Hacker News Reader

Comment by seiji | Hacker News Reader

Someone had to know that Maps was a bad idea. A huge step backwards. They had to know.

They know exactly that. Luckily, it's a server driven service, so over time everything can magically get better. They are going to aggressively harvest user data throughout the getting better process. Their options were: launch bad and get better as fast as possible (using millions of devices for feedback) or spend a few years doing things by hand then still releasing something subpar. Rip the bandaid off now.

In every other case HN screams "more data > clever algorithms!" -- this is their way of more data. It'll be painful at first, but it will get better quickly. It has only been a week since public public release. It has been sucking in betas for months too. None of this is new, exciting, or newsworthy. Release your burdens and get back to work.

View on HN

People keep saying "this is their way of more data". Can someone explain exactly how this data is acquired? If I enter an address, and it gives me the wrong location, how does it know that it was wrong? How does it get better data?

You can tap "Report a Problem" under the page curl.

And what minuscule percent of people even notice that there is a page curl?

And what even smaller percentage of people go through the trouble of reporting an error event if they do see the curl? It feels weird to pay for a service, and then have to actively work to make it better.. (by paying, I mean buying an ios device..).

Then I understand the release and iterate strategy, and I'm sure it applies in many situations. But when you are Apple and have painfully built a reputation of releasing only great products, you may want to think more than twice before crippling your star product, even if it's just a little.

As these threads have demonstrated, I think a segment of the population is actually highly interested in accurate maps.

Weird as it sounds (not being sarcastic), I think people will actually report problems, especially in their own neighborhoods or places where they travel and know well.

The problem for Apple is, they'll get a bug report something like:

Your maps is rubbish, I couldn't find the park, where is streetview too!

They will then have to have someone look at that data, try to find the area, a park nearby, compare with whatever licensed satellite data they have, and try to redraw that section, or they have to request new data from tomtom, wait a year, and hope it is fixed. That is a huge task, and not one I'm confident Apple will ever manage to pull off. The satellite data they have put into the app is appalling so if they try to use that as a reference...

And that's for the 0.5% of people who bother to report a problem - the rest will just lower their opinion of Apple and look at other solutions next time they're shopping for a device.

If one query matches multiple results, you can observe which result users tap and sort that one to the top. Or, if the user types in a query, gets no results, then types in a similar query and selects a result, you know that the first is likely to have been a misspelling/alternate spelling. If lots of users type in a query and don't get results, then that query can be flagged for further investigation. And so on.

All of this in addition to users reporting erroneous results.

And just entering the address/location queries (whether they succeed or not, and whether the user's own satisfaction is implicit or explicit) gives Apple bulk who-queries-what-from-where data that can be manually reviewed or automated-tested in followup processes.

For example, they know how dense their geo-data is; where does the volume of queries most exceed the existing density of data? Which queries actually result in someone following instructions to their terminus, and then not requerying again for a while?

Also: while Apple doesn't have a toolbar reporting every query and click into Google Maps -- as with the Google-Bing search results dustup of February 2011 -- they could be using their own automated process to trickle queries into Google Maps APIs and highlight major discrepancies with their own answers, for engineer/data-entry attention.

using their own automated process to trickle queries into Google Maps APIs and highlight major discrepancies with their own answers

Cool idea; I wonder if it'd run afoul of copyright laws though.

Good question, I've wondered this myself.

I'm not in this domain, but they do have the location of millions of phones as they move around during the time the Maps app is open. This could solve problems like adjustment of roadway locations, finding one-way streets, identifying highly-used locations. If people take a different route than the one recommended, this is also a clue.

The location information would also provide a focus-of-attention mechanism (concentrate manual effort along oft-driven routes).

I'm not familiar enough with their privacy rules to understand what portion of this data was already available to Apple versus stuff that is only available on the server side.

Since they also own the phone OS, they could get the location of those millions of phones at all times, no need to have the maps app open.

No they cannot. That sort of "track every phone all the time" behaviour would be illegal under Data Protection Directive.

There's a button in the app to report inaccurate data on all POIs. If something is wrong, use it.

They would do well to make that more prominent especially in the short-term to better handle their growing pains here, but it is a quite solvable problem. Unfortunately for Apple, Google has invested nearly a decade and who knows how much money into Maps (they were straight-up paying cash for business listings at one point, and may still do so) - so the standard with which they're competing is extremely high.

So, use your paying customers as beta testers? Not an ideal solution.

c.f. every google product ever

Seems exactly wrong to me - Google uses its free customers as beta testers, whereas its paying customers (Google Apps users) don't get to see the products until a year later.

You don't think Android 1.0, 1.6, 2.1, and 2.2 weren't a beta test of sorts? There wasn't even a software keyboard for a year after Android showed up.

iOS maps is in the same position that AndroidOS was in a year or two after the iPhone appeared. Substandard in comparison, largely functional with big gaps, but an expectation of quick improvement.

Google does not "sell" Android to you. It's free for anyone to use. Take it up with the OEMs.

You are right they don't "sell" Android to you they enter commercial agreements with manufacturers to provide them with Android enabled with Google Services imposing significant conditions on them.

Then they sell you and your data to advertisers.

I'd rather pay for something than be sold.

You mean the ones that are explicitly labelled "beta"?

Google's 'beta' is equivalent to spiderman's broadway 'previews'. When they open to the public, start amassing huge amounts of user data, using the public for R&D, call it what you want, the public is paying.

How many Google products have you used?

How many did you pay for?

Whether monetarily or otherwise, you are paying.

"Release early, release often" is good advice, and in that sense there's nothing wrong with using customers as "beta testers". No matter how good your QA department is, there're always unanticipated issues when you release at massive scale and push something to dozens of millions of users.

But you don't just rip out something millions of people depend upon for a function as critical as safe navigation in unfamiliar areas and replace it with something that has constant, pervasive issues performing its basic operations. A custom Maps application and dataset is a good business move, but you have to make sure it's at least a minimum viable product before you do the kind of rollout that Apple has done.

When Google replaces Gmail with the shinier "Googmail" and 25%+ of mail to the new service comes back undeliverable, and there's no reasonable method to revert to plain old Gmail, you'll have a comparable quagmire. The fact is that while Apple's Maps may occasionally work sort of well, the failure rates are unacceptably high; too high for anyone to trust the program any more.

It will take years to undo the damage from this, and I fully expect a good portion of people to swap iPhone for Android as a result.

It is a minimum viable product. If Apple maps shipped on the first version of the iphone no one would be complaining about it. As it is it now it takes away features that people are were used to using.

It's not a minimum viable product. If it were, Apple wouldn't be writing this letter. Maybe it would have been viable in the past (also more widely considered viable in the past: paper atlases), but it's not viable today in this context (perhaps it could have been viable as an optional beta, instead of an irrevocable feature regression) as demonstrated by this apology letter. You don't have to write apology letters for viable stuff.

iOS is a free operating system. You are only paying for the device.

A fuzzy distinction at best. You can't get an iPhone without iOS, and you can't get iOS without an iPhone/iPod.

What a specious argument. And in other stories, blogs and comments, we hear that Apple takes 80% of the mobile industry's profits and about the high margins on iProducts and the $100B+ cash hoard that they have. But now the tone is all about how poor Apple was forced into a corner by Google and others and had to subject its users to bad maps.

Guess what? Maps is hard. Why did Nokia buy Navteq in 2008 for a whopping $8B? Why do you think Amazon recently choose to license Navteq maps from Nokia for their Kindle tablets' Maps API?

Given that Amazon makes $7 million profit a quarter and Apple makes $4M profit an hour like Gruber sneeringly likes to remind us[1], is it wrong to expect better maps from a phone that costs much more than a run of the mill cheap Android phone? What excuse is there for driving directions that direct you to drive on train tracks? [2]

[1] http://daringfireball.net/linked/2012/07/27/amzn-profit-corr...

[2] http://theamazingios6maps.tumblr.com/image/32188632467

> What excuse is there for driving directions that direct you to drive on train tracks? [2]

Judging by all the cars parked on both sides of the track, there's probably a driveable surface there too. Not uncommon.

How many people would buy an iPhone without iOS, or even with a poorly implemented iOS? Far fewer.

Apple sells an experience — the whole package. They always have. That's why Antennagate, a hardware issue, and Maps, a software issue, are equally frustrating and newsworthy. Apple proposes to remove the hardware/software distinction and deliver a magical device that just works.

Maps no longer just work for many people. And that's the problem with magic. It's very brittle. The clock strikes midnight and everyone turns back into a pumpkin, the ball is over. And so many people are frustrated, because they pay a premium for this abstraction, this magic, and it has disappeared, and that annoys them, and it scares them, because if Maps can stop working when it used to work, then so can my Phone, and my Email, and all my kids' baby pictures, and overnight, Apple's customers no longer feel in control.

I like the new maps (I live in Austin, the data seems pretty solid.) But this letter is a big deal. For the first time that I'm aware of, Apple has publicly admitted that there is no Wizard of Oz, and they're just a man behind the curtain, and would you please bear with us while we iron out the glitches?

What alternative OS can I run on the device?

The old Maps app was also coded by Apple. Is there any (legal? practical?) reason why they didn't collect usage data from that?

Completely agree. Perhaps I don't rely on the app as much as others, but I do find myself using it at least once a week. I personally haven't had any real problems since launch. But expected there to be quite a few hurdles at first. I'm an iOS user, but don't consider myself a fanboy. But I expect them to really push this and create something beautiful and useful over time, hopefully sooner than later. I'm anxiously awaiting.

Just want to point out that back-end improvements alone aren't sufficient to have Apple's Maps be on par with Google Maps. Adding something like Street View to Maps will require non-trivial updates to the iOS app (not to mention the colossal task of gathering and processing the street view data).

There's no indication they'll ever have something like Street View -- or rather, what's "like" it is the 3D flyover mode.

In practice the back end is really the main thing that they need to worry about. I'd argue, in fact, that the first order of business is straightening out the point of interest database, which seems to be where the majority of genuine problems are coming from. The second order of business is ensuring cartographic completeness and correctness, i.e., making sure that roads (and even the occasional town) are actually present and in the right place. Things like fuzzy terrain tiles or bad 3D meshes may be the most fun to take screen captures of for point-and-laugh purposes, but it's far more important to me that the maps application can tell me how to get to the Brooklyn Bridge than whether it can render a pretty 3D model of it.

>this is their way of more data. It'll be painful at first, but it will get better quickly

Reporting mapping errors on a small device is painful and cumbersome, especially if you're outdoors or trying to get somewhere. Like Google, Bing or Nokia maps, can you access and report errors in the maps from a laptop or desktop? If not, this process is going to be way slower.

Reply on news.ycombinator.com