Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Bump (tbray.org)
112 points by wglb on April 22, 2010 | hide | past | favorite | 30 comments


The first time I ever used Bump, I "bumped" in the air (just moved the phone back and forth). It asked me if I wanted to exchange contact info with someone who I had never met; I tried it again and it asked me to accept contact exchange with another unknown person.

All of my attempts since had no effect. Granted, I haven't tried to bump that many times.

My theory is that they use a GPS location to determine where you are. Then when the phone senses a particular type of motion, it considers that a "bump". The server looks to see if anyone else is "bumping" in that GPS location, and, if so, they connect and exchange contact info or whatever.

If this is true, it explains my original success with bumping into strangers: these two people were probably in a nearby house trying to bump! The server thought they were close to my GPS location, and connected me to the stranger.


Second thoughts and notes: 1) Clearly, bluetooth has its place. But, Bump is for the rest of us who'd rather get things done quickly. Ask a random iPhone user how they'd share a file or contact with you using Bluetooth. But Bump is easier to explain and it works fine. The bump is a mere ritual for explanation sake and coordinating accelerometer data and is probably not meant to be phone shattering (And seriously, why did anyone pay $600 for a phone these days?)

2) No, the bumping together seems to be perhaps calculating two similar accelerometer signatures within a certain probability (of course location seems to be their secondary heuristic).

3) I don't see why this should this insecure. They are probably using Diffie-hellman for some pretty badass security measure - http://en.wikipedia.org/wiki/Diffie%E2%80%93Hellman_key_exch...

4) To those who don't think this is not going to make any money on the "business" side, maybe you should tell Sequoia, one of Bump's funders, that.


If you look at Apple's recent earnings report or listen to the conference call, you'll find that the ASP for the iPhone was exactly $600. This does not include any additional carrier payments.


True, but I mean at a consumer level :-)


This is the kind of technology I like; a simple effect, instantly understandable, that requires rocket science under the covers to pull off.

Historically, most "rocket science" grade technical implementations in software/Web fail, except for stuff by NASA (and the likes). Keep it simple under the hood, but no simpler than what's absolutely required.


I think dropbox is an app that is instantly understandable and takes quite a bit of work to get functioning seamlessly.

The big thing isn't the literal "rocket science" part, just the "easy" part where you don't realize that the problem might have been difficult at all.


Perhaps I'm old fashioned, but what's wrong with using Bluetooth? The last thing I want to do with a $600 phone is smack it into somebody else's $600 phone just to transfer some pictures.


Turn Bluetooth on. Set one device to discoverable. Send contact. Wait for approval. Enter PIN. Wait for transfer. Accept transfer.

Compare to: lightly tap your phones together.


More accurately, based on the Bump FAQ[1]:

Bump your phones together. Bump them together again. Wait for both to connect to the central Bump server, potentially failing to do so. Approve the transfer. Accept the transfer.

Now, lets assume the same amount of development effort goes into writing a proper interface for Bluetooth transfers. It could be something like:

Put two phones near each other. Drag an image/contact/money onto the "Bob's Phone" on my screen. Bob drags it from his inbox to wherever he wants it.

That seems easier to me: it doesn't involve third parties[2], complex bump-matching heuristics, or physical contact.

[1] http://bu.mp/faq

[2] Even if Bump's servers have 100% uptime, how do you know that the local cell towers will have similar records? What if you're in San Francisco or New York, which have notoriously poor iPhone reception?


This is a pretty loaded example. You're using the absolute worst case scenario as the use case for bumping (requiring two bumps and that their server is down). 99 times out of 100, you only need a single bump, and their uptime has been great so far!


"the local cell towers will have similar records" I didn't see anything about cell towers keeping information in the faq and why would they. Bump manages all the handshake (and probably uses something like diffie-hellman for security) and the location information seems like a secondary heuristic.

I am sure Bump uses more than mere cell phone reception, for iPod Touch is supported, which means wifi must/can be involved as well.

In a perfect world, yes, Bluetooth would do what it's meant to do, but in the real world people like to bump into each other :-)


Was my choice of words poor? I didn't mean to imply that Bump or cell towers are retaining information. Rather, I don't trust cell phone companies to maintain their infrastructure more than absolutely necessary.


While I appreciate your sense of doubt about the cell companies, I personally wouldn't go as far without proper data on their high availability. Again, I am sure Bump has thought about this extensively enough to get some money from Sequoia.


I think this is an excellent point. Bump is (IMO) an example of too much dependence on the "magic cloud" and an odd combination of over and under engineering. Data shouldn't have to travel to a datacenter and back to move 3 inches from one phone to another, although I can see how it's easier to leverage a cloud server for some of the functions instead of making a 100% stand-alone app.


I agree. I went to their website and saw how the technology works, and it seems really iffy. The way they recognise that two phones were bumped into each other is by having each phone contact a central server and having the central server compare the timing of the two events as well as some rough location information.

So this means that you have to send your data through their central servers. Yes, it is encrypted, etc. but I would feel much safer if it were encrypted and sent locally instead.

Also I am sure as it gets popular there will be issues with timing, such as different people bumping at the same time, or the bumps being mistimed, due to differences in the way the wireless networks of two different providers time stamp the various notifications.

So it seems that while the act of bumping may give an impression of security (i.e., that you know for sure who you are sending your data to), this impression is illusory which means you will have to authenticate the recipient anyways, so you might as well use bluetooth.

And as somebody else noted if they devoted the same efforts to bluetooth, I am sure they could make their app just as nice. And they can also use bumping to initiate bluetooth transfer too.

It seems that they are eerily insistent on transferring the data through their servers.


Bluetooth connects devices that are close to each other, but your phone’s Bluetooth driver can’t tell the difference between a device that is touching the phone and a device that is ten feet away. So in order to establish a Bluetooth connection between two devices, they need to go through a mutual authentication handshake before anything else happens. You wouldn’t want some stranger on the bus to accidentally get a message on his phone saying hristov wants to send you $100, right?

I think there are some folks at the MIT Media Lab who were experimenting with using touch as a way to establish relationships between devices (e.g., you touch a light-switch panel to a lamp, then stick it anywhere on the wall, and that switch will control the lamp it was bound to), but obviously current cellphone technology doesn’t support that.

The alternative to mutual authentication is to have a trusted third party, and that’s what the Bump server provides. Using the server to create the illusion that information has passed from phone to phone by osmosis is, IMHO, a neat hack.


It seems that they are eerily insistent on transferring the data through their servers.

Judging from their job board, it's because they plan to mine the data.

http://bu.mp/jobs#data


To give them the benefit of the doubt, I don't think it's the user's transferred data they are talking about:

"At Bump we have a diverse collection of data sets: network performance, user behavior, social network data, and many types of handset data [...]"


We do not log any information that people exchange. The data job is to better understand usage/users and to improve our matching algorithms.

http://bu.mp/privacy


That's disappointing, I had expected them to analyze the vibrations of the bump on the two devices. Maybe that would be too much black magic after all.


Bluetooth doesn't work very well or reliably between two of the same phone (such as iPhone to iPhone). And lots of people have Bluetooth turned off to save battery (you can't turn it on programmatically, users have to leave your app and goto settings). And it definitely won't work well between any two different phones (such as iPhone to Android).

The official recommended bumping techniques is hold phones, and bump hands. Like a fist bump. That eliminates the risk of damaging your phone and actually works better.

The long-term goal is to bump more than contacts and pictures. For instance you can bump money using the new PayPal app on iPhone.


you can't turn it on programmatically, users have to leave your app and goto settings

Uh, on Android? False:

http://developer.android.com/reference/android/bluetooth/Blu...


Yes that statement was about iPhone OS devices which are used by the majority of our users.


> And it definitely won't work well between any two different phones (such as iPhone to Android).

Interesting -- could you elaborate more on this? I've never had a problem getting two Bluetooth devices to communicate. If my old laptop, older cell phone, and modern headset can all interoperate easily, I don't see why two months-old phones from the world's leading technology companies won't work together.


I never had trouble connecting my Nokia to any other phone, including Samsungs, Sony Ericssons, etc. Bluetooth has been nothing but easy to use and reliable. Couple of UI things on some phones could be improved, but I do not see a reason for switching technologies. IR to bluetooth has been a vast improvement, this bump thing seems ridiculous.


> And lots of people have Bluetooth turned off to save battery

As opposed to GPS?


Location services has three sources: cell, scanned wifi, and GPS. Not having GPS does not mean not having location. In many cases we do not get a GPS lock because it can take a loooong time.


You're seriously scared about destroying your phone by lightly tapping fists?


Ah, exchange personal information with somebody in immediate proximity to you...oh, but through a central server. Wait, what?!

I'm sure the central server is and always will be benevolent and benign regarding the knowledge it gains about its users, their associates, and the information they convey.


I like this idea a lot, though I'm not sure there is any possibility of making money on the business side.

I think I might have to code up an implementation of this to see how well the vibrations between phones match.




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

Search: