A Distant Sound of Hooves: EKOS

Is this the race we’re watching?

Canadian mainstream media knows only one way to cover an election: it is always a horse race with polls coming out weekly or even daily in which one party or another edges ahead or falls behind by less than the margin of error.

Polls are funny things: they give a particular picture of the race at a particular time without providing much by the way of explanation. And, in Canada, the most reported “national” polls measure a race which does not exists. We don’t vote nationally or even province by province: we vote riding by riding.

The bright boys in the Conservative and NDP war rooms know this and, apparently, someone has been kind enough to explain the rudiments to the geniuses surrounding Trudeau. The fact is that the election turns on, at most, 100 ridings scattered across Canada. Amusingly, these are not the same ridings for each party.

With less than a month to go to election day, but with a month of campaigning and polling behind them, each of the parties will be able to focus its efforts on a) marginal seats where that party’s sitting candidate may lose, b) competitive ridings where that party’s candidate might win a riding previously held by another party.

Talk of the Blue Wave or Orange Crush is like the English pre-WWI talking about rolling the Huns up by Christmas: now we are in trench warfare. And now, small differences are all that matter. Exciting as it may be for the Greens to run 5% nationally, they are running more or less even in Victoria which would up their seat count to 2 and knock an NDP held seat off Mulcair’s search for a plurality of House of Commons seats. And there are ridings like this across Canada.

At the same time, the trench war is influenced by the perception of who is actually winning the overall election. Political scientists talk about bandwagon effects. Here Harper has the huge advantage of incumbency. For every Harper Derangement Syndrome voter out there, there are at least one or two voters who, while they don’t love Harper, prefer the devil they know.

EKOS is out with a poll which has the CPC ahead with 35.4, the NDP with 24.5 and the Libs with 26.3. And there is this:

The poll results now show the Conservatives with clear leads in British Columbia, Alberta, the Prairie provinces and in Ontario, where 38.7 per cent of respondents are backing the Tories compared to 30.3 per cent for the Liberals and 19.9 per cent for the NDP. toronto star

If that Ontario number is even close to right it suggests that for all the noise, neither Trudeau nor Mulcair have actually connected with the voters they need. Because this poll deviates from the rest of the polling suggesting a very close race indeed, it is going to be called an outlier. And, coming three weeks before the election, it is not terrifically predictive; but it will certainly motivate the CPC troops as they fight riding by riding.

 

Form Factor Future

Tackle Box Robot

A lot of the substantive content with my homeschooled boys is looking at interesting things and then discussing them and digging a bit deeper.

For no particular reason this past couple of weeks we’ve been discussing:

• pallets and containers
• delivery drones
• 3D printing
• autonomous and electric cars
• economic aspects thereof

Now the fun part of this is that we can find useful things on the internet like Captain Capitalism’s thought provoking post on “Post Scarcity Economics” or Walter Russell Mead’s post on “Is Downton Abbey the Future of the US Economy?”.

While the economics are interesting, the actual history of things like containers can be riveting for boys and their father. (Susan decamps with a good book.) An article about the humble pallet gives the boys a fair bit of insight into everything from the global economy to the logistics of Ikea cup design.

One of the recurring themes which has emerged is how standardization improves efficiency. Containers have to be the same size, pallets – in an ideal world – would be the same size. A decision would be made about metric vs. imperial. Merchandise packaging would be optimized to max out pallet efficiency. There’s room for a bit of math and some basic geometry.

Once you start talking about pallets and containers you can also consider the “last mile” problem. How do you get the goods to the customer? Amazon uses the mail and UPS. Walmart wants you to actually go to their store. Nothing is more fun than thinking about how that last mile can be crossed without a trip to the mall or the UPS guy finding you are not at home.

In Australia there is a company set to deliver text books by helicopter drone. (And, yes, we did discuss this quaint idea of a paper “textbook” in a Kindle world.) No question that for things like pizza, prescriptions and drycleaning the flying delivery drone makes sense.

But, for my money, the autonomous vehicle is a better bet for the day to day business of getting stuff to people. (3D printing is still a distance away for every day use.) The excitement and hype in the autonomous car world has been about passengers – essentially moving people rather than things. But moving things is a huge business and it could get much bigger if an easy, inexpensive, means to get your groceries to your house could be devised.

We already have automated warehousing. (Worth looking at this video at Amazon’s Kiva systems site. However those robots merely pack the boxes with the customer’s order. Now what?

The last mile problem is going to get a lot of attention in the next few years. Autonomous delivery vehicles are one part of the solution; but the other part is actually delivering to the customer. Canadian start-up Buffer Box (recently bought by Google) has a fairly elegant partial solution. Your stuff is delivered to a Buffer Box kiosk which has electronic lockers you can open with a passcode from your phone. Nice for your online kite purchase, not too brilliant for the chicken you want to cook tonight.

There are lots of ways to attack the last mile problem. Each family gets a personal box and deliveries are made right to your box. Or, if price can be brought down, a person or family would have one or more “boxes on wheels” which would travel on some sort of schedule to the various places the family needs to have stuff picked up.

Part of the educational process here is that these are not questions which, at present, have answers. But they are not abstract issues: even a partial solution to the last mile problem is a billion dollar business. And it is a business which will occur very, very quickly. The horse was replaced by the internal combustion engine in less than 20 years.

Marrying a GPS/internet/Google Map aware computer to the rather well understood technology of the golf cart or electric scooter and you have an autonomous delivery vehicle prototype. With a secure storage capacity – think trunk of a car – you could probably build such things for $1000. They don’t have to go fast and, with good logistics design, they would not likely have to go far.

Now, think about what the introduction of such personal pick up and delivery ‘bots would change, especially in cities. Just one example: at the moment even if the Lady of the House is hyper aware of which items are on sale at which grocery stores this particular week, she is not particularly willing to make five stops and go through five checkout lines to save a total of, say, $10-20. But her pick up agent would be delighted to make those stops. (And, of course, now the supermarkets – if they are smart – are going to want to attract the pick-up agents so the scope for price matching and agent loyalty programs is huge.) On the other hand, think of the congestion these pick-up agents might cause. How to solve that problem. (Two hints – first, the number of car trips to get stuff would go down as would the number of manned delivery vehicles, second, it should be possible to build anti-congestion imperative right into the software which runs the pick up agents.

As I have pointed out to the boys, one interesting thing about robotic pick up agents is that they can operate continuously and therefore quite slowly. If your agent has to make four pick ups all within a one mile radius of your home over the course of, say, three hours it can accomplish that at a walking pace even allowing a ten minute stop at each pick up location. While I am not all that thrilled with the prospect of driverless semi-trailer trucks running at 60 miles an hour (which is an irrational fear but there it is), box toting pick-up agents sauntering on the sidewalk or in designated road lanes seems pretty manageable.

Thinking about thinking – or why philosophy matters

Now, the truth is that knowledge consists of conjectured explanations — guesses about what really is (or really should be, or might be) out there in all those worlds. Even in the hard sciences, these guesses have no foundations and don’t need justification. Why? Because genuine knowledge, though by definition it does contain truth, almost always contains error as well. So it is not ‘true’ in the sense studied in mathematics and logic. Thinking consists of criticising and correcting partially true guesses with the intention of locating and eliminating the errors and misconceptions in them, not generating or justifying extrapolations from sense data. And therefore, attempts to work towards creating an AGI that would do the latter are just as doomed as an attempt to bring life to Mars by praying for a Creation event to happen there. david deutsch

The article will make your brain hurt but it goes a good distance towards explaining why Artificial General Intelligence is a very tough problem indeed.

Interestingly, Deutsch adopts a Popperian position – anathema to many fuzzy thinkers simply because it takes seriously the idea that the process of thinking is about coming up with right answers,

The lack of progress in AGI is due to a severe logjam of misconceptions. Without Popperian epistemology, one cannot even begin to guess what detailed functionality must be achieved to make an AGI. And Popperian epistemology is not widely known, let alone understood well enough to be applied. Thinking of an AGI as a machine for translating experiences, rewards and punishments into ideas (or worse, just into behaviours) is like trying to cure infectious diseases by balancing bodily humours: futile because it is rooted in an archaic and wildly mistaken world view. david deutsch

Dislocation

“The [robotic] technology proposed appears to allow cutting and sewing at costs LESS THAN in China,” according to Softwear Automation’s website. “There is only one basic innovation required; that the metric of motion should not be meters or inches but rather thread count in the fill and warp directions.”

Success could spell out huge disruptions for workers as robots continue taking over human jobs in manufacturing and other industries. Low-paid workers in developing countries stand to lose out the most in this case, but U.S. workers won’t gain much, either. Still, U.S. businesses could once again regain a foothold in the garment industry and win back a share of international trade. mother nature network via the prof

The direct substitution of machines for men (and women) has been going on for a couple of centuries but we still really do not have much of a handle on its implications or its economics.

Propose for a moment that you can robotize clothing manufacture. What would be left is building the robots and the buildings in which they would function and then designing the clothes themselves and the fabric and doing routine maintenance on the ‘bots.

So while the clothes produced would be very cheap indeed there remains the question of who could actually afford to buy them?

This is an increasingly troublesome question as various working and middle class jobs fall of the table. Economically, from a Canadian or American perspective, a Vietnamese seamstress is pretty much the same as a robot. However, from her perspective, she is at least making some money which she will spend on goods and services of some sort.

Robots don’t spend money.

Now, various economically savvy types will suggest that the robots’ owners will make tones of money which they will spend. Perhaps. It depends on whether those owners are public companies with lots of shareholders who actually consume stuff, or if they are tightly held companies with very few owners who have a lot of money to spend…buying more robots.

And the economically savvy will also point out that the clothes made by the robots will be really, really cheap. Well, so are the clothes made by our Vietnamese seamstress. Making a t-shirt cost 10% less is not going to make it vastly more affordable to people who essentially have no income.

Trade economists talk about comparative advantage and Marxists mutter about the ownership of the means of production; but underlying such conversations is the assumption that it is, as a practical matter, impossible to directly substitute capital for labour in its entirety. The evolution of robotics is making that ever more possible.

Thinking hard about the implications of labour replacement devices and their impact on how we allocate income is something we should be doing right now. Before a robot comes up with a solution which makes sense for, well, robots.