Thursday, March 15, 2018

Do indexes dream of humans?

I tweeted a link to a book indexing conference with a comment about indexing maybe being the stuff of thought. Reiver kindly replied with some tweets on how search and indexing were viewed in computer programming. So here I'm trying to explain my thoughts more thoroughly, and then addressing the difference in computer programming. The follow are my mild musings, wild speculations, and some thoughts whimsical. I welcome constructive comments.


Imagine a host of nodes, or if you prefer something more concrete, biologists.
They are using many terms which have the possibility of having different meanings to one another.
The meanings are already there they just aren't evenly distributed, or consistent.


I've heard a possibly apocryphal story about L. Wittgenstein that after a conference in exasperation he said it sometimes seemed to him (some very high percent) of philosophy was discussing word meanings. (I've heard different versions about just how high a percent.) Wittgenstein explained language as a kind of eternal guessing game where participants tried to discern the meanings of others. It seems simple and natural to humans but is in fact extremely vague and complex. We would now call it a kind of instinct blindness. Like how fish minds think about swimming in water, of course everyone knows how to do it - it's so simple!


Chomsky, perhaps inspired in part by the emergence of computational linguistics around the 1960's, argued against Skinner's Verbal Behavior book that the possibilities for babies learning the meanings of word were under-determined. That there was a 'combinatorial explosion' of possible meanings babies could ascribe to teacher-behaviors and that there must be some kind of cognitive module providing a scaffolding for grammar and the learning of word meanings.


Pinker in The Language Instinct explicitly discussed such scaffolding as evolved, and in the context of reasoning instincts - humans are more capable of reason because we have evolved new instincts, not because we have lost instincts.


Imagine a host of biologists, before Carl Linnaeus. There are many taxonomies, often conflicting, designed around different ideas of what it is best to optimize (and around scientists' egos too no doubt.) Eventually the systems merge or become forgotten and what remains is the Linnaean system. (One might alternatively look at the nomenclature rules in chemistry, for example.)


At one level of description, indexes are pages at the back of a book with names of ideas and where those ideas are discussed in the text. Rephrasing, such indexes are collections of pointers to the meaning of ideas.(Aside: A very common index item is ".. (word), definition of" ..) It seems to me that in a very real sense scientific taxonomies are also collections of pointers to ideas. Only they are guides to how to converse rather than to a page where an author discusses an idea.


Okay then, that is all at the level of communication between individuals. What about within?


"Neurons that fire together wire together." - quip about Hebbs' learning rule (1949.) But a quip does not an explanation make. But even if it did here, I'm more interested in meaning and communication theory than the rabbit hole of neurology, at this point.


How do different parts of the brain know they are communicating about the same thing? Currently brain waves are being touted as the unifying principle. Okay. They probably have something to do with it. Maybe a lot. But I'm much more interested in functional explanations than mechanical ones.  More ultimate causation and less proximate causation. Sometimes you can understand all the proximate causation you want, but still not understand the functioning. (Which is probably a good way of roughly describing the state of knowledge today about individual neurons.)


There have been some explicit comparisons of the hippocampus as a kind of an index to the memories in the brain. (Sorry, no cite on hand. Only general memory.) I like this comparison. It makes sense to me that the index should be in a particular place, centralized, in one of the most phylogenetically ancient structures of the brain. I like to imagine a brain wave travelling across the hippocampus like the eyes of a reader travelling down the entries of an index.


So this is what I implied about indexing being at the heart of the thought process - The meanings of a concept are distributed in different brain regions, like discussions in the text of a book, and linked together and accessible from different small areas of the brain's index. This is speculation. Or a thought in process perhaps.


(By the way, the part of the brain that includes the hippocampus continues on to form the amygdala, which has over the decades been a subject of much interest in brain studies of autistics. I've wondered just about as long if the claims around enlarged amydalae in autistics have been somehow overlooking enlarged hippocampi in autistic savants.)


Now as to what Reiver tweeted about the terms search and index in computer science. First of all, thank you. That's important information. By the way, cognitive psychology which often draws on computers as metaphors for the human brain, seems to define the terms similarly.  (Search seems to be often operationalized in cognitive psychology as eye saccades, or some such thing.) Defining the terms that way is very useful, particularly in computer science. They are defined in terms of discrete proximal operations by actual equipment and registers you can point at.

But for me they are the same function, only one is an accumulated operation of the other. In short, roughly speaking perhaps, indexes are accumulated searches. The reason why indexes make searches faster is because they have done all the searching beforehand. Perhaps I will think on this more and in the future come to a different view. But for now, some metaphors -

In psychometrics the general intelligence factor is sometimes broken down into crystallized intelligence and fluid intelligence. (Gc and Gf.)

In accounting you have Income Statements covering a period of time, and Balance Sheets at a point in time. Usually the Retained Earnings section of Balance Sheets shows the change in retained earnings from the prior balance sheet, in effect summarizing the income statement. Roughly speaking, Balance Sheets are an accumulation, summation, and transformation of all the previous Income Statements.

Maybe all those metaphors don't make any sense. Maybe they don't even make that much sense to me. But hopefully they help.

Perhaps explaining a little something about how indexers index might help. Typically they carefully read through the text (sheets from the publisher sometimes) and mark up the likely index main headings and sub-headings directly on the sheets. Then they type the headings and the page numbers and such into one of the professional software programs, which is mostly a sorting program (but not entirely.) Those are the easy parts. Then comes the editing. Deleting and re-organizing old entries, looking up new entries, over and over and over.  I knew one highly experienced indexer who said he usually spent twice as much time editing as all the other tasks. The point is that the first part of indexing is search in reverse. You are at the location where the topic is mentioned, and you then sort that into the constructed index. The editing process involves looking at the collection of links/locators/locations in the initial index, and .. working the indexer magic to make sense of it and make it useful.

By the way, my impression is that In psychometrics people still write about Gc and Gf in the journals, but it doesn't seem they do it as much as they used to. I suspect it's because the tools being used to measure general intelligence tended to fall into the categories of measuring by way of crystallized intelligence (mostly vocabulary or world-knowledge tests) or fluid (computations, insights, etc.) Those issues don't seem to dominate academic interests the way they used to.

And that is why for me the concepts of search, index, and thought are all right next to each other in my hippocampi.

And now some things whimsical.

Why are back of the book indexes organized alphabetically? Why are they not organized more like a detailed table of contents, according to the structure of the material in the author's mind? It seems to be because the reader's may have a different sense of how the meanings are related to one another, and what would be a logical ordering to the author is not necessarily the same for the readers. Of course there is always the table of contents there at the front of the book for the reader to access instead. But I find it curious that the injection of a kind of randomness, a somewhat arbitrary ordering of the meanings in a text, can actually improve the communications from the author to the reader. (Of course, being alphabetical makes it easier to directly look up a term of interest. But again, the term of interest then is de-contextualized from the overall structure of the meaning of the book as the author sees it.)

Do hippocampi dream of editing indexes?
It has often seemed to me that the accessing of never-before accessed old memories results in very clear images at first, but they rapidly degrade. What if dreams are the by-products of accessing memory components in the process of editing out the old no longer accessed memory index items? It might help explain why so many dreams go unremembered.

Monday, October 30, 2017

The Clintonites are continuing to destroy the Democratic Party.

I've been puzzling why the Democrats have been going all-in with the campaign to paint Trump as a traitor. (The Dems in general and the Party-friendly Brookings Institution in particular.)

Hillary ran against Trump's personality, leaving the campaign issues open for him. -A disastrous overall strategy during the 2016 election. Normally for a first term President there would be some reaching across the aisle to the other Party to develop new programs. By continuing an all-out attack on the person of Trump the people at the top in control of the Democratic Party prevent any moderates in the Democratic Party who might be willing to work with the Republicans toward less extreme Republican policy positions. They keep control of the party by branding anyone who would vote with Trump's policies as failing to work together to brand the administration as traitors to the country. Even if Hillary doesn't run again in 2020, the Clinton faction will most likely be able to choose one of their own as the candidate. There's a lot of love for alternative candidates outside of the Clinton circle of power but let's face it, that by itself doesn't win political primaries.

The Democrats don't have enough support in Congress to impeach the President, even granting the charges of accepting Russian support and intelligence against Hillary are wholly and entirely true. And even if they are much more than the amount of involvement indicated so far, which is pretty minor compared to regular US involvement in foreign elections of other countries over the last century, including very probably Russia itself. Some history -- Nixon was impeached because (.. considering using all capital letters here....) he made an enemies list that included sitting congressmen and was trying to use the IRS as a political weapon to go after those congressmen (and if memory serves) some of whom were sitting on the very hearings committee to investigate the Watergate break-in. The political campaign 'dirty tricks' of Donald Segretti and others, the break-in at Watergate, the suppression of the Pentagon Papers, were very big deals but even all together it wasn't entirely clear a vote recommending impeachment hearings would have been fowarded to the full Congress.

I grew up as a Democrat. Most of my life I have been a Democrat (though I consider myself an independent now.) It pains me to see what has become of the Democratic Party under the current leadership. (Which to a large degree I think goes back to the 1974 special convention to deal with the aftermath of the 1972 McGovern debacle. The book to read is - The Whole Damn Deal: Robert Strauss and the Art of Politics.)

Most voters didn't think much of the Watergate break-in. Most voters don't think much of the Russian involvement in the 2016 campaign either. Like the Watergate hearings, most congressmen don't think much of the Russian involvement either. They haven't been put on any enemies lists, unless it an enemies list of the Democratic Party leadership.

There's the old idea that if the weather is good then the crops are good and the King is well loved by the people. There are many reasons why the US economy should do well over the next four to twenty years. China's economy is probably over-stretched. The shale oil and gas revolution are likely to keep the US energy independent for a long time. (The video to watch is ... global strategic analysis - Peter Zeihan on “The New President & the World .. ", https://vimeo.com/215078563  .. 68 min.s. Agree with almost all of it, but one big caveat is China's drive already underway to create a major high speed rail network from Beijing to the Baltic.)

Predictions: In the absence of any major changes (a big IF) the Clinton loyalists will remain in power and get one of their own nominated for the 2020 election. The 2018 mid-term elections will largely be a replay of 2016, as will the 2020 elections. (This doesn't make me happy. Trump brings change but he doesn't seem, to me at least, like he has a sense of overall strategy, or even worse, is trying to shoehorn everything about running a presidency into a standard simplified corporate business model.)

harpersnotes

Friday, June 30, 2017

Soil biology as a subject and it's informational structures and flows. Part I. 

During the early 1990's I happened to be present during the founding of a new academic field. A sequence of steps is usually involved -- Small meetings. Then meetings as a track at a larger academic organization's conference. Then small one-day mini-conferences. Then a journal for publishing results of studies done and of studies presented at the conferences. Then conferences of larger sizes. Then the full works -- award ceremonies, funding announcements; job announcements; web presence and so on.

So now when I look for the information structure and flows of a new area I'm interested in I look for those sorts of things. So I went looking for the equivalent for composting, or at least some near-equivalent. There were several dead ends to searches where the organizations I found were mostly set up to disburse information or sell various useful products.

The best I found, most relevant to what I was searching for seems to be the --

United States Composting Council (USCC).
   Mission Statement: http://compostingcouncil.org/mission/
   The official magazine is Biocycle: https://www.biocycle.net/ ..
   Biocycle Events Calendar: https://www.biocycle.net/conferences/event-calendar/
   The 2017 conference is in Portland Oregon starting October 16th, http://www.biocyclerefor.com/
   Previous conferences (scroll down). https://www.biocycle.net/conferences/about-biocycle-conferences/

Sampling the Tuesday presentations of the 2016 conferences:     http://www.biocyclerefor.com/2016/tuesday.html
 
   It turns out you can download the pdfs for the talks apparently but when you try to open the pdfs on your computer you can't access them without a password. I tried searching unsuccessfully for a couple of names to see if they'd posted previews of their talks elsewhere on the internet. In some fields that happens a lot but probably not so much here.

So I searched World Cat for Biocycle.
   https://www.worldcat.org/title/biocycle/oclc/494598926&referer=brief_results
   .. Which indicates it is accessible at most of the academic libraries nearby: UNM, UNM HSLIC, CNM, and some others further from where I live.

   I used to go to the New Mexico Health Sciences Library and Information Center on Sundays because it is usually open from noon until late and the parking is free. Ideally one of the places has actual physical copies of Biocycle and the conference proceedings books as I do so much more enjoy browsing through those.

Richard Harper aka harpersnotes

Saturday, January 28, 2017

The limits of biological designs

Allometry, biological constants, and theoretical limits to human IQ augmentation. .

Karlin tweeted - Are there any good articles making the case that IQ augmentation thru GWAS/CRISPR will be *harder* than Hsu/@razibkhan/@charlesmurray think?
Source: https://twitter.com/akarlin88/status/824990887977549825

That discussion thread mentions Kevin Mitchell's piece on mutation load (genetic load) as decreasing human intelligence from some ideal (Platonic form?) That is the first idea that came to my mind and it almost certainly plays some large role. But I'd like to suggest another way of looking at limits to IQ augmentation.

Allometry scaling is a way of getting at biological features that are constant across species by adjusting (typically) for body size. (Allometry seems to have some of the appeal in biology that physical constants do in physics.)

For flavor.
Sizing Up Allometric Scaling Theory (2008) http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1000171
Wikipedia for general background, https://en.wikipedia.org/wiki/Allometry

Wikipedia mentions (quoting here)
West, Brown, and Enquist in 1997 derived a hydrodynamic theory to explain the universal fact that metabolic rate scales as the ¾ power with body weight. They also showed why lifespan scales as the +¼ power and heart rate as the -¼ power. Blood flow (+¾) and resistance (-¾) scale in the same way, leading to blood pressure being constant across species.[31]

The limit on total lifespan (i.e. maximum lifespan) of species being scaled to body weight suggests a limit across species on the evolutionary design of heart muscle (or cardiovascular systems.)
(Which by the way is why I find hummingbirds interesting, for if any species has broken the design barrier on hearts it is most likely them.)

It doesn't seem implausible to me that in the evolutionary design process of human brains those that are functioning at around 160 are already at or near the limit of maximum biological design, in the same way as heart muscle tissue -- Though with the difference being that humans are the solitary species that have reached such a hypothetical limit first. (I tend toward skepticism about the ability of current IQ test designs to meaningfully discriminate between IQ's in the highest ranges.)

Friday, January 6, 2017

Ottoman Millet System

My very quick and tentative take on millet system threads on Twitter. Caveats (see last part). 1820's. Royal court re-takes absolute power. Embarks on many expansionary wars. More territory, everything else (e.g. trade) worsens. Diminishing returns to millet system. Too many guns, not enough butter. Worsening resentment over time. Baki Tezcan might frame as a burgeoning proto-constitutional monarchy with strongly growing trade replaced by old system of royal absolutism and extractive militarism. I'd add that unlike the millet system trade can better help forge a national identity. The millet system 'national' identity was largely limited to religious, Sultan worship, and the military. http://www.cambridge.org/us/academic/subjects/history/middle-east-history/second-ottoman-empire-political-and-social-transformation-early-modern-world  

(Caveats: I spent three months in Istanbul 1989-1990. Read Bezcan's book, more than once. Have read a lot of history over the years. No professional status as historian or directly related to topic at hand.) 


Tuesday, December 13, 2016

Too Many Chess Players

Too many chess players spoil the strategy.

December 13, 2016

It was a common thing at the club when I was young - to pit a chess master on one side of the board against many on the other. The many often included at least a couple of chess experts as well as A and B rated players. The intuition of most people is that it should be an even match, or that the many would have the advantage, but the master always won. With lots of eyes on the board the many would catch all the tactical shots often even better than the master. But the long run strategy was consistently inconsistent. First one of the many and then another would persuade the rest on following one course of strategy and then on a different course of strategy. So they lost. Too many chess players spoil the strategy.

It has become for me a broad metaphor that I see in many different forms. There are the political implications of course such as in international relations and in trying to explain why the US Presidency has over time become increasingly about foreign policy -- you need a consistent strategy for winning The Great Game.

Today I saw an article about an interview with Marc Andreessen in the Financial Times. He talks of how investing has become more risk-averse, comparing the declining returns over time to stocks in new tech companies. He talks of how there are many who can individually say no to going ahead with a technology investment idea. So naturally I think, 'too many chess players!' Each of those who can say no has a range of acceptable risk they are willing to allow. In order to put together enough funding to ensure an acceptable probability of success the innovator has to satisfy the most risk-averse of those who can say no. The end result is to end up funding only according to the most risk-averse of the many.

And what about the 'wisdom of the crowds'? There is a lot that has been written about that now, but I'm only making a minor comment in that direction. At it's simplest level and as originally conceived the wisdom of the crowds is about taking the average of the individual estimates of something like the weight of a steer at a farm festival contest. I think what happens in chess strategizing and in what Andreessen is describing is that what is being sampled is the most risk-averse of the crowd.

All metaphors are not exactly the thing they are being a metaphor for, so of course all metaphors break down, which is for me when they become even more interesting. Then it becomes about the skillful use of the metaphor as a tool best suited for particular tasks and the search for new metaphors can more effectively begin. If those who can say no in the tech-investing case were to be restricted in their communications with each other, and in effect only allowed to do a silent bidding auction of whether to invest or not, it would probably make the most risk-averse less likely to discuss and move more toward the center. But in the case of chess, .. well here it's probably best to imagine postal chess - an pre-internet form of extremely slow chess still occasionally practiced where people living in remote areas find suitable competition by engaging in play-by-mail one move per one postcard. If the many were to first send a separate round of postcards to each other on the overall strategy they thought best for the next three moves or so (control the center versus immediate flank attack for example) and then vote, that might actually improve the strategy.

harpersnotes

Links:

The Great Game at Wikipedia (Central Asia) https://en.wikipedia.org/wiki/The_Great_Game ..

Andreessen interview discussion nub, https://www.isegoria.net/2016/12/the-financial-times-has-lunch-with-marc-andreessen ..

Those who say no (The Knights who say Ni! Monty Python) https://www.youtube.com/watch?v=KHqy_AyKIUI ..

Friday, October 14, 2016

Obese Thoughts


The conflict over whether to cut sugar or cut fat may become more clear if we control for the patterns of eating that provide cues which in the environment of evolutionary adaptation reliably predicted famine.

Here is some of my current amateurish thinking on obesity. Largely this is just an expression of the idea of the thrifty gene but perhaps more in the context of adaptationist thinking. Someone has probably already said these things.

Organisms are adaptation executioners. Genes have been selected upon to respond to cues of information about the environment of whatever the current replication-vehicle host is and to express accordingly. Famine cues include the number and intensity of hypoglycemic events. If famine cues are sufficient, a famine adaptation response might be triggered. Once triggered it is likely active for the rest of that gene-host's lifespan. (Here skipping some possible one-generation carry forward epigenetics discussion here through plasms of eggs and sperms in the host body.) If the famine adaptation response is not activated and there are sufficient fat stores then in times of low blood sugar burn those stores. Alternatively, if the famine adaptation response is activated then in times of hunger lower the metabolism rate and increase food-seeking behaviors ('hunger', 'appetite'.) Note that ingested fat generally burns slowly allowing a steady supply of energy, and low fat diets allow for more and for greater sugar-carb roller-coaster hypoglycemic crashes. One way to test this idea might be to compare populations for the amount of consumption of non-fat and low-fat milk, and perhaps butter too, in terms of subsequent development of obesity.

(Does some quick searching ... )

Of course there's a long history on whether non-fat milk is good, but in any case --

This study looking at dairy consumption and obesity got a lot of publicity back around April of this year. https://www.ncbi.nlm.nih.gov/pubmed/26912496 .. Paywalled.

The Time magazine article that discusses it some: http://time.com/4279538/low-fat-milk-vs-whole-milk/ ..