Thursday, October 1, 2020

 I've recently enjoyed a number of "Sherlock Holmes pastiche" stories by Donald Thomas.  He's written nonfiction about Victorian crime, and his depth of knowledge shows - the stories evoke a wonderful atmosphere.  I always liked the Conan Doyle stories, but you can only read them so many times, so finding his work was a real pleasure.

As my trade involves Artificial Intelligence applications, a few weeks after finishing most of his Sherlockian works, a thought began knocking around.  Could one write a "Sherlock Holmes AI system"?

It might work like this: taking an image (or consecutive stills from a video) it would, for instance, examine a person and categorize what it could: is the clothing a good fit, is it in style, is it worn in unusual ways; are there environmental indicators such as being wet from rain or snow.  Concerning the face, are there any telltale signs of illness, stress, makeup, etc.  Gait might be analyzed for its medical or employment implications.  I think we all know the typical Sherlock Holmes story passage in which he deduces amazing stuff from such information.

I realize that the author of such stories has the freedom to put in what he wishes Holmes to get out of his observations, so it's somewhat contrived, but some of that works in reality.  Years ago when commuting in Boston I used to play a kind of game (not really inspired by Sherlock) of trying to guess what I could about someone from the appearance of their vehicle.  Bumper stickers give away a lot.  I recall riding with a friend and I just commented on the driver ahead of us - "A former marine, with a son in the Army, and a daughter at <I've forgotten the private school> who's into equestrian activities".  This did surprise my companion, but it wasn't my intent.  The whole game was driven by the boredom of being stuck in slow-moving traffic.  One could also tell a lot from the car itself - had it been in accidents, the nature of the repair history (and possibly the economic status of the owner), the residential or academic parking stickers, the cleanliness of the windows.  So I do know that some thinking along these lines really does work.

An AI system might similarly also look at other aspects of scenes - lighting, mirror reflections, furniture type and condition, indicators of taste or travel, family photos, pet presence, education level, political orientation, level of social activity - to be the basis for further analysis/deduction.  At this point it's just a matter of basic information collection and interpretation.  In fact that may be enough for such a system to be useful; making that information available to a human in some way that allows exploration and correlation with other data might be as far as it needs to go.

The thing that seems "hard" about what Sherlock Holmes would do is it's comprehensiveness - normal people only pay attention to the key elements in a social interaction, not all of the ancillary detail.  This of course is a key trope in the Sherlockian world, but it's very hard to do in reality, certainly for any length of time.  But a computer system could do it, being "always fully on".  Toss in facial recognition and the like and it might be very powerful.

The other element in the stories is that Sherlock had a sort of encyclopedic knowledge of arcane topics, with a library at 221B Baker St. to back it up.  Now we have the web, which might even be better.

Maybe such systems are out there for specialized use; not sure anyone would fund one for practical use, but maybe there's grant money out there... if not, it might be a great basis for a story.

Friday, July 15, 2016

The upcoming Olympics are to be in Rio; the news is worrisome.  Crime, disease, a marginally-functioning security and public infrastructure all raise questions about whether it will be a success or failure.  Thinking back, many cities have had "successful" Olympics, but big financial problems - it took Montreal 30 years to pay off the debts it incurred hosting the 1976 Olympics.

Why not have a permanent home for the Olympics, say in Greece, near Olympia?  Other countries could still act as hosts for specific games (and receive significant revenue thereby); but the permanent facility itself would be improved each time.  Ongoing maintenance would be much cheaper than build-from-scratch every four yearas.  Our current alternative leaves former Olympic sites without any clear purpose.  See, for instance, "What Abandoned Olympic Venues from Around the World Look Like Today".

Greece might not be enough; Winter Olympics, for instance, might also want a permanent home in a colder climate.  There might be other specific sports that need their own unique sites; but the principle of establishing a long-term venue still seems a sound one.

Tuesday, November 26, 2013

A retrospective bit on Obamacare

I remember a conversation I had with a Democrat years ago, when Obamacare first burst onto the national stage. I had expressed skepticism on a number of points, to which he replied (as best as I can recall):

"Bryan, you just don't get it. This will result in a permanent Democratic majority - and the fun part is that your tax dollars will be paying for it. In a little while the benefits will be so central to people's lives that [it] will be impossible to remove. Republicans will always be the ones wanting to limit benefits, so we'll always be 'the good guys'. Not only will people vote for us because they know we'll not cut what they need, but we'll be able to paint you as 'meanspirited' from here on out."

It was a very depressing thought, but it looks like it's not quite working out that way... 

Thursday, April 18, 2013

A term popped into my head the other day: "Theoretical Genealogy".  By it I mean the mathematics of populations - genetic diffusion, the probability of surname survival, how population growth skews the pool of descendants, etc.  This post is about just one topic, suggested by some recent research: diffusion.


While using Family Tree Maker’s search capabilities, I found I could add quite a lot of ancestors; there is no sound way to be sure they’re correct, and it’s hard to get an intuition about the reliability of the data, but it is a relatively rapid process.  As I chased lines back into the Middle Ages the doubts increased – but if nothing else the information may be used in the future as a starting point for, say, using DNA to verify links.  That is, it’s better than nothing, though perhaps not as reliable as I would like.

I noticed one phenomenon early on – many of my lines run through England (1600’s, 1700’s), and occasionally one of these would lead to some minor nobleman.  That person would in turn lead back to more important nobility, and finally to some recognizable king (e.g., William the Conqueror).  After this sort of thing happened a few times I began to be concerned that this really represented wishful thinking of someone, or some genealogist, along the line.  Am I really descended from Billy (above), as well as Charlemagne (several ways), along with various Viking and early Irish kings?  I didn't think so  because it seemed so surprising, and began that slightly unnerving process of figuring out where in the line of descent fiction crept in.

Now I’m not so sure.  The records of the minor nobleman and his ancestors seem fairly sound, I expect because property, wealth, and some level of authority changed hands with each generation.  The main weak point was the link between my known ancestry and that nobleman – that is a big jump.  Often it seemed to be the third daughter or so of the nobleman marrying my ancestor, who might have been a well-off farmer.  That sort of thing doesn’t seem unreasonable.  In the cases when this happened in the 1600's-1700's, sometimes coincident with travel to America, it was documented.  It might not have been were it just one more element of small town life in England.

Moving backwards in time from the minor nobleman the links seem fairly solid, for the same reason as above, it mattered at the time in substantial ways.  I’m sure there is questionable data in there – perhaps the occasional infidelity not identified as such ("Yes, he's your son, ignore the red hair") – but otherwise the genealogical data seems as sound as one might expect.

However, looking at the totality of the lines I’ve traced, I find I’m related to all sorts of kings and notables, in Ireland, France, Scandanavia, etc.  Is this reasonable, or is it the cumulative effect of past genealogists fudging the data a bit to claim famous ancestors?

To address this I turned the problem around, and imagined one of these notables (king, count, whatever) – they often had some semblance of wealth and a number of children.  The oldest few, particularly the males, would likely have married into other noble families, and it is from those lines that the people holding the noble titles today are descended.  The other kids did as well as they could, but often married significantly lesser nobility.  When this process repeated itself it eventually would mean some descendants were marrying commoners - perhaps on the well-off side, but non-nobility nonetheless.  This is not at all unreasonable - and if you think about it, the descendants of the original king/count/whatever would end up spreading out throughout the population.  Working backwards, then, as genealogists do, it isn't implausible that one might be caught up in that spread and then subsequently be led to the original notable as an ancestor.

What this really is is just genetic diffusion in the population, and when looked at from that perspective it seems unsurprising.  There have been recent articles about Genghis Khan’s descendents – apparently about 10% of males in the area of his former empire are related to him (or his family; one might imagine a brother, for instance).  As future generations arise the mixing will of course go further, until nearly everyone will be able to claim him as an ancestor.  Of course the number of generations back to Genghis is so large that the percentage of genetic data from him is quite dilute, and will become more so, but this is just the other side of the genetic diffusion coin.

Of course just as many people might trace their lineage back to Charlemagne, or Genghis, there are also some likely unknowns who cast similar genetic shadows over the future - some unknown peasant father and/or mother, who had numerous children, healthy, good looking, both sons and daughters, whose descendants spread out just as widely.  We just don't know their names.  The nobility, even minor nobility, shows up in church and civil records; the farmer doesn't.  Perhaps DNA analysis might at some point in the future reveal his existence, even if tentatively. 

(It is one of the striking things about doing this kind of research - how many people left so little trace other than through their children.  Even their names are missing.  I wish I had a at least a page, or even a paragraph, of information about each - I'm sure their challenges weren't fundamentally so different from ours.)

So - is there any way to quantify any of this, even crudely?

My background is European; the population of Europe in 1700 was about 50 million.  That is about the number of your ancestors back 25 generations (assuming no repeating ancestors).  At 20 years per generation, that’s 500 years.  By 'repeating ancestors' I mean no ancestor appearing twice or more - saying it that way makes it seem easy.  Turning it around - it means that say, at the 15th generation, two people who marry must have absolutely no ancestry in common.  I think this would be rather hard in a society where long-distance travel was rare.  The 500 year figure might well be half of that - 250 years - which, while long, is perhaps close to accessible oral history, particularly in fairly stagnant populations.  In either case, 300+ years on, a crude analysis suggests that pretty much anyone in Europe then might well be one of my (or your) ancestors.

The process of genetic diffusion is an intriguing one.  One might imagine as a simplified ideal a uniform population in which any person is likely to marry any other person in the population with roughly equal probability (excluding close relatives).  In that case the mixing will be maximized, that is it will happen as rapidly as possible.  Of course such populations may not really exist – social stratification, for instance, will lead to several (possibly overlapping) subpopulations that intermix that way, but not so much with each other.  Or there may be geographical separations that cause small ‘pockets’ of population to intermix internally but not externally.  I would think population genetics might be able to detect traces of such historical isolation from the genes and histories of people today.  It would take the right sampling to be able to draw solid conclusions, but it might be done.  While it seems an abstract notion, I think it might be practical, at least for recent cases.

For example, a number of my ancestors were among the early Dutch settlers of what is now New York City.  When the city became British in 1665, the Brits considered the Dutch second-class citizens, and intermarrying with them was rare; consequently the Dutch who remained ny necessity intermarried among themselves.  The end result is that, very loosely speaking, if you have a Dutch ancestor in that group it’s not unlikely you’re related to many of the other Dutch families who were present.  This phenomenon should be detectable by examining the DNA of descendants of that time – there would be areas of commonality due to that (socially) isolated population mixing as it did.  It would be more interesting to deduce such isolated populations entirely from the DNA as a way to augment history.  Doing it in the more distant past might depend on tracking mutations, and these might not happen rapidly enough to spread through a location such as 17th century New York City.  It might be different for, say, 6th century Naples.

The resulting model might be one of relatively static pools of population, connected by some punctuated diffusion.  This might not match historical reality exactly (say, an individual might have married into an immigrant family in his town, then his descendants might have moved back to the source of the immigration), but it might be a useful model nevertheless.  It seems ripe for mathematical modeling.


Saturday, October 6, 2012

The God Abandons Antony (C.P.Cavafy)

When suddenly, at midnight, you hear
an invisible procession going by
with exquisite music, voices,
don't mourn your luck that's failing now,
work gone wrong, your plans
all proving deceptive - don't mourn then uselessly.
As one long prepared, and graced with courage,
say goodbye to her, the Alexandria that is leaving.
Above all, don't fool yourself, don't say
it was a dream, your ears deceived you:
don't degrade yourself with empty hopes like these.
As one long prepared, and graced with courage,
as is right for you who proved worthy of this kind of city,
go firmly to the window
and listen with deep emotion, but not
with the whining, the pleas of a coward;
listen - your final delectation - to the voices,
to the exquisite music of that strange procession,
and say goodbye to her, to the Alexandria you are losing.

(Edmund Keeley, Philip Sherrard, translators)

Thursday, July 14, 2011

Thoughts on the Multi-Universe Interpretation of Quantum Mechanics

Just looking around the 'net I see much written about quantum mechanics that isn't, well, sound; I hope I'm not contributing to that genre, and fear that I might be.  I should start by saying I have studied it a bit, and have a degree in Physics from MIT, but I'm certainly not an expert, and many of its intricacies I'm sure lie buried by the detritus of years of other thoughts.  I'd like to lay out the outline of an idea, as much to get it down so it won't be forgotten.
Very briefly and somewhat loosely, the multi-universe interpretation of quantum mechanics says that everything that can happen will happen; when a random event occurs (say one with two outcomes), the universe splits into two universes - in one the first outcome holds, in the second the second outcome holds.  Wikipedia has a fairly decent overview here, worth taking a moment to scan.  At first this explanation sounds a bit extreme, wasteful in universes, so to speak.  But it does address some fundamental issues that are otherwise hard to make sense of, like quantum mechanical wave collapse (when an observation transforms a system properly described by a wave function to one described essentially classically) .  In the multi-universe interpretation, this never needs to happen, because each split universe has its own observer and its own result.

I have a small (and possibly testable) modification to suggest, which I'll get to shortly, after I describe its genesis - or at least the thought that provoked the idea.  It was years ago, and I was driving to work, and pulled into a bank near my office to get some cash - something I'd done many times over the years.  As I pulled in I saw a decent place to park about halfway down the lot, and a thought sprang up unbidden: "If I park there I'll be in an accident."  So, I didn't park there.  I went into the bank, stood in line, got cash, came out - and a car was in that spot, the driver exchanging papers with a second car that had hit it.

This was of course somewhat shocking - it'd never happened before, this isn't the sort of thing that occurs in my life (I can't think of another example, certainly not as striking).  I'm not a mystic, so I began to wonder how it could have happened at all.

I've always wondered about the instant we call "now" - a Euclidean point on the time axis, of zero thickness, has always struck me as somewhat absurd.  Einstein  reportedly said "There is no 'now' in Physics", meaning there is no model of it, no description of it, indeed it doesn't appear in physical theories.  If you give this a few moments of thought it's quite amazing - our only experience of life is in the instant we call 'now', and our physical theories don't consider it at all.
My thought after the bank parking lot episode, many years ago, was that maybe we don't proceed through time linearly moving forwards - on average we do, perhaps for thermodynamic reasons, but maybe we move forwards a little, back a little, oscillating about what we consider to be "now".  (If you like you could take the furthest point into the future we go and label that 'now', so all of this oscillation is entirely in the past - it is only a naming convention).  Moving backwards in time would change the physics of brain processes so that it would be very unlikely that we'd have any coherent memories of the future.  Occasionally these might persist, perhaps with enough internal consistency that they would be recognizable as useful information - so I might have been 'remembering' an accident, in a manner that felt like a strong intuition to avoid the situation.  This thought has been lying dormant for a long time - perhaps appropriately, as there seemed no way to test it or explore it further.
Back to the multi-universe interpretation: all of those splitting universes seem a bit unnatural.  What if, instead of splits, they are different excursions into the future, that is they all occur in this universe.  Let me describe this by analogy.  Imagine a snow-covered football field, with distance along the field analogous to time (think of one goal line being a few minutes ago, the other being a few minutes on, and where you are being "now").  You start walking at the first goal line, go a few yards, walk back, go forwards again, but perhaps not in the same route; you're moving down the field slowly, but making many tracks forwards and backwards as you go.  These tracks represent the excursions into the future mentioned above, each being something like one of the split universes; and, as you walk, you may interact with the tracks you've already laid down, stumbling over earlier footprints (analogous to the sort of effect one sees in the two-slit experiment, when a photon can 'interfere with itself' and seemingly go through both slits at once), or gravitating to well-trodden areas (analogous to high-probability outcomes).  Eventually you get to the final goal line, and can think over what you've done.
Wave function collapse reappears because we're back in one universe, but it might be a tiny bit clearer: Your consciousness holds memories, and those of the future excursions are very weak.  As your mind makes sense of where you are it sees multiple tracks behind it, as perhaps it should (see this interesting paper on the possible literal meaning of the 'sum over histories' technique:
I'm glossing over lots of details, such as the nature interference can have, what provokes the random wandering in time, etc.  Some of these may be fatal to the entire approach.  One implication of this idea is that the movements forwards and backwards in time are discrete (you will walk across a given yard line some integer number of times, indeed an odd number of times); given that interference between the excursions is possible, there might be an experimental way to detect this, how long they might be, etc.


Wednesday, December 8, 2010

The Ideal Minimall

Large supermarkets I don't really like, with the exception of Whole Foods - and even then there is much in the middle of the store (organic toilet paper and the like) that I have no use for.  I tend to shop daily now - life circumstances change fast, with kids doing this and that, so it is hard to plan long-term, and I don't particularly want to most of the time.

For a few years an idea has been bouncing around my head - the 'ideal minimall'.  It would have four separate stores as its core.  One would be a good butcher shop, with a knowledgeable butcher and a wide selection of meats - quality meats, meaning no hormones, antibiotics, etc.  Local organic meats would be ideal, but that's probably not practical.  I'd also like to be able to find the kinds of things that supermarkets no longer seem to carry, like bones for making stock, or the more unusual organs (for that occasional haggis craving).

Next to this would be a greengrocer, a thing so rare that the term seems to be falling out of use.  A place to get quality vegetables, run by a person who knows something about them.  I would add fruit to this store as well.

The third store would be a fish market.  We have some fish markets in the area, and given the proximity to the sea and the existence of a local fishing fleet, these are pretty good.  The selection can be limited in some, and in others the staff doesn't seem to know as much as it should.  One of my favorites was one in Waltham, MA (I don't remember the name) which seemed to have 3 or 4 people working at all times, had a large selection, and was a place one could buy things like fishheads or lobster bodies for stock and stew purposes.  Of course there has to be enough business to justify the stock size, and enough turnover to maintain quality, so this might be difficult.  One think I love about Whole Foods is their seafood, which seems to be of high quality at every store - so I'm sure it's doable.

The fourth store would be a bakery.  Supermarket bakeries seem to bake premade or preformulated mixtures - I'm sure it guarantees uniform quality, but that quality isn't very high.  One of my fond memories from childhood is walking into a (good) bakery, and smelling the buttery sweetness that seemed to hang in the air.  A good bakery should do the full range of baked goods, from breads to cakes and pastry. 

Of course all of these stores would be able to handle custom orders, and ought to have knowledgeable staff that could offer advice as necessary. 

There might be other stores as well - a liquor store (perhaps more focused on wine than hard liquors) would be a nice addition, and a cheese/dairy outlet would be a fine addition, particularly if the products were from local farms.  A coffeehouse at one end of the minimall might provide a nice gathering place as well.

One could drive in to such a minimall and stroll from store to store, assembling a dinner.  One would likely have to make occasional trips to supermarkets for soaps, napkins, etc., but I think this sort of minimall might be quite an attraction.  One way to do it would be to find an investor willing to put up the money to buy or build the physical infrastructure, then lease to the individual markets, perhaps also with some equity in those businesses as well.  Such a plan would have to cover the contingencies of a market failing, or key personal leaving, but that shouldn't be insurmountable.