Ben's Website

Serious musings

CommuniCam: A camera to build communities

I think there's a natural opportunity to use connected cameras to improve communities, using the form factors pioneered by Ring, Nest, and Wyze but with a business model designed to guarantee local control and increase trust.

The basic model would be:

  • A technology company engineering, manufacturing, and operating (eg firmware updates) connected cameras and other sensors, charging predictable fees to:
  • Franchisees to deploy hardware and form customer and community relationships, enabling community members to buy/rent, aggregate data, analyze, and understand aspects of their communities.

Today's cameras are rudimentary, processing locally to determine when to stream video to the cloud for analysis and action. This is generally efficient, allowing the units to be battery-powered and concentrating the analysis to servers managed by a single entity. Customers paying a subscription get defined value: in the case of Ring, photo or video of any doorbell ringers alongside a more general feeling of security provided by backing up footage to the cloud. Amazon (Ring's owner/operator) provides the camera units and doorbell notification service but, having the raw footage, they develop additional, generally non-consumer products off that data. While the consumer subscription and hardware fees may cover the bottom line, the aggregation and analysis of the data is where (Amazon believes) the money is.

But that second part isn't actually necessary and does limit adoption, sparking both intimate (filming of family life) and big brother (government surveillance, product usage) privacy concerns. If the CommuniCam company were structured as a franchise nonprofit, co-op, or B-corp, it would be able to resist abuses of its network, focusing simply on developing and deploying commoditized hardware. This tame structure is its competitive advantage; any privacy compromise breaks the whole enterprise, leading it to join its competitors in brand-competition.

Franchisees would play their part in communicating and maintaining the standards of the parent. Their larger value is in helping communities to understand themselves, to coordinate local actors to better use their CommuniCams. They would do this by running defined studies, accessibly and accountably to hosts.

Imagine that a road is going to be rebuilt and the city is to conduct a traffic survey to understand road usage. The city has its standard process of counting vehicles and modeling traffic flows; this data is inherently limited. CommuniCam hosts could contribute to this by running a traffic study on cameras along the route. This would first validate the basic cars/hour of existing traffic surveys, but also estimate destination flow (by matching vehicles) and speed. Personal, commercial, and industrial users could be distinguished, along with bicycles, scooters, runners, and pedestrians. Driving behavior might be observable, depending on camera orientations and available analyses. The CommuniCam driving study would provide many of the outcomes of an in-person traffic survey, in a more scalable way over a longer time frame.

Franchisees would generally make money by subscription and study commissions, with some of the study commissions possibly paid to hosts in the form of subscription credit. They would likely (eventually) be compensated for running the traffic study and formatting the results for the municipality's consumption. But if these hardware and analysis capabilities exist, how will CommuniCam franchisees compete against municipal vendors of the same technologies? Well, the same hardware might equally be deployed to search for a missing cat. That is, the benefit and control lies with the CommunicCam hosts. If they want to improve (or check) the municipal traffic study they have that ability. But they need not be restricted to that single outcome; the traffic study could be on-going to highlight problems for the city or they could compare their traffic patterns to other parts of the city.

Some non-road studies include:

  • rainfall, snow, and hail estimation,
  • sun and light quality
  • vegetation monitoring (grass growing rate and color indicating sun and nutrient)
  • urban animal population (deer, turkey, rabbit, squirrel, racoon)

The camera systems would be designed or configured to be absolutely trusted. What that means is, whereas virtually all of today's solutions are internet-connected, leading to frequent abuse by private, government, and unauthorized actors, these connected cameras would be configured for local storage and processing. To run a study the franchisee would push a study filter to individual units which would screen their data, returning responsive elements to the franchisee for aggregate processing. This bifurcation is essential: individual units steward their data according to the hosts's retention policy and the device's limits, releasing it only when they can internally verify that their data is responsive to a study.

Surveillance capabilities are inherently negative, but the degree of negativity is worth examining. Hard negativity is the dystopian use, where people and communities are subject to ever greater powers which stand at ever greater remove from their interests and control. Soft negativity is the sense that CommuniCams are stand-ins for their hosts (which might be termed positive surveillance), that the community aims could be accomplished by having every host sitting at their mailbox, noting the same information for the same studies, but for the demands of time and attention. This is the soft sense, that individual units are less than the fullness of their hosts, that only a subset of the benefits of being present on your driveway are realizable by units.

To be frank, participation in crime or speed enforcement studies are immediate sources of notoriety and may obstruct other possible benefits, tending toward a broken-windows tallying of problems without encouraging root remedies. The recent launch of SpeedcamAnywhere, a phone app that estimates the speed of passing cars, is one of many upcoming tests of technical norms enforcement. In fact, my starting thought for CommuniCam was as a speed and traffic law observation system, allowing, say, the inhabitants of a corner house to document turning-without-stops, increasing their ability to request enforcement or the redesign of their intersection.

The question of surveillance is the challenge for hosts and franchisees, how to use technology to advance their communities rather than divide them. The larger point is that this tension is best resolved by individual people, as every other resolution involves a greater chance of division. With respect to the creeping surveillance state, CommuniCam does advance it by virtue of deploying surveillance-capable technology more widely. The hope is that this can head-off worse uses of the technology, and by having it widely deployed may educate society to its capabilities and thereby advance the norms and laws surrounding its use.

Hosts, of course may revise their study criteria or change franchisees whenever they become concerned by the franchisee's actions. They may also not use a franchisee and instead operate their own island of cameras for their own purposes, such as pet or child observation. The beauty of the local study filter design is that individual camera units can participate in multiple studies, from different organizations at the same time. That is a host may run local domestic studies, while participating in school-crossing and municipal traffic studies brokered by their franchisee, and at the same time contribute to a nationwide flora observation, while also being able to provide retrospective wind- and hail-damage documentation to their insurer.

The franchisee becomes necessary when spanning property domains, as it is organized and insured to satisfy whatever legal requirements exist. I think it likely that any data communicated beyond the host would be reduced to some form responsive to a study or at least transmitted along with the originating participation criteria, so that downstream uses may be audited. Beyond their subscription fee, franchisees would likely compete on their track record of study success, their community impact, and their audited study rating.

What's to stop a nation-wide, anonymous franchisee from dominating the system? The only technical limit is on the individual devices, that they are in the control of their hosts and subject to the host's study participation criteria, one of which may be a limit on the distance the data may be used, eg within-block, -city, or -state. The more active limit is on the decreasing utility of nationwide data, that as data from individual cameras is removed from its source, it loses value. And this is the great contribution of the franchisees: they, being locally rooted, can find new uses for CommuniCam data in a way that a remote organization cannot. This anticipates the commoditization of both the camera hardware and the analysis methods. After these, the only roles are those of host and value-finder, the one who connects the dots to find a new use for the data that is matched and responsive to a real problem. With a little time, franchisees will be able to discover and define community, profiting from their presence in an understanding of the community.

And there's nothing that prevents the franchisees from participating in national studies, again provided the study filter satisfies individual units criteria. It is just that larger studies will have comparatively smaller compensations, owing to the ever larger institutions that must be engaged to affect any change.

I think this could work, and suspect it will be better than whatever Amazon is planning. Would you like to join CommuniCam? What concerns do you have? Are there other design choice to help CommuniCam avoid the errors of Nextdoor? Please email or comment on twitter.