D Eli, Angelica go in the fuzzing business, sculpties Done 1.7

Early in the morning, before Wonk was awake, Yolanda, CB and Angelica were sharing a cup of 5x water filtered coffee at Mojo's.

I'm talking about deep state's growing ability to track people. The problem, Yolanda said talking a sip from her travel mug, is that though a number of sources the government has been compiling facial recognition for X million people. It's amazing, the government doesn't have the money to address the nerves, opiates, health care, areas struck by hurricanes, maintain the military, or pay down the debt, but they do have the money to spy on US citizens at an unprecedented rate. Blue Circle has had personas for over a century and used look alikes in public, but the recognition is getting so good that we are reaching the point where this won't continue to work. Personas cost booocooo bucks to create and maintain, so that is a serious issue.

I like your hair up like that, answered CB, kind of a sassy look for you. Whoops, a drop of coffee had fallen from her cup to her linen tunic of many colors.

Well, the good news is that nobody will notice, quipped Angelica.

Girl, I saw that photo of you on Snoutbook in the courtyard, I see you have the whole Eastern inspired thing going, Yolanda shot back playfully.

It was fun, great HNMNBie oriented resort they chose for the mindfulness weekend,  Angelica smiled remembering, then returning to the subject at hand.

Extreme data collection is becoming a serious problem.  The deep state activities seem to be taking on a life of their own amidst an absence of adult supervision. Then they don't protect the data that they shouldn't collect in the first place, organized crime gets it, or worse in my case, the Robustos.

We have a game plan for 2D images. Eli's robotic portraits can be adjusted to produce similar photos, almost exactly like a photo of the person, but slightly different that we insert in HNN and other media outlets. Early experiments are very promising, computers do not see like we do. We can even keep the facial structure 100% the same, and just goof with the grey scale and cause a match to fail entirely, or reduce the image match confidence level.

Here it is, CB interrupted, she pulled the picture Angelica had posted from and showed it to Yolanda. Does every room at the Paradis Plage come with a pillowed courtyard, Yolanda teased Angelica.

No, but every room does come with pillows, Angelica answered flippantly.

What we need though is 3D. So that the photos other people take, reporters, bystanders with smartphones etc, they don't go through our system are slightly different. That is a lot harder, in fact, right now we are stumped.

Angelica  nodded, we have male and female robotic dolls of course and a face is just a skin. But how to put them into circulation at scale?

Clara Bow suddenly said, I have an idea, we are going to need Fad King to make it work though. What about convincing people that anyone who is anyone has a sculptie.

Sculptie?

What if we could manufacture, 1000s of robotic heads for wealthy people, and also BCs and HNMNBies, that look slightly different than the real person. Eli started it with the portraits, people want a portrait of themselves, or an ancestor that looks like them and are willing to pay. So now we start pushing busts, sculpture.

I get it, said Gloria, yeah, like a selfie, but 3D. Only we tweak it just a bit, so that it pattern matches differently than the real person. Usually an enhancement, smaller nose, stronger chin, lose a wrinkle, stretch it to make it look thinner, we need to feed egos. But if we do this, we can also put the skin on a bot from time to time and get it photographed.

Obviously we need to see this among all the A list, the haves and not just circle and HNMNBie.  Like busts, in fact, the same 3D model that we use for a skin can produce marble busts.

So what's the point, I don't mean to be slow, Angelica said.

No problem. Right now, it is too easy for deep state to use software to ID a person based on an image, open a file, and add the event record.

What if 50 robots with skins appear at a professional football game?  They always have image recognition to scan for terrorists, or see who doesn't stand for the anthem. Kinda like the bowler hat people in "The Thomas Crown Affair."

They don't exactly match, so they have to open a child file for each one. They currently require a verifiable >95% confidence in a match to add an event to an ID record. Purging bad data is hard and frightfully expensive. These guys aren't geniuses. The facial match stuff they use is straight off github, open source, right down to the academic papers.

Woah, slow down, Angelica commented. The big social media companies have been claiming face matches of near 100% for over a decade, where does the 95% come from?

Clara Bow smiled, social media is a probability score, deep state is an actual, you are comparing lemons and limes and limes cost 8 times more than lemons. If your favorite friendsite brings up a photo tagged to be you, or a buddy, and it is wrong, no harm, no foul, you probably just get a good laugh. As I just said, to clean up an intelligence file, cross linked across several agencies is a nightmare, it would end up costing more than most people make in a year. So they are much more conservative about what they file and how they file it. After all, there is so much data flowing around that neural nets make connections of unlabeled information all the time.

So, from the top. We get some celebrities and so forth to commission sculpties, make sure it gets written up in the media. Start a couple shell companies to make these. The folks with disposable income get the marble, brass, whatever bust. We have the model, so from time to time we 3D print skin variants targeting 94 - 96 percent match. Robots show up in public, look like a famous politician, actor, actress, photos are taken by the press and people using their smart phones. Images get clipped, run through the deep state image recognition SW, get flagged as too close to call. Human analyst takes over, that not only adds cost, but whether they get it right or wrong, it makes personas more effective, it will only take a couple hundred before someone starts thinking about changing the threshold. As soon as we get a couple hundred in circulation, we can start with our people. At the same time, we never again run a totally unaltered facial image through HNN, tiny changes, 50 shades of grey scale. Within a year or two, if we do this right, there will be so many fuzzy matches that they will have to back off taking action on their analysis. It only takes a couple cases of harassing a senator at an airport, or not allowing a famous actor to pass through customs and policy will change.

Let's run it past Wonk, he has a pretty good idea what will sell. This idea can't work if people don't buy the sculpties.

The 3 ladies descended on Wonk's office, excited. Shared their idea. Wonk chose his words carefully. Apartments are 20% smaller than they were just 10 years ago. Horizontal space is at a premium. It doesn't do anything. I get the play to ego, but it's not enough. It has to do something.

Besides think of all the movies you have seen where the statue moves. Can we make it like a smart speaker? Can we make the mouth move?

Making the mouth move is no problem at all said Angelica, there are some cost and reliability due to moving parts issues, but a bot is a bot all day long. Smart speaker might sell, you want a cylinder, or a self portrait, I like it.

Hmmm. What about your delivery bots, they have skins, why do they all look alike anyway, asked Wonk?

Those were chosen mostly by focus groups to be accepted by humans, but we have been running them for years, we could use the skins that way, sure we could.

Maybe you could even have some fun with it, Wonk said, license a Leonardo DiCaprio pizza delivery bot, but mess with the grey scales.

Actually that is almost exactly what he does, he looks the same in his movies, but there are subtle differences and it isn't just facial hair. He is probably outside of our price range, but I like the idea.