Saturday, February 16, 2019

Omniscient, Malevolent, Autonomous Robots Threaten Consensual Reality



I was very unsurprised to read this headline which would normally sound very exciting to me:
"Researchers create 'malicious' writing AI"

Not because it isn't true, but because it's already been true for at least five or six years. The only thing different is the word "AI" in there.

What's also different, not with this technology but the greater technosphere, is that we can now see what intentionally-generated low-credibility information can do when propagated at the scale of globally-connected digitally-mediated social networks. (A robot wouldn't use that many hyphens in one sentence by the way, too suspicious; credibility-threatening.)

The idea of holding back your natural language generator for fear that it might create a dystopia sounds more like a natural publicity generator at work.

Nonetheless, some real talk. The thing that makes this new intelligentity special is that it's got a good natural language generator AND a good database of high-credibility content to inform it. That's important, because without the ability to first filter good news from bad news sources to begin with, the output would not be noteworthy.

So this is the next step. As soon as we figure out how to deal with the fake news problem (i.e., by creating databases of credible sources, or eventually criteria that can auto-select), then bad actors will create the automated re-narratization of that "good" information to send fake shit right under our new radar.

I look forward to re-reading this post in another seven years.


Notes:
Researchers create 'malicious' writing AI
Feb 2019, BBC News

Narrative Science
They write programs that write. For example, they will take your quarterly report data, and write about it so people can read it like a story using their Natural Language Generator.

Other Writing Robots:
The Policeman's Beard Is Half-Constructed
written by a computer in 1984
[main]
[PDF]
[html]

Verified Facts
http://www.verifiedfacts.org/
The greatest website ever for conspiracy theories, created by robots, for robots.

Communications From Elsewhere
http://www.elsewhere.org/pomo/
"The essay you have just seen is completely meaningless and was randomly generated by the Postmodernism Generator."

Robot writes LA Times earthquake breaking news article
March 2014, BBC

Post Script:
I'm not sure if this is still happening, but for a while (circa 2013) you could buy a book that was basically a copy of a Wikipedia page. The entire process, from the purchase order, to the printing of the book, was automated. Obviously, there were no authors listed, and no editors. Also obviously, this blew my mind, so I soon owned a physical copy of "Stigler's Law of Eponymy," taken from the eponymous Wiki page.


Eponymy relates to names of things, like how Weezer's first album is eponymous because it's named after the band itself. (Actually, because all their albums are named this way, they get new meta names based on colors, and so the first album eventually became the Blue album. Beatles come to mind.)

Stigler's Law of Eponymy states that discoveries, for example, are  assigned to people other than their inventors.

And from a passage in this very Wiki page, I will now paste, via an unknown volunteer contributor:
"It takes a thousand men to invent a telegraph, or a steam engine, or a phonograph, or a photograph, or a telephone or any other important thing—and the last man gets the credit and we forget the others. He added his little mite — that is all he did. These object lessons should teach us that ninety-nine parts of all things that proceed from the intellect are plagiarisms, pure and simple; and the lesson ought to make us modest. But nothing can do that."
-Mark Twain, "Letter to Helen Keller," 1903   
A real-fake book about eponymy written by an author with no name FTW!
Network Address, 2013


Monday, February 11, 2019

Perpetuity, Posterity, and Pretense


Europe has had enough - the Right to Repair law would require  manufacturers to make things that last longer and are easier to repair, instead of making things that are designed to stop working after a specifically designed time horizon and which can not be repaired, no matter how simple the design failure.

I will now spend the next 6 paragraphs transforming into a crotchety old man complaining about how things are not like they used to be.

Let's begin with how software updates are built obsolescence on steroids.

Now that we've covered that, let's talk about how things in fact have to be designed with their eventual failure in mind. And let's start with duct tape.

I have a problem with duct tape, and I understand I am in the minority because of it, but hear me out. Let's say you have a wall outlet, a duplicate receptacle if you will, and you're plugging a bunch of things in there, so you have one of those adapters that turns two outlets into six. But the adapter keeps falling out of the wall because of all the plugs hanging from it. How do we solve this problem?

Duct tape. And that would be a great solution, except that duct tape, for most of its purposes besides taping ductwork together, will fail. And when it does, you'll have to take all that tape off, only that it doesn't come off. It leaves a very sticky residue behind, marking its territory everywhere its been. (Then again, you might not be using enough duct tape.)

And this is my problem with duct tape. The almighty MacGyver child is not a permanent solution; it will fail, and when it does, the thing you taped it to is now ruined, which means the thing you taped it to has also failed.

Extrapolating this back to European manufacturing policy and built obsolescence, we must remember that nothing lasts forever, and that designers know this. To pretend that your washing machine was not designed with its own eventual failure in mind is to ignore one of the most basic laws of nature - nothing is safe from entropy.

The other side of this fundamental law is that nothing is free. You want a washing machine that will last 100 years? It can be done, but you have to pay for it. It's that simple. When a washing machine is designed, one of the problems measured is the cost of resisting entropy. "How much money do we have to spend to get this thing to last long enough that people would still buy it for more than it costs us to make it?"

We can make a washing machine that lasts 100 years, but it would cost $4,000. We could also make one that costs only $50, but it would last about as long as it takes to get delivered to your house. The two variables have to meet somewhere in the middle. I'll pay $500 for a machine that lasts 10 years. That's $50 a year for a washing machine, which is way cheaper than the cost of a laundromat. (Actually, if I were ever in a position to own a washing machine, I would probably buy it used for $150, and plan on repairing it myself a few times over the next 10 years.)

Finding this sweet spot is a part of the design process. It doesn't make a manufacturer sneaky or immoral to build a product that is 'designed to fail'; failure is a part of the design process.
This is a shot of a jail cell at Eastern State Penitentiary, built in th early 1800's. The original intent of a penitentiary was to repair the damaged inmate by allowing them to reflect on their mistakes. The word is related to repentance.


This is why the Right-to-Repair movement is such a great idea. It reminds me of a design strategy called Design for Disassembly, where you consider how hard it would be to unbuild your building before you build it, and then you design it so that it would be easier to unbuild. In other words, you imagine what happens when you eventually need to remove the duct tape from your wall outlet, and you modify your design accordingly. Or take the Japanese temples that are meant to be rebuilt every so often, the sacred cedar wood replaced, thus rejuvenating the spirit of the building.

Dialing back one more step, this idea can be derived from a Japanese aesthetic called Wabi Sabi. The Wabi aesthetic is to Japan what perfection and beauty are to the West. Under this paradigm, imperfection is beautiful. You can see it in their tea bowls, which are asymmetrical and austere. They have cracks in them, and its ok. In fact, depending on the crack, a tea bowl can been seen as even more beautiful because of it.

Japanese Teabowl

But that's not the best part. Over time, the crack will fill with the leaves of a thousand cups of tea, and they will seal-up. And then the cup becomes even more beautiful. Transience and endurance are held in high regard. The Wabi aesthetic reveres age, aging, and the aged, and sees things as having value because of their age, which is in direct contrast to Western values of perfection.

This distinction also informs the obsolescence-vs-utility consumer calculus of Westerners who will pay for a product that they know will fail in a few years. Under a Western consumerist paradigm, the older a thing gets, the less valuable it becomes, no matter what. Under that formula, why would a manufacturer make something that lasts forever?

When I look at a thing with my Wabi goggles, I don't see what it is, but what it will become. Colors fade, metal rusts, and stone gets dirty. That building may look perfect today, but perfect doesn't last forever. And so its future gets superimposed on its now, thus changing its value.

The reason Right-to-Repair is a thing in the first place is because the force in the market is not strong enough for manufacturers to do it themselves. But asking consumers to change their behavior is way harder than forcing businesses to do the same thing. 

Image source link

Post Script:
The most popular understanding of the concept for Built Obsolescence relates to the making of a product with a deliberately short lifespan. Granted, there are products that are 'made to last' and those that are not. The problem is when the consumer is misled into thinking they are buying something of a higher quality than it is. This version of the concept is itself potentially misleading of the realities of design and economics.

Let's face it, if enough people wanted a mobile phone that lasted 10 years, there would be one on the market. And if more people wanted their cars to last forever, why would so many be leasing? (And then again, there's always inkjet printers to prove me dead wrong; I'm looking at you Espon.)

Moore's Law now has a lot to do with all this. Although it refers to limited aspects of a product's performance, it says that in 18 months, there will be another product twice as good as yours, but for the same price. This affects long term research and development more than short term product design, but in a world where the software has become the product more than the hardware, it should be a part of any discussion about product design.

Notes:
Climate change - Right to repair gathers force
Jan 2019, BBC News

Here's the Truth About the Planned Obsolescence of Tech
2016, BBC Future

Fact Sheets on Designing for the Disassembly and Deconstruction of Buildings
EPA, 2019

4th International Symposium on Environmentally Conscious Design and Inverse Manufacturing, 12-14 Dec. 2005, IEEE, Tokyo, Japan
ISBN: 1-4244-0081-3

DOI: 10.1109/ECODIM.2005.1619391

The practice of sustainable design requires an analysis of the environmental, social, political and economic impact of a product throughout its entire lifecycle. There is much research into the manufacturing and waste management phases; however, little attention has been paid to the 'use' phase of the life cycle. Joseph Fiskel acknowledged in 1996 that, "the longer the life of a product, the more eco-efficient it is, since the same amount of material delivers a larger amount of economic value". This paper will evaluate how the premature obsolescence of technology, the obsolescence of quality and the obsolescence of desire can be managed by using Design for Disassembly (DFD), and applying it to the 'use' phase of the lifecycle to increase product durability.

Deconstruction and Design for Disassembly
Here's a paper I wrote circa 2010

Cradle to Cradle, William McDonough, 2002
This book outlines and encourages the design of a product in light of its entire "life-cycle".

Cradle to Cradle Products Innovation Institute