Bartlesville Radio: The area’s best source for news, weather and sports!

The continual discussion about Flock software product features are a gigantic red herring in the discussion.

The flock software product, as with all software products, exists as a full stack of compounding architectures. In technical parlance these are known as “layers” as in layers of a cake.

In the case of the flock software product, the bottom of the stack is the physical installed device itself as this is where every piece of data the software product consumes originates.

To understand the software product fully we must first fully understand every single supporting layer below it. It’s not enough to say “Flock only takes photos of the back of vehicles and their license plate” as this is only technically true if sufficient computing resources exist at the physical device to make this happen—they do not.

To keep the product simple and as inexpensive as possible, the physical hardware only has minimal physical components: a large battery, a daylight camera, an infrared camera, a motion sensor, microphones (on some models), and finally a cellular LTE device to transmit all data for centralized processing in Flock’s Amazon Web Services database layer.

At this point, we have at minimum tens of thousands of individual devices transmitting likely billions of images every single year into a raw imaging feed on AWS.

From there, the data can easily be replicated for a host of purposes before being copied into a production pipeline for the actual Flock product.

At this point, Flock software engineers can use near infinite computing resources to process these billions of images with artificial intelligence models to classify them into two initial buckets:
1. Images that meet criteria for what we tell customers product does. (note it’s only at this step when the data actually reflects what they say it does, but a litany of other processes can easily occur BEFORE this step)
2. Images that do not meet criteria for what we tell customers product does.

We can now claim bucket 2 images are “deleted” (loaded term by default but is meaningless in the context of any prior replication or parallel processing steps not used for the production Flock product)

Next, all of the bucket 1 images are fed again through our proprietary artificial intelligence logic to apply further metadata to every single image. Things like plate number, make, model, color, bumper stickers/other distinguishing features, routes the vehicle travels frequently, other vehicles that travel in close proximity with it, etc etc

Only after all this occurs does a dataset exist that actually represents what the flock software claims to be doing.

Keep in mind that this is still one gigantic dataset of commingled images from hundreds of thousands of cameras in thousands of jurisdictions.

"THE ONLY THING PROVIDING THE SECURITY TRUMPETED BY THE FLOCK SALES STAFF ARE MERE SOFTWARE CONTROLS SETTING PARAMETERS FOR WHICH USERS CAN SEE AND QUERY WHICH DATA BUT THE UNDERLYING DATASET IS LIKELY ONE HUGE CLEANSED FOR PURPOSE MEGA SET.

KEEP IN MIND THIS CAN EASILY BE ONE OF MANY SUCH DATASETS EASILY CREATED FROM THE ORIGINATING RAW DATASET BUT FOR LITERALLY ANY PURPOSE YOU CAN DREAM UP WITH THE RAW DATA."