A city does away with traffic lights
It looks like a recipe for chaos.
Yet, one city in the US has managed everything from reducing traffic accidents to saving fuel by doing away with traffic intersections. And converting them to roundabouts.
Intersections tend to have more idling traffic than roundabouts that increase the flow. The directions in which traffic can move is not a maximum of four.
And here's the unbelievable thing. People don't have to receive special training. Instinctively, traffic flows far more smoothly than the start-stop-start rhythm of traffic lights.
The typical response to an increase in traffic density is one of two things - build an overbridge or set up traffic lights.
Now managing traffic lights is another major field of study. Several things have been tried - from aligning traffic lights in a row, so that traffic goes through a series of green lights rather than get stuck at frequent intervals.
Roundabouts are apparently a British way of managing traffic,
It's not as if they can only be used at major junctions. Small roundabouts are perfectly workable in the smaller intersections as well.
Drivers slow down when they come to a roundabout. They don't have this competitive streak of trying to beat the signal because there isn't one in the first place.
The reduction in accidents at Carmel is not trivial. It's an astonishing 80%! Plus there's no maintenance and upkeep of traffic signals required.
Since cars and trucks are not idling at the signals, the fuel saving is huge as well.
But it's going to be quite hard to convince people that traffic does not need regulation as much as a way to keep flowing.
And that's where we will keep idling for now.
Can we articulate our own preferences?
The idea that artificial intelligence is different from human intelligence is gaining cachet.
All purchase data captures the action, not the process of consideration - now people would argue that it is unimportant, but without understanding the process, how can we be sure of the results?
I remember a discussion initiated at the agency on the choice of toilet soaps.
No one was able to give a clear reason for preferring their brand apart from the obvious responses like – ‘It works for me, it smells better, lathers in all kinds of water, keeps me feeling fresh’, etc. – things that the advertising has put into their minds in the first place.
When pressed further, they were just plain confused. Then factors like ‘Mum always bought it’ were touted as reasons.
So, if we are not able to explain simple preferences for the trivial necessities like soaps, shampoos, and moisturizers, how do we get machines to understand what preferences are?
If physical properties like tall and short get vague the moment you change the context ( tall among basketball players vs tall among the general populace), imagine how hard it is to define desire. And from a status standpoint.
So, if we want machines to think (whatever that means), we’re constantly confusing it with emotional connect – the essence of being human.
There are lots of terms – humanoids, robots, and the concepts have been explored by some of the world’s best directors in movies, but we are no closer to understanding how to construct them.
Sure, fuzzy logic washing machines and microwaves claim to understand which cycle to use – depending on how dirty the clothes are or what needs to be cooked. But the variables again have to be clearly defined.
Incidentally, how is dirt defined? Dust, grease, curry, oil? On what surface area of the clothes does it have to be to change the settings in the washing machine?
See where this is going?
Torrents - the misunderstood marvel
For most people, torrent clients equal piracy.
But the innovation that created it has a fascinating backstory.
In the early days of the internet, the late 90s and early 2000s, exchanging large files was impossible.
Even an MP 3 file, at barely 3-4 mb would take at least 10-15 minutes to download over slow lines.
So, a 2-3 GB movie file would literally take days or weeks.
In one stroke, torrents changed that completely.
Bram Cohen, the inventor of the Bit Torrent protocol came up with a remarkable solution.
One of the several jobs he moved through after dropping out of college was at a company called Mojo Nation.
The company was trying to build a decentralised system to safely store confidential files.
The concept was that if a file was broken up into pieces, encrypted and distributed over several systems, hackers could not download the file without active co-operation from all the systems involved - which would be tough.
Bram made the connection that this would be the perfect solution for file sharing.
He worked for months to create the system.
Basically it involved exchanging parts of files stored on several computers in parallel.
The genius idea was that people could simultaneously download and upload parts of the file at the same time with anyone else who was online at the time. The program acted like an automated traffic management system.
It would manage the whole process of assembling the file into a single one after it was complete.
So, over a fairly 'good' internet connection in the early 2000s, you could download a movie file in about an hour or so.
Unfortunately, it was always attacked by the established interests in the movie industry legally and a business model around the original could never take off.
But everyone from Netflix to Facebook to Google and of course cryptocurrencies use it even today
The public perception is completely different from the reality