Comment by jschrf | Hacker News Reader

Comment by jschrf | Hacker News Reader

I've read this article multiple times now and I'm having a hard time reconciling it with reality.

The app "intelligently identifies nearby smartphones, whether at a local park or at a concert, using advanced proximity algorithms".

OK, I'm listening...go on...

"Lots of people are trying to create location-based services and using GPS ... The problem with GPS is that it doesn't work."

GPS doesn't work, really? Is there more context to this quote? Maybe CEO Bill Nguyen means that GPS doesn't always work. Like, in a huge city with signals bouncing off buildings everywhere.

OK, in that case, there's always cell-tower triangulation and/or Wi-Fi positioning. Maybe that's what he meant?

The article goes on:

"Color does things differently by collecting these various data points from the phone's sensors and then looking for proximity by looking for identical inaccuracies."

This is where my eyes start to glaze over. You don't get magic quantitative inaccuracy values from sensors, sorry. Unless you have some sort of magic real-time control experiments - you don't have that information. Am I missing something here?

Even if you can guage inaccuracies in arbitrary sensor data, how does this possibly correlate to a 3-dimensional position in space and time without GPS?

I understand and appreciate the premise and value of the app here - no-effort photo albums based on proximity, and a social graph prioritized by real-life interaction - but this article triggers my B.S. sensors in a big way.

Okay, so these "sensors" come down to a lot of things. The compass, the accelerometer, the camera, the microphone, bluetooth, wifi. And the tech here is both about location as well as proximity. And actually, looking in my phone settings, it appears that Color uses GPS too. It's on in the location services part of settings on the iPhone.

I think the big takeaway is that it doesn't rely on it and it uses all of these signals to provide a far more accurate sense of location and accuracy. And I have to say, I'm not sure what about using more data points to get a more accurate reading would set off B.S. sensors.

I agree with you 100% on your definition of sensors, and I see that the app is about proximity as well as location.

What I'm curious about are how these "advanced proximity algorithms" and "sensor inaccuracies" can be used to locate co-users.

The only viable "sensor inaccuracy" example I can think of is network-based location. Mobile devices give lat/lon network location readings with error margins expressed as radii. Perhaps you could correlate users that are possibly close (e.g. users P and Q might be close, and are also experiencing a location error of X on the same cell carrier infrastructure, so let's strengthen their spatial relationship). Maybe it's another data point, but it's a pretty poor one.

Where this all falls down is with the notion that "GPS doesn't work" and implying that other non-positional sensors (e.g. accelerometers) can be used to get a location in the absence of GPS bearing.

Maybe I'm just missing an awesome, groundbreaking, (patented) aspect of their technology, or maybe there was some hyperbole involved in the article.

Anyways, sounds cool and I'll definitely try it if I get the chance.

Doesn't any app that uses the iPhone camera API to take photos need to have permission to access GPS data because the iPhone embeds GPS coordinates of where each photo was taken in the EXIF data. If it didn't it would be a back handed way of accessing a persons location w/o them realizing it.

If you turn off the location service for Color app, the app itself will explicitly tell you that it can't work without the service.

Reply on news.ycombinator.com