47 comments

[ 2.8 ms ] story [ 99.1 ms ] thread
We posted Density a while back when we used mac address tracking (https://news.ycombinator.com/item?id=7305097). We've since moved to a wirelessly connected door counter, would love some feedback.
Glad to see the change in methodology! I loved the concept then and still do, but freaked out at the thought of my MAC address being logged everywhere I go.
We're happy with the change too. Simplified things. Any sensor trying to mass measure foot traffic across a city needs to be accurate, real time, anonymous, and cheap.
This is presumably one reason why Apple has stopped broadcasting your real MAC address while doing SSID probes, instead randomizing it.
And Android followed suit. The policy change pretty much bifurcated the industry. Half died or sold themselves off - http://fortune.com/2014/10/29/after-a-blow-from-apple-nomi-s...

The other half doubled down on MAC address tracking tech, arguing that it still generates meaningful data - http://www.euclidanalyticsblog.com/2014/08/beacons-and-wi-fi... Although I've heard they've started to move toward video and facial recog.

We were so small at the time of the announcement last June that we thought... "Why don't we just build a door counter. How fing hard could that be?"

...turns out it's like way, really fing hard.

But here it is - http://i.imgur.com/zLUMAG8.jpg?1

I had a similar idea, I wanted to place small microphones all over the city to know which places were "quieter" than others at certain times of the day etc... I was looking for a place and I wanted to live in a street that was quiet on weeknights
SceneTap is doing something similar with facial recognition technology to layer on age and gender split

http://scenetap.com

Super cool idea. Best of luck with this.
Will there be an API for this data to allow business owners to extract it?

One of my clients is building a dashboard that correlates data from various sources, and this kind of data for physical businesses has long been a dream.

Yes. Actually, that's our model. We designed the hardware to be cheap enough to be free (situation dependent) and then we monetize the resulting data.
It would be nice if you let people know before they signed up for an account what cities you are in...
Good point, definitely something we overlooked. Next build will include that, as well as the ability to request locations from anywhere.
Nice, however given the name, the triangle logo associated a lot with destiny the game. http://img3.wikia.nocookie.net/__cb20131219001817/destinyped...
Looks interesting, how do you do this exactly?
Looks like the business has to purchase (proprietary?) hardware to install on-premises.

Usefulness will depend on whether the location you are interested in has the required hardware.

Generally, we give the hardware away for free (circumstance dependent). We'd prefer to monetize the data. We don't really want to build a company on $49.99 / mo. There are a lot of other players in that game. Here's a picture of the sensors if you're interested: http://i.imgur.com/zLUMAG8.jpg?1
This is great, I was just thinking about a similar idea when I waltzed into my empty barbers the other day, good luck!
Had the same idea a while back whenever I called a restaurant to check the wait time and it went to voice mail. App looks good, I will try it out this weekend.
Local business voicemails are the worst :)
Very nice! You've taken a big step towards blowing up the assumptions behind the canonical version [0] of the minority game.

I imagine you're already thinking about how you can enrich your dataset and provide analysis that will help people solve real problems of consequence. Parking might be a big one: there's surely money to be made directing drivers to available public parking spots or showing businesses and real-estate developers the potential to share parking among uses that are statistically complementary with respect to parking demand.

And please add Kansas City.

0. http://en.wikipedia.org/wiki/El_Farol_Bar_problem

Re: Minority report. Yeah, I think there's a space in the future where utility and privacy don't have to be traded for one another.

There's another cool company doing parking called Parkifi I think.

Someone else mentioned El Farol Bar. If we're causing too many people to show up or too many to stay home... that would be an awesome problem to have.

Re: Kansas City. Friends with some of the founders / early people over at Zaarly. We're definitely keeping Kansas City in mind ;)

Minority games, not Minority Report the movie. :P In game theory, minority games [0] are a class of games that includes the El Farol Bar problem, a class in which the payoffs depend on the number of actors who make each choice. For the most part, the strategic aspect of these games disappears when you have enough information about who has choosen what so far.

On the parking point—there do seem finally to be some real efforts underway to gather data about parking availability. At an average cost of about $40,000 per spot in a parking structure and low overall utilization, it's astounding that it's taken this long for serious attempts to emerge to find ways to allocate parking more efficiently. In addition to solutions that involve putting sensors in situ, car-makers are starting to think about how they can use the ever-expanding array of sensors in their vehicles to map parking spaces and track their availability [1].

But there are surely other open problems of the same sort that Density's data could help solve: how many people are jamming into the subway right now, or how many are at the farmer's market already? In this common sort of situation, your experience depends on how many other people are doing what you're contemplating doing, and your behavior will (or would) change with better information about who's doing what.

Look for those sorts of problems to apply your data to, I'm suggesting. There might also be opportunities to map demand irrespective of supply. In the parking situation, for instance, tracking the number of open spaces only gets you so far; you also want to know how much aggregate demand there is, because that helps you figure out whether to build more capacity. Second-order statistics like how many people are checking a location's activity in the app can help you estimate that sort of thing.

0. See, e.g., http://www-f1.ijs.si/~rudi/sola/MinorityGame-Seminar.pdf [PDF]

1. http://arstechnica.com/cars/2015/02/remote-valet-mode-and-re...

re: Minority Re..port... I've always been bad at reading. Do you know another example of a minority game? And is it a paradox or just a method of thinking about game strategy?

Parking - I agree. That one company has a unit that senses the presence of a battery in a parking spot.

Other open problems - I think your talking about how datasets affect one another. What's really interesting about that regarding foot traffic is that there currently isn't a way to know how many people came and went. Which means it's currently tough to draw connections between how foot traffic might affect the farmer's market or the subway or the financial district in San Francisco.

Can you explain map demand irrespective of supply?

Checking PDF..

Neat idea, though the lack of an Android client is disappointing; is one in the works?
We're focusing on iOS initially, but certainly plan on an Android app.
This is really cool. My cousin and I had a similar idea but a phone would be used instead of a custom sensor.

Question - why would a business be OK with broadcasting a "quiet" status? Wouldn't that reflect poorly on them?

Can you give me an example of how quiet could be bad?
Bars, clubs, etc.
Yeah.

Well, I guess it could be bad if I knew that it was always quiet and therefore "a drag" (sorry, that was a weird word to use). But I suspect any bar or club that's reasonably solvent, especially in populous cities, has busy times.

Atm though we only show the traffic right now and notify when something is triggered, quiet / trending.

Second, I think there are people out there that are not looking for busy bars. Rather quiet ones. I certainly fall into that category.

We're not in any bars right now, though. I'll get back to you once we have more data.

What does "trending" mean? It's... popular right now for being quiet?

It seems like if you had the data, you could build both an app that's marketed to tell you when places are quiet, and an app that's marketed to tell you when places are lively.

EDIT: Or do you envision the app to do both? I'm having a hard time squaring "The Heartbeat of a City" with "Is your favorite place quiet right now? Check it out in realtime and get notified when it’s convenient to drop by."...

> Second, I think there are people out there that are not looking for busy bars. Rather quiet ones. I certainly fall into that category.

I'm curious what the market size is for people looking for quiet public places vs. lively public places. Is there already a popular app that targets the latter? Has anyone successfully filled the gap left by Foursquare eliminating check-ins?

(comment deleted)
Does this only work with places that have one entrance / exit?

Or could someone rig up say, an entire hospital with this thing and see stuff like admission / availability of the waiting rooms at nearby ERs?

Rig up... We can normalize multi-entrance places. Actually Blue Bottle is a multi-entrance place.
(comment deleted)
Good concept! Just curious - can the device accurately track multiple people (or group) entering / exiting together in one door swing?
We're essentially identifying patterns that occur in the voltage readout of two parallel sensors. Two different people, even if passing through without a gap, show enough of a difference in their pattern to tell them apart. It simply takes time to root out all of these granular edge cases.
Got it, thanks! Not sure what use cases you have already thought about but certainly see this being quite handy from a business perspective e.g. retailers can run real time customized offers/campaigns depending on the levels of density at their place to engage with the users better and improve overall experience (of course bottom line as well:)).
Like "reverse surge pricing"... free coffee until the place gets busy.
Exactly or maybe 20% discount on your favorite meal depending on the store/time of day/user profile etc. This can to some extent ensure the overall utilization of the retail store and thus profitability are improved. I'm pretty confident retailers wouldn't mind paying for this on a saas basis or some other similar model. This will also help on the user acquisition side as there's now a really cool incentive for downloading/using the app!
Are there any plans to make the data available via API or maybe offer webhooks? Would be cool for other folks to mashup this data in various ways.
Yup! That's actually our model. I'm really excited to see what people do with the data

Love the webhooks idea, I can see users hooking into something like ifttt.