantly.

Really sorry about that. When I moved it to my CDN, the error happened a lot less though. I do have plans to add a buffer between searches, as well.

I too got that quite a lot.

Hi guys. Ever heard of Google Instant, Youtube Instant and Google Maps Instant?

Well, I made Google Picasa Instant. By the way, I'm looking for a job too!

... Okay, Actually I didn't but it could be possible.

I did make an Instant Google Image Search.


Hi, I'm the creator of YouTube Instant. This is pretty awesome. Congrats!

Google Maps team, at least much of it, is in Bangalore.

Like to relocate to India?

The Geo team has a whole building in Mountain View, I'm sure they'd let him work there :).

Maps engineering is vastly in California.

I think waiting for a space to be typed so you know the newest typed chunk is a usable discrete piece of data along with a small delay for when no space is typed would make this work better.

In thinking about how this compares to youtube instant, I'm not sure an autocomplete would work well at the global level without more thought and refinement.

This is really impressive. I wonder if the guy will get a job offer - he certainly seems to be a pretty quick coder.

You should give it a short time lag. Instead of it immediately trying to find out my location, make it find my location after I stop typing for 2 seconds.

This is clever. Who cares if it's copycat? I bet you if it was slickened up slightly it'd be a perfect fit on Google's stable. I think transition animations and auto-zoom would help.

And how about this for an idea: Why not split up the boxes in two, one for city and one for address? It seems this might work better with a box for general tuning and one for fine-tuning. Just a thought.

Well, cool but... first place? I wonder how many other "clones" we'll see in the next days

edit: don't want to be mean

Well, it got my house, which is in Spain, albeit after a long list of JS alerts. Nice job! Even if you can't get around the errors, at least don't inform the user via JS alerts, they steal the focus from the search box and hence are indescribably annoying.

Are the map pins ever cleared? As I continue typing a long place name, I can see the 'shadow' behind the pin getting darker and darker with each keystroke.

It's 'neat' but I don't think it's useful, per se. Possibly no more or less useful than YouTube Instant though.

It's broken

Example: 110 W 75th st, new york, ny

Compare to the correct location on google maps: http://maps.google.com/maps?q=110+W+75th+st,+new+york,+ny...


Modal popup came up about three times. There's gotta be a better way of handling that - I wasn't even typing very fast.

It wasn't your fault. Unfortunately, it's a combination of no buffer between queries and a lot of people using it at the same time.

I did remove the alert window, but it will take time to propagate to my CDN.

You get a job too I guess but you should get rid of the query limit.

Plus get an auto-zoom feature.

And clear markers from the prior keystroke.

Little too popular already...

I type "New", it shows me 3 cities on the map (New York one of them). I then continue " York" and it doesn't do anything.

I think Google's phone call might delayed a bit.

Update to v0.2 just went live

Major features: - Intelligent zoom - More efficient API usage - Markers clear with new searches

I really like this - can you make an iPhone application that does google maps instant search?

It works on iPhone and Android. :)

It does too! I have added it to my home screen but the accuracy does not seem to be that great - e.g. searching for a Roka restaurant in Canary wharf london doesn't work, postcode search in UK also doesn't seem to work

Is the "scroll up to zoom out" behaviour intended? Usually it's the other way.

I didn't modify the default events. It could perhaps be a bug with the Google Maps v3 API. I will look into it, and report a bug if necessary :) Thanks!

Whoa! It only needed four letters to pinpoint my home street in Finland.

"Geocode was not successful for the following reason: ZERO_RESULTS"

Haha, I like your try! And it works really smooth.

instant is cool no doubt.. but the proactivity becomes irritating after a while

I like how typing "Paris" went to Paris, France and then adding ", Texas" takes you to the location expected as well. I like to see tricky cases like that handled the least surprising way.

any Facebook/Twiter Instant?

btw, there are lots of good ideas about command line - reverse search as in readline, context help as in Cisco IOS or SoftIce debugger and Zsh of course.

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact