Category Archives: Poetry

Why Driverless Cars Can’t Understand Sand

Sand is a fluid such that driving on it can be hard (pun not intended) even for humans.

It’s like driving on snow or mud, yet it seems to be far less well studied by car manufacturers because of how infrequent it may be for their customer base.

Source: Simulator Game Mods “Summer Forest”. Snow and mud computer driving virtual environments can easily be found, yet sand simulations are notably absent.

Traction control, for example, is a product designed for “slippery” conditions. That usually means winter conditions, or rain on pavement, where brakes are applied by an “intelligent” algorithm detecting wheel spin.

In sand there is always going to be some manner of wheel spin, causing a computer to go crazy and do the opposite of help. Applying brakes, let alone repeatedly, is about the worst thing you can do in sand.

On top of that the computer regulation of tire pressure sensors has no concept of “float” profile required for sand. When the usual algorithm equates around 40psi to safe driving, deflating to a necessary 18psi can turn a dashboard into a disco ball.

The problem is product manufacturers treat core safety competencies as nice to have features, instead of required. And by the time they get around to developing core competencies for safety, they over-specialize and market them into expensive festishized “Rubicon” and “Racing Design” options (let alone “WordPress“).

In other words core complex or dangerous scenarios must be learned for any primary path to be safe, yet they often get put onto a backlog for driverless. Such a low bar of competency means driverless technology is far, far below even basic human skill.

Imagine it like exception handling cases or negative testing being seen as unnecessary because driverless cars are expected only to operate in the most perfect world. In other words why even install brakes or suspension if traveling parallel to all other traffic at same rate of speed, like a giant herd? Or an even better example, why design brakes for a car if the vast majority of time people don’t have to deal with a stop sign?

Recently I put a new car with the latest driverless technology to the test with dry sand. I was not surprised when it became very easily confused and stuck, and it reminded me of the poem “Dans l’interminable” by Paul Verlaine (1844 – 1896).

Dans l’interminable
Ennui de la plaine,
La neige incertaine
Luit comme du sable.

Le ciel est de cuivre
Sans lueur aucune.
On croirait voir vivre
Et mourir la lune.

Comme des nuées
Flottent gris les chênes
Des forêts prochaines
Parmi les buées.

Le ciel est de cuivre
Sans lueur aucune.
On croirait vivre
Et mourir la lune.

Corneilles poussives,
Et vous, les loups maigres,
Par ces bises aigres
Quoi donc vous arrive?

Dans l’interminable
Ennui de la plaine
La neige incertaine
Luit comme du sable…

“The uncertain snow gleams like sand.”

Bailing Sand

the truck transmission whined in protest, the computer gave up. then, bailing away soft flowing sand from our door sills, shovel burning my hands even under a cool moonless starry night… something was truly exhilarating about digging out.

this machine would never understand. sat quietly and waited for rescue by a tool thousands of years old.

in a way, hacking machines is like driving off-road so far that you’ll maybe never make it out again. and that’s why to do it. humans are driven by curiosity, machines are driven by humans.

Reduce Your Vulnerability to Fraud by Listening to More Stories

There is measurable vulnerability in people who lack experience outside a single story. We documented this in terms of who falls victim to Advance Fee Fraud (419 scams) but it also applies to the AntiFa and BLM scams spreading lately.

A story-teller on TED may explain it far better than I ever could:

Army Captain Florent Groberg, who received the Medal of Honor for charging a suicide bomber in Afghanistan, calls this opening up to others so you can “have a conversation”.

Deepfake Map History

Here’s a weird footnote in deepfake history…

Old publishers of “facts” like map makers and encyclopedias gave up inserting fake data after the US Supreme Court ruled their facts can’t be copyrighted, as a New Yorker article from 2005 explained.

If Mountweazel is not a household name, even in fountain-designing or mailbox-photography circles, that is because she never existed. “It was an old tradition in encyclopedias to put in a fake entry to protect your copyright,” Richard Steins, who was one of the volume’s editors, said the other day. “If someone copied Lillian, then we’d know they’d stolen from us.”

The clever habit of sprinkling fake facts among the real ones declined when money couldn’t be made from proving someone copied facts.

It kind of makes sense that they lost the copyright.

Stealing a fact seems absurd, made only more absurd by polluting a fact to prove it was stolen (an act essentially proving it impossible to steal a fact), and then made even more absurd by trying to prove the decoy (deepfake) is NOT a fact.

“Its inherent fakeitude is fairly obvious,” she said. “We wanted something highly improbable. We were trying to make a word that could not arise in nature.”

They had to make the fake obvious, yet hidden, because they wanted to prosecute people for copying facts that couldn’t be distinguished from fakes. Uh-huh.

In the end there were better ways to create uniqueness (such as very specific distances) versus these silly falsifications. Money for copyright violation seems like such a low bar though, in terms of why someone would insert fakes into a map.

Food for thought when you look at your next map and wonder if you can detect a deepfake. As I wrote in 2020, maps are inherently political and about power.