Seeing Forests

There’s been a lot of discussion about the automatic tagging of images. While I think the technology has improved greatly, it’s still pretty limited when it comes to really detailed tagging. That’s why we use a crowd-sourced approach. Because we first need to figure out if an image is a room in a house (rather than a picture of kangaroos), then we need to figure out what room in the house for the image (kitchen, bedroom, living room), then we need to figure out what’s in the image (along with highly subjective things like style), we don’t think an automatic solution can come anywhere near what we can do with a combination of computers and people. For this image, we can tag it as “Kitchen, Modern, Oriental, Exposed Eaves, Pendant Lighting, Range Hood, Splashback, Side-by-Side Refrigerator, Sculpture, Floorboards, Breakfast Bar, Flat Panel Cabinetry, Hardwood, Stainless Steel, Frosted Glass, Reds, Whites, Browns, Silvers”. Now, we might only get five of those tags but even so, I submit that there exists no software now, nor in the immediate future, that can come close to tagging this photo with these elements without human judgement.

Tags: No Comments.

The New York TImes article, In a Breathtaking First, NASA’s Voyager 1 Exits the Solar System was virtual blast from the past, being launched the same year I graduated from high school. I honestly don’t remember it as back then we had done the moon, had space shuttles whizzing about and pretty much assumed we’d be on Mars by now. I do remember learning to program a computer in Assembly, doing so with punch cards for grins, and knowing which end of a double byte register was up. I thought it was quaint but I learned discipline in programming (like numbering your cards so when you dropped them you could put them back in order). Was reminded of all of this in the article when it mentioned that they had to find Lawrence J. Zottarelli, a 77 year old engineer to program the 8-track recorders on the craft. The reason given was, as Suzanne R. Dodd, Voyager project manager said, “These younger engineers can write a lot of sloppy code, and it doesn’t matter, but here, with very limited capacity, you have to be extremely precise and have a real strategy”. It’s that first clause after the assertion that “younger engineers can write a lot of sloppy code”, that I’ve been thinking about: it doesn’t matter. I studied Computer Science. For the most part, computer programming is no longer a discipline but an avocation. Cutting and pasting Wordpress plugins is enough as it just doesn’t matter. I think the good news is that pretty soon that’s not going to matter anymore either. Software will soon just start writing itself.

Tags: No Comments.

Using the new Google Maps on iPhone and iPad. Reported this business, Murder by the Book as closed. It’s been closed for over a year. There was a fire. It’s been gutted. Here are photos. Reported this to Google and was ignored. Facebook doesn’t list it. I’ve never been a big Facebook user but it’s worth speculating that they might have a leg up on having more relevant data over the long term.

Tags: No Comments.

I have an almost terminal obsession with local discovery. I say almost in a denying reality kind of way. I may be doomed. I want to solve it myself but at the end of the day I just want it solved. It seems like I may as well be talking flying cars.

The new personalized Google Maps on the iPhone just breaks my heart. With all the money, with all the data, with all people, it ends up looking like yet another directory. Eat, drink, shop. 2, 3, 4 miles away. Just like so many apps before. Sure, it remembers what I searched for and may soon use that info more but my heart is broken now. Seriously? That’s the best to be done?

On the other hand I am heartened by Facebook local. Just showing nearby businesses that are actually nearby a great is a great start. Why I might want to see a nearby brokerage firm is questionable but rendered somewhat moot by the bulk of other more relevant businesses. And their use of small brand images is a qualitative differentiator. For me, my friends aren’t as relevant to addressing my local needs but accessing my demographic will be. Think I’ll be paying more attention to Facebook for local than Google…

Tags: No Comments.

I just did a cursory review of some of the posts for local products and services in my neighborhood on NextDoor.com.

• Plank & Pearl - Curated Vintage for the Home
• 40% off TODAY Massage Therapy at Body Mind Energy Healing Center
• Recommendation for AC/furnace install
• Where to get art framed?
• Audi Mechanic Recommendations?
• Leak Detection
• Need Handyman with Extension Ladder
• WTB duffle bag
• Gutter cleaning recommendation
• Arborist
• Who did your tuck pointing?
• Finished Basement Flooded Last Night
• Roof gutters
• Check out Brews on Broadway
• Good Carpet Cleaning Service
• Electrician

Mix of small retail shops but healthy representation of services - roofing, electrician, plumbing. Some local business actively (and shamelessly) plugging themselves, some being promoted by “friends”, others don’t even know they’re being recommended. No attempt seems to be made on behalf of NextDoor to capitalize or regulate. Reached out to see how an independent might get involved but no response. Interesting that you have all of these ventures chasing such leads but you have one venture that has all these leads and it doesn’t capitalize on them it seems.

Tags: No Comments.

Last week’s Street Fight Summit West 2013 in San Francisco had some great sessions (particularly Mark Canon’s Round Table on Impact of Mobile Retail Devices on Local Commerce with Leonard Speiser from Clover) I was mystified by the Next Generation of Hyperlocal session featuring 500 Startups companies. While there seemed to be a story surrounding LocBox, the CEO, Saumil Mehta, presented how to hire software engineers - not a service to hire them, just an admittedly useful methodology for hiring them in a crowded marketplace. Wanted to know more about how to “Empower Local Businesses” as the LocBox web site. Others seemed to show some very little familiarity with any history of Local, seemingly making up figures to support their positions, and otherwise showing that a good business plan is best presented with bluster. Again, useful information but seemingly off-topic.

Tags: No Comments.

It seems like the new discovery capability in Google Maps is still just based on search results and some basic categories. The example given was trying to figure out what to do with some kids on the weekend. Clicking on a museum that mentions kids somewhere in related content and then having a new map presented with a bunch of other museums that mention kids somewhere in their content is a step in the right direction but it’s still just text matching with some basic categorization of points of interest. Another example starts with clicking on a sushi restaurant and then shows all of the other restaurants with “Sushi” in the name nearby. Highlighting those who have been reviewed by friends helps only if you have friends you trust know anything about sushi. And the other example of “personalizing” the map by highlighting all of the places I’ve already been doesn’t seem to help me discover anything new, merely reminds me of how I don’t get out enough.

Tags: No Comments.

A recent article on Techcrunch announcing that AroundMe has passed the 6 million user mark, prompted me to take another look at this application. I thought I’d try to understand why this app may be popular and, while I was at it, take a look at some other major local discovery apps. I thought I’d propose the following use case. Suppose you’ve come down to the lobby of your new hotel first thing in the morning and you’re hungry but you’re in a hurry. What do you do? (Let’s say asking the concierge is not an option). Glancing out on the street reveals nothing obvious so you whip out your smartphone. How do you “discover” something right next door? You don’t necessarily know exactly what you want. You’d like something new or surprising or different. So, you whip out your smartphone and consider the options. (In these examples I just searched around my house, so experiences may vary.)

FourSquare will now give you a bunch of options with their “Explore” feature. Although their first selection in Explore always seems to be Walgreens (presumably paid inclusion) it starts asking you if you’re up for a meal or maybe some coffee or just want to go shopping. At least the UI is trying to engage you in a conversation, attempting to offer you a wide range of options at least tangentially focused on determining your needs. At the top of the list is their “Top Picks”, invariably restaurants, but at least there’s a chance you’re going to want what a lot of other people already had. I do appreciate some of the time-sensitivity they display (such as asking about taking a coffee break). And the photos really add to the experience but they’re not always organized well and you’ll end up doing a lot of scrolling. You can search, of course, but then again, remembering the use case, you can’t really say you’re hungry and the term “food” is just a category synonym for restaurants and bars.

Yelp and their “Nearby” functionality cuts right to the chase: Restaurants, Bars, Coffee & Tea and other assorted top categories. They have an “Everything” option but it is really not much help, mixing as it does Auto Repair garages with nearby Insurance Agents. Leaving aside the “Hot New Businesses” option (for another article perhaps), the “More Categories” offers a general “Food” category but again, this is more a collection of categories for Restaurants and Coffee Shops - no results for either the 7 Eleven or Whole Foods around the corner (although there is a separate category for Convenience Stores under the Food Category - they just don’t show up in “All Food”). Their augmented reality version called “Monocle” seduces but is limited to the well-reviewed restaurants and in doing so ends up showing results miles away. In fact, their default radius of search results seems to be about the widest of all the options, requiring a great deal of work trying to zoom in.

Where makes a promising start with “Eat & Drink” but we’re driven yet again to another mega “Restaurants and Coffee Shops” category. One of the more concerning things about Where is the quality of the listings themselves. Multiple results for the same business are accompanied by results for closed businesses as well - even businesses that were closed before the other business that was closed at the same location are returned. As I alluded to with the “Hot New Businesses” option for Yelp, managing new and closed businesses is both an opportunity and a challenge for Local Search and Discovery applications. Purging closed businesses would seem to be essential for sustaining credibility. It’s probably just quite noticeable around me as there have been some rapid closings and opening of new restaurants.

AroundMe, like Yelp, cuts right to the chase again, providing links to Restaurants, Bars, and Coffee Shops from the main screen. Here, some of the differences start to emerge between AroundMe and the other apps. AroundMe has aggregated many more “local” feeds than the other major local apps, singling out things like Parking, ATMs, and Movies. In addition, the default search radius seems to be more in keeping with walking than driving that other apps assume. It may not be much but these little things might just add up to why AroundMe seems to be doing well - a focus on a simple user experience, a broader set of local categories, and more immediately actionable result sets.

UrbanSpoon, that darling of shaking roulette UI design, drives down better than most on addressing this use case, which makes sense given their focus on food. Fortunately, you don’t have to scroll too far through their magazine layout to get to a “Breakfast/Brunch” option (thought I wouldn’t exactly term that a “cuisine”). You’ll be rewarded with results much more tuned to your needs than the rest but still it assumes you’re gonna want to sit down to eat. And they can be forgiven for not including results for fruit at Whole Foods because their proposition is clear from the beginning - they’re all about the sit down. Plus, UrbanSpoon gives you additional ways to slice and dice your result set, such as by cuisine or food “type”.

It seems you can group almost all of the Yellow Pages derived apps (YellowPages, YellowBook, YPMobile, and SuperPages) together as they are all remarkably similar. Each tries to extend their offering like AroundMe with multiple content sources (well, at least with Specials and Coupons). All take a pretty programmatic approach to UI design with an emphasis on search. Surprisingly, all of them seemed a bit challenged in the listing quality as they can’t quite seem to leverage the power of something like Google or the proprietary and specialized content from Yelp, FourSquare and UrbanSpoon. AroundMe is similarly challenged but the tradeoff just might be overall simplicity of use.

You can look at the competitive offerings in a number of different ways. From the analysis I conducted here, I put together a local mobile product positioning diagram based in two dimensions. Vertically, I positioned the offerings with respect to their content sources. Those offerings closer to the top attempted to differentiate themselves with more proprietary and/or specialized content sources; those at the bottom attempted to do so by offering a wider range of content that might be more generically available. Horizontally, I positioned the offerings with respect to their user interface experience. Those offerings towards the left took a more programmatic, search-based approach; those towards the right took a more browsing, discovery-based approach. The relative positioning of each offering is meant to be more illustrative than definitive.

One site that I think best illustrates an approach that helps the discovery of local products with the use of specialized content sources is TheFind. I searched for something specific on the site (knives) and was presented with a list of results that showed availability at Office Depot (office supply company), Sports Authority (sporting goods company), and Ace Hardware (a hardware supply company). While not quite responding to a simple need, like I’m Hungry, and having a list of breakfast choices returned, it does show how products can break out of the “category jails” they are in for most local discovery applications. This is a step in the direction of having an Internet of Things where the products you might be interested in almost advertise their own locations. A long way from there but the direction seems clear. Consumers will ultimately connect directly to the things that satisfy their needs regardless of the kind of company that just happens to be providing them around you.

Tags: No Comments.

We all lament the problem with local data: its inaccuracies, non-existence, and irrelevance. We all dream of Deep Content shimmering below the surface of Big Data. Companies like SinglePlatform are collecting full menus of businesses like restaurants and spas (and end up being worth 8 figures). Many applications are trying to “get the goods” on the offerings of different businesses through various means, like taking photographs of your favorite dishes at a restaurant (lobster cavatelli with ricotta gnocchi). When you’re in a hurry or driving by at 60mph, though, you don’t always have time to do this perusal though. So, we have this problem where on the one end of the spectrum, we don’t even have a business listing and at the other end of the spectrum for select business listings we have more information than we can reasonably use in some critical cases.


We certainly have to continuously work on getting the baseline data complete and correct - business name, address, main phone number. After that, we start throwing businesses in category jails: Italian Restaurant, Pizza Parlor, Bakery. This has value but depending on how this data is used it can quickly become meaningless - a map of 10 Pizza Parlors around you is still pretty generic (and can miss the great pizza in the Italian Restaurant).


So, I’m advocating for soliciting what I call “trigger terms” - those key phrases that summarize a business so succinctly that we can rely upon a person to “fill in” the rest. Terms that trigger almost a complete picture of what a place provides without necessarily picking through the indexed regurgitation of an entire menu. Examples of trigger terms to me are “New York Style Pizza”, “Authentic French Pastries”, and “Fresh Local Pastas”. Immediately, you can start differentiating offerings (New York, Chicago, Colorado Style Pizzas). You can almost smell the difference - and there’s nothing more powerful for evoking imagery than smell.


Not quite going so far as to try to develop some kind of digital olfactory offering but I hope the point is reasonable. Of course, there are a lot of interesting questions surrounding this approach (like what’s Authentic or Local) but at least this is a little further away from content derived from opinion and review (like BEST PIZZA). And there certainly is some fuzziness around categories (Northern Italian Cuisine) but still feel like getting the right trigger terms for a business is the missing link between getting the basic business information right and getting Deep Content for a business. At the end of the day, it’s all about getting away from these anachronistic Yellow Page-derived categories (a legacy of the print book) and moving towards a more genuine and authentic way of describing a business quickly.

Tags: No Comments.

TechCrunch just announced that Ford is promoting a hackathon for Automobile Application Access via Applink. It’s always been a blind spot to me since it always seemed the car was a closed system but to see at least some measure of open access sets the mind to wandering. Have submitted our ThingsThere app as a candidate but not holding breath. Still, I think this is the ultimate for the LoMo of SoLoMo but wonder where the social is in the local mobile in this context. Nothing like a friend finder at 80 mph.

Tags: No Comments.