Android News

Sherpa developer talks Android, privacy concerns, and mobile future

August 11, 2009 | by Andrew Kameka

Android News, Featured post

Sherpa developer talks Android, privacy concerns, and mobile future

Few apps have hit the Android Market with as much buzz as Sherpa. The location-awareness, point-of-interest finding, user behavior learning app has become one of the most talked-about Android offerings since it debuted on the U.S. market last week. had a chance to discuss Sherpa with Rahul Sonnad, Founder & CEO of Geodelic, the small startup that created Sherpa. Below are Sonnad’s comments to address privacy concerns, Sherpa’s usefulness, and issues affecting the app’s availability outside the United States.

Androinica: How was Sherpa conceived?

Rahul Sonnad: We really felt now that mobile phones have accurate location awareness, the consumer scenario is really powerful. If you know where you are, you should be able to come up with information that is really relevant. If you’re in front of a restaurant, you may want a review of that place. If you’re at a beach, maybe you want surfing information. I had a laptop with an EVDO modem and a little GPS and we just figured that you could really build the software to make the process of connecting internet information with locations very well.

What made you incorporate the carousel?

We looked at various [methods] for showing information like on a map or augmented reality interfaces. We decided the carousel was a very appropriate interface because it’s a little bit forgiving in that you can have an arbitrary number of items in the carousel; whether they’re spaced in a very tight area or across a wide area, you still see them all. The problem with a map is that it’s so literal that often you get crowded out or things are too small. We liked the carousel and find that it’s very conducive to letting the user take the last 50 or 100 feet and skewer it to what they really want to focus on.

One of the things that caught my attention about Sherpa was that it reminded me of the scene in Minority Report where Tom Cruise gets virtual ads target specifically for him. Obviously Sherpa isn’t that detailed, but exactly how targeted/relevant are the offers in Sherpa?

Right now, we’re splitting things out by content and then advertisement. At this point, we have an almost exclusive focus on getting you the right content. If you’re in front a Blockbuster, you’ll get movie reviews and information, so that content is targeted based on your current location. On the advertising side, we haven’t really focused heavily on it, but I think the idea is that you will get targeted offers and those offers may or not be relevant to your location. They are going to be contextual to you and what you care about. If you’re a business traveler and are heading to your hotel downtown, you may still get the same type of the targeted ads shown at the airport because we know you’re in that demographic…That is very easy and will open up as soon as the scale of this type of advertising gets large enough, which is probably towards the end of next year.

When I made the Minority report comparison, a few people expressed concern that the adaptive learning might be too intrusive or that it would reveal information to marketers. What would you say to people who fear that the learning feature is more about business than user relevance?

I think that’s a valid concern and the industry needs to deal with this as we move. But when you use Sherpa, we really don’t know who you are. We use a random number, but we don’t know your number or your e-mail, so we feel that affords you a large amount of privacy. Then, you can do things like turn off the adaptive learning if you don’t want it.

There’s a tradeoff in trying to do something useful for the user and making sure you don’t get intrusive. My feeling is that it comes down to what is your goal as a company or product? Is it to create a great user experience or optimize short term dollars? In our case as a [venture capital-funded] company, we have a little bit of luxury to create a great experience and not be intrusive. Our goal on the advertising is to figure things out slowly, keep the user’s best interest in mind, and give people as much transparency and control as we can.


How long does it usually take for Sherpa’s “Adaptive Learning” feature to offer suggestions?

Well, there’s a few ways the learning feature works. It’s basically what you do, what you say, and where you go. If you explicitly say “thumbs up” or “thumbs down” on a point of interest info page, then it’s going to immediately start taking that into effect, especially with things like chains. So if you say you like Subway or Wells Fargo but don’t like Chain X, you’ll see that happen immediately.

On the other hand, if you’re just going about your activities without making any explicit selections, that takes a little bit longer. The challenge we have there is that it’s kind of like the Amazon shopping challenge: if you do one search for a baby show item, you don’t want to be recommended diapers for the rest of the year. In terms of looking at both search behavior and POI’s you’re digging into and pulling up detailed information about, each time you do that, you’ll get a very minor increase in the priority of that POI or category. That takes a little while to determine what you’re digging through is something you’ll care about in the future and not just something you’re standing in front of at the time.

When do you think users in other countries like Canada or the United Kingdom will be able to use Sherpa?

The challenge with Sherpa is that it tells you everything nearby you and gives you all the standard information about those places. In order to do that, we need to know where everything is and where to get the content. That is very challenging and we’ve only been able to do that in the U.S. I’m not sure when we’ll do that in other countries.

The other capability that Sherpa offers — and you’ll start seeing this later this year in the U.S. — people will be able to author out their own location. If I were a hotel, zoo, or theme park, I could create a very customized experience around my specific location. I might walk into a zoo and see the monkey cage and get information on monkey behavior or [habitats]. Those types of experiences are much easier for us to allow anywhere in the world because we don’t need to get all the data. It’s a small concentrated area.

For the first experience where it knows everything in the area like in the U.S., that may be a little bit longer. That’s going to be country by country where we need to find good partners and figure out who can manage the service. You might expect some pilot projects in countries, but wide scale will be a little bit later.

Sherpa does not perform as well on the G1 as it does on the myTouch3g. What are some challenges to developing for Android since the platform splinters into multiple devices?

That’s a great question and I think what you’re alluding to is potentially an issue moving forward. Right now, it’s very manageable where you’ve got the G1 and the myTouch. There are some slightly different processors, but we haven’t found to be a huge issue. I think you may find in the future where you’re on a device with a smaller screen, our carousel may not be the most appropriate view, and the memory allocated to our application may force us to change things around. It’s challenging to develop apps like that and I imagine it will take us a little while to figure out how to make the best experience for every user. Initially, we really optimized it around the myTouch and tested to make sure it works on the G1 also. I think as you look into next year and beyond that, there are a lot of things to figure out about the Android platform and how it’s used.

You’ve said that GPS and smartphone technology is leading to a new global web. What do you think apps like Sherpa are going to offer to people in the future?

The basic proposition is pretty simple: as soon as the device knows where it is, the idea of information changes. You won’t need to search to have a pretty good idea of what people want to know about. For example, if I walk into the Hyatt hotel, it’s likely that I want to have access to restaurants to hotels, specials, gym classes, happy hour, and all of this information is something you can push to a location-enabled device. We think in 2 or 3 more years, almost every phone is going to automatically do that. If you look at the dominant amount of mobile activity, it’s not going to be the standard web browsing search. It’ll be that I’m right here; automatically tell me what I need to know with no effort on my part…We think very quickly that people will realize that if they have a physical presence at a business, they are going to want to have a mobile experience at that location. At Geodelic, we’re focused on building apps and tools that will enhance that experience.

Rahul Sonnad is Founder and CEO of Geodelic, the company that created Sherpa. For more information on Sherpa, watch’s video review and visit the