Albert Einstein on Capitalism

Perfect synopsis.

“Private capital tends to become concentrated in few hands, partly because of competition among the capitalists, and partly because technological development and the increasing division of labor encourage the formation of larger units of production at the expense of smaller ones. The result of these developments is an oligarchy of private capital the enormous power of which cannot be effectively checked even by a democratically organized political society.

This is true since the members of legislative bodies are selected by political parties, largely financed or otherwise influenced by private capitalists who, for all practical purposes, separate the electorate from the legislature. The consequence is that the representatives of the people do not in fact sufficiently protect the interests of the underprivileged sections of the population.

Moreover, under existing conditions, private capitalists inevitably control, directly or indirectly, the main sources of information (press, radio, education). It is thus extremely difficult, and indeed in most cases quite impossible, for the individual citizen to come to objective conclusions and to make intelligent use of his political rights.”

— Albert Einstein

Calcium

[Not intended as medical advice, see your doctor for that.  I don't know your situation at all, which may preclude some of the ideas here, especially if you are or will be taking any kind of medication.  Likewise, see a Nutritionist for nutritional advice targeted to you.  As always, my posts are intended to be thought provoking, but never the fully edited final word on anything.  I am not an encyclopedia and I don't have time to go back and fix everything I've written as new information comes in.]

Back in 2017, a friend of mine in her 50's got osteoporosis and started a treatment program.  For a few years she had to take Fosamax, which helps make weak bones more rigid but also can make them more brittle than normal if you take it too long.  The best thing is to reach the point where you don't need Fosamax any more in a few years and from then on get the proper nutrients every day for normal bone remodeling which will, over time, restore full bone normality.  Fosamax is a band-aid you don't want to keep wearing.

This made me start thinking about my bones and took a look at the Recommended Daily Allowance for Calcium.  I was shocked to see that it was 1,000mg for adults and 1200 mg for seniors.  I thought I was getting plenty of calcium from (then) 2 cups yogurt, but that was less than half of the RDA.  And not much else in my diet had calcium.  My multivitamin didn't add much.  So, I started supplementing calcium with a single 330-500 mg Calcium/Magnesium tablet every day to fill the gap.

Visiting my sister and brother-in-law in 2019, my attorney brother-in-law noticed my supplements, and gave me a scathing critique.  "You should not be supplementing calcium," he said.   "You'll get atherosclerosis.  You'll get kidney stones."

I replied, "You only get those things if you are out-of-balance somehow.  Meeting your minimum daily calcium requirement is not out-of-balance, it's in balance."  I could not change his mind at all.  He was highly opposed to supplements almost in principle (as it seems many doctors I've known have been).

I was very concerned about him that since he is lactose intolerant and avoids dairy altogether (except 2 cups of special ice cream with lactase for lactose intolerant people) he was not getting enough calcium.  And my sister too, who did have small pieces of cheese with her salads but less ice cream.  They were both already in their 70's and starting to look like they were rapidly shrinking.  I've tried and tried to get my sister to consider supplementing calcium but it's never happened so far.  She'd have to face her husband's disapproval for one thing.  And for another, she doesn't like to do anything technical, like calculating the calcium supplementation required, preferring to leave everything technical to her husband.

Nine years later now they both look extremely shrunken.  My sister has constant neck and back pain and her neck has become flexible.  Her two upper front teeth simply broke off one day and she had to get a partial dental.  The shrinking is not just in statue but in depth.

Everywhere I go I see elderly people who look like they do now.  I doubt any of them are getting enough calcium.  Few seem to even be aware that seniors need more calcium than adults and how much that is.  (BTW, this is a well known fact that most US seniors are not getting enough calcium.)

Retrospectively I now blame lack of calcium for my short stature.  In freshman year of high school, a friend of mine convinced me that drinking milk was for babies, and I quit doing so.  I really didn't have other significant sources of calcium in my diet then.  I didn't regularly eat yogurt or ice cream.  That was surely bad advice.

*****

Doing this can get a little more complicated than people might like.  You should not take simply take supplements having the RDA for calcium.  You need to figure out the gap between your diet and the recommended amount for your age, and then fill just that gap.  You don't want to get too more more calcium than the RDA because then you could get kidney stones, etc.  

But there is some slack here.  You can have somewhat more than the RDA without issues, so long as you stay below the upper limits.  It's when you go past the upper limits that you could get the kidney stones, etc.  The upper limits are a bit more complicated since they vary for young people, pregnant mothers, adults below 51, and adults above 51.  For non-pregnant adults under 51 the upper limit is 2500mg and for adults over 51 it is 2000mg.  Thus the range for eldery people is the most narrow: from 1200mg recommended to 2000 mg upper limit.

It's not really hard to do, though it takes a little time and arithmetic.  Nowadays you can simply ask your favorite search engine for the calcium in any kind and quantity of food.  You could probably even get it to add up the calcium in a long list of foods that constitutes what you eat in a given day.  I do the adding up myself.

Better than using a search engine is actually looking at the Nutrition Facts on any labeled product.  Then you get the exact numbers for the specific products you are eating (though lots of fresh foods are not so labeled--there you need to use the search engine).

I wrote everything down on an index card with columns for calcium, protein, and sodium.  It's a worthwhile exercise to do at least once, I wish I had the time to do it every day.

I don't recommend the example below as an ideal diet for anyone, even me.  I'm more interested in showing the method of adding up the calcium including everything.  This just happens to be what I ate yesterday when I was keeping track of everything, everything that went into my mouth.

It might seem like "too much" peanut butter but you may notice that the total saturated fat is still fairly low because no eggs, little nuts, little meat, and no red meat, at least yesterday.  Anyway the most recent research seems to show that dairy and peanut butter saturated fats are actually somewhat protective, but the saturated fat dogmas haven't been updated to reflect this.  I like natural peanut butter (nothing other than peanuts and salt so you have to stir it) because it's convenient and almost as much protein as eggs.   Yesterday I didn't get any of the most dangerous saturated fats, those from red meats, but sometimes I do have 3-4 oz lean steak instead of chicken, which still wouldn't be very much.  I've also switched since writing this to snacking with regular or low sodium Ezekiel bread, I only have regular bread (usually the whole wheat variety) for breakfast.

I have never had a bad A1C measurement but recently I've become very focused on keeping it low because I believe it is actually the number one factor in heart and other diseases for elderly people.***  You will note that there's very little sugar in my diet now.  I do use orange juice but only in the very tiniest amount for flavoring.  I eat a tiny bit of sugar and a tiny piece of chocolate.  I need the artificially flavored electrolyte drinks right now because I have diarrhea from taking an antibiotic.**  I didn't have any fruit mainly because that's not compatible with current diarrhea.  Normally I'd eat a fresh apple or banana.  On nutrition focused friend of mine thinks its pointless to eat fruit at all but I find the fiber beneficial.

Food / Calcium

1 oz Orange Juice (mixed with no sugar Citrucel) 2 mg

4 large slices Dave's White Bread            80 mg

4 tablespoons all natural peanut butter     36 mg

7/8 cup Chobani Nonfat Greek Yogurt    225 mg

1 scoop whey isolate                                105 mg

Centrum for Men 50+                               210 mg

2 Ultima Electrolyte Drinks                     94 mg

3 oz cooked chicken breast                       10 mg

3 oz Protein Pasta                                      0 mg

8oz can no salt added green beans            40 mg

2/3 11.5oz can low sodium V8.                33 mg

2 Tbsps Olive Oil                                        2 mg

1 Tbsp Heavy Cream                                   10 mg

7 cups RO water (coffee, mixes, ice)           7 mg

(Local Tap Water would be 105-140 mg)

1/2 cup Low Sodium Marinara Sauce.        20 mg

1/5 oz (one tiny square*) Chocolate w nuts  4 mg

3/4 tsp Sugar (in whey drink)                       0 mg

Before calcium powder: 876 mg calcium

1 tsp Calcium/Magnesium powder            267 mg

Total Calcium: 1143 mg

I was a bit below calcium target, but not far.   I was a bit too conservative when measuring out the Calcium/Magnesium powder.  I didn't actually know how the numbers were going to work out.  Going forwards, I think I may use the 1 1/2 tsp suggested amount.

That powder was my calcium supplement yesterday (sometimes I take a 333mg pill instead, which would have put me just slightly over 1200 mg).  But I'll explain below why I think the powder is better than nearly all the pills.

The total 'supplements' contribution to my calcium was 567 mg (multivitamin, electrolyte drinks, calcium/magnesium powder).  Almost half.  And that's in spite of eating slightly more than the standard serving of yogurt and about as much calcium as most people in US eat.

I love to have a spaghetti dinner every day and I think that's fine with exact portion control (pasta is weighed on my scale) protein pasta and low sodium marinara.  Pasta itself is zero sodium (and I put no salt in the water) which helps stay within optimal sodium. I do aim to cut back the extra two pieces of regular bread and peanut butter but it was the only snack I had available to go with my whey drink in the afternoon yesterday.  Today I had the low sodium Ezekiel bread with the lightest rub of peanut butter possible.  Going forwards, I plan to use low sodium Ezekiel Bread for snacking, with less peanut butter, and so reduce the daily peanut butter to 3 tablespoons or less.

Good calcium supplements (including pills and powders) are are sold with Magnesium included because you need to have calcium and magnesium in balance.  The recommended balance for calcium:magnesium is 2:1, so many pills have this exact ratio.

But it actually doesn't make sense if most of your other calcium comes from dairy, because dairy products are 10:1, the supplement should try to bring that back into balance.  So a 1:1 calcium magnesium supplement powder makes sense for me whose other calcium comes mostly from dairy.

The 1:1 Calcium Magnesium powder I get (NOW) is also made with citrates, which are more easily absorbed than the calcium supplements made with carbonates.  Most calcium magnesium pills are made with carbonates.   Adding it to my whey isolate drink makes it seem less watery and more like milk.

(* It wasn't long enough ago that I couldn't help myself but to eat a whole 3.5 oz bar of chocolate every day.  My recent near death experience with pneumonia and dehydration has made me a different and better behaving person now, and hopefully from here on.)

(** The day I started taking electrolytes the situation dramatically improved.  It's as if part of the problem was the interplay between the anal and urinal sphincters, which are connected, and of course are connected to the nervous system and all of its electrolytes.  My antibiotic lists no food or supplement interactions but out of an abundance of caution I apply a 4 hour spacing rule used for others, no supplements within 4 hours of taking the pill, before or after.) 

(*** From NIH published research I've read, the biggest risk factor for heart disease in elderly is in insulin resistance, aka type 2 diabetes, not in fats****, anyway.  The saturated fat dogmas haven't been updated to reflect this.  Lots of my thin elderly friends on lowfat diets are still taking statins, which I wonder about, then stuffing themselves with candy.  I think they should be more focused on insulin resistance, as I am.  I quit taking statins when I retired and could do my own low fat cooking instead of eating at Chinese buffets and steak bars every day. When that happened, I started noticing that statins made me feel weak  My doctor wanted me to continue statins anyway, perhaps a different one, looking at my BMI category, she said my LDL was still a few points short of target.  I don't think BMI reflects my stocky and above-average muscular body style correctly, I see Percent Body Fat, as I get on my biometric scale, a better indication of how much excess fat I have, which is as I understand it the relevant factor for heart disease.  I have always done better on percent body fat than BMI, and as of right now I am no longer even "obese".  Right now I can't seem to find the very impressive NIH published research I read last year on insulin resistance vs hyperlipidemia that put insulin resistance on top for elderly people while discounting hyperlipidemia a lot, but there is much research and writing that makes a similar case that can be easily found, such as https://www.optimalwellnessmd.com/blog/uncategorized/atherosclerosis-is-it-due-to-lipids-insulin-or-inflammation which mentions many studies.)

(****It bugs me no end that most people have come to regard low density lipoproteins (the entities, not the LDL relative blood measurement) as diseased.  In truth, low density lipoproteins are the essential lubricant the body needs like an engine needs motor oil.  You could not live without them.  High density lipoproteins are like the oil additive that keep it flowing smoothly under different conditions.  You can have too high HDL--when it keeps the low density lipoproteins from doing their job.  Not that you need to eat saturated fats, the body can synthesize LDL's.)

Here is another (Vegan) POV on calcium.  They claim the US RDA are inflated because of high sodium and animal protein diet.  May be some truth to that, if your sodium is lower and your animal protein* is lower it not be necessary to achieve US RDA but 300mg or so lower.  (This is still more than many Americans, especially seniors, get.). (*"animal protein includes dairy, which is probably the #1 dietary source for most people, including me. If you're Vegan and not consuming dairy, you will almost certainly have to supplement even if you follow their lowered recommendations.  Vegan recommendations also therefore do not apply to people like me, even though I use relatively little meat, I still seek high protein which increases the need for calcium.  So the Vegan POV does not apply to me and I'm back to RDA.)

https://www.vegancanada.org/news/article/2020/01/23/misleading-calcium-recommendation.html?gad_source=1&gad_campaignid=18597142512&gbraid=0AAAAAC2laoEA6W5grs7fEVO0TycsLLRLX&gclid=Cj0KCQjwr4jSBhCSARIsAOX1E-Iq_SgorrjeBbbfbP3oh2iB5ATOWisbL-ufLmJ7iQPrAGFlD7lKPLcaAnNpEALw_wcB

Another caveat is that one shouldn't have more than 500 mg calcium at one meal.  My plan of having yogurt plus multivitamin (totaling about 500 mg) in the morning, then whey drink with calcium supplement (totaling about 500 mg) in the afternoon or evening already does that.

Truthfully, I was much less careful about the total amount of calcium back when I argued with my brother-in-law in 2019.  That argument got me to start thinking about how the supplements add to natural sources so I only need to supplement the gap, and so I've been more careful about supplementation since then.

Petroleum Jelly

White Petrolatum, also known as Petroleum Jelly, often recognized under the trademarked name Vasoline, is one of the most useful skin protectants.

Vasoline has always been sold to consumers in small tubs.   This makes zero sense unless the goal is to sell as much of the stuff as possible.

With a tub, every time you use the stuff you dip your finger in the tub, pulling out a small bit for whatever purpose, you leave all sorts of germs, skin oils, and other crap (sometimes literally) on the remaining stuff.  After I've done this for awhile, I'm motivated to buy a new tub next time, wasting most of the earlier tub contents.

OR, if I use one tub of Petroleum Jelly for mechanical purposes, such as lubricating the corner of the door threshold, I'm motivated to get another tub for personal uses, because I don't want any of the crud from the doorway getting into my personal care.

It's much more reasonable to have Petroleum Jelly in tubes, allowing you to squeeze out exactly what you need and leaving the rest unadulterated.  This is the way it is sold to dermatologists and other medical doctors.

But since Vasoline was first sold to consumers in tubs, virtually everyone else who sells Petroleum Jelly in a retail store sells it in something like a Vasoline shaped tub so they'll recognize what it is.

The now disappearing Walgreens was the one retail store where you could buy Petroleum Jelly in nice blue flexible plastic tubes where you can squeeze it out like a professional.

It looks like when I need to buy more Petroleum Jelly in a tube, I'm have to get some professional variant at Amazon.  And I worry that the tubes may not be as nice as the Walgreens ones were.

100% pure Petroleum Jelly is generally far superior to all the other sorts of products which add extra ingredients, such as Aquaphor.  Now, Aquaphor IS conveniently sold in tubes.  But those extra ingredients are almost never necessary or useful but just create the conditions for the product to degrade over time.

Paradoxical and other Information Processes

[This is still a work in progress.  However, illness and long runs are slowing progress.  I will link back or copy to a future post so you don't have to keep looking back here.]

Introduction and definitions

In the discussion and experiment that follows, I propose to define an Information Process as something that takes an input stream of information, if any, and produces an output stream of information, if any.

I imagine 5 basic types of Information Process:

Lossless, where more or better input, no matter how small, is reflected in more or better output.

Truncated, where the tiniest changes in input may make no difference in output.

Paradoxical, where less information can sometimes produce more or better output (such as tiny amounts of information being worse than none). 

Independent, Where input if any is completely ignored in producing output.

Dumb, Where no output is produced.

Paradoxical sounds strange but may be the most common one.  It seems to be a description of human beings, and, I would have thought, most computer programs of any significant complexity.

I was shocked to find out that a fairly complicated computer program I have been developing over the past few years is actually Lossless.  No matter how little the input information it is given, even amidst increasing misinformation, it still performs better than with no information.

The only ways I could think to make it Paradoxical require that there be be information loss in the program itself.  I can't say I've proven this but it looks to be correct for all such processes.  Information lost within the process interacts with the quantity or quality of input information to produce the Paradoxical effect.

But if information loss causes Paradoxical behavior, what causes Truncating behavior?   Well, the only answer for my program appears to be the injection of random numbers.  Something like this appears to be generally correct, and I note that digital audio processors add dither when lowering resolution.  I cannot prove that it always requires noise or random numbers, but it is an interesting conjecture that some sort of active operation has to take place rather than just being a passive loss where information just isn't copied, for example, because passive losses produce Paradoxical and and not Truncating.

Methods

The program MakePlaylist program randomly selects songs or albums for playlists to play as background music.  I've worked on things like this for about 30 years ago and realized that you must use Random Without Replacement (RWOR) operation instead of Random With Replacement (RWR) to prevent a great and largely undesired disparity in how much things are played.

But I have been concerned with two different but similar problems.  In the first situation, I was continuously playing songs in a background process until I feel like listening to music, then I turn the amplifier on and hear them.  But I might only listed to the background process a few hours a day, and the rest of the time the songs are being marked as 'played' by the RWOR system even though they haven't actually been listened to.  That was the kind of thing I was doing in the late 1990's, when I scarcely imagined such things as playlists, I scarcely imagined them until 2021 when I got Roon and an Oppo player I could use it with.

In the present situation, with MakePlaylist, I still might not actually listen to all the songs in a particular playlist before replacing it with a new one.  Around 2023 this always happened with my playback program (Roon) after I shut off the streamer remotely.  Roon would hang and when restarted it would start the playlist from the beginning.  So that was a big problem then, but Roon has fixed those problems now, but it can still happen anyway that I don't listen to every song in every playlist because I decide to listen to a bunch of other individually selected items in the interregnum, then my place in the previous playlist may still be lost because Roon doesn't keep your place in a playlist if you start listening to other things. (Inconveniently, Roon never shows the number of the item in a playlist that you are playing either.)  I can imagine it's pretty commonly done that way too, Roon is a High End player program that requires a subscription or large initial payment and generally has all the features available in other programs.

And then the question becomes, in cases like these, where the information about the songs played is combined with and perhaps even dwarfed by misinformation about songs being played that weren't actually played, is it possible that RWR performs better than RWOR ?  And I can define this performance in a singular and objective way: how many playlists do I need to create to listen to every song in a library?

(Though one could also look at the distributions of times things were played and details like that, it's much easier for the present experimental purposes to boil the question down to one easily described and measured number, which obviously is related to the distribution too but it doesn't define the spread or kurtosis of the distribution which can vary among methods, but in all cases I have seen, RWR has higher kurtosis than RWOR meaning longer 'tails'.)

With a properly working RWOR system, and when you listen to every item in every playlist, it will take ceil (L / S) playlists to listen to everything where L is the size of the library and S is the size of each playlist and ceil is the ceiling operator which rounds up fractional bits.  I call this number of playlists, or the number of songs listened to (depending on context) in such a count of playlists, one epoch.

I realize that the problems' I am investigating could be solved analytically with probability formulas, but I chose to solve them experimentally by building test shells around Make Playlist: first TestPlaylist and now TestPlaylistRange, because it was a lot of fun.  Ultimately my results can be compared with the true means from analytic probability theory, though it is more difficult to get the distributions and other ancilliary information like that.   Then my statistics such as confidence intervals can be assessed.

Tests show that MakePlaylist works correctly in both RWOR and RWR modes.

For convenience here, I have stuck with a library size of 1000 and playlists sized 100 items each, though that is longer than most people make playlists (20-30 is more typical) the even numbers are easier to think about for experimental purposes).

I run experiments in which the last item actually listened to in the playlist (<LP>) can be set to any number between 1 and the playlist size, and playlists are created over and over until every item has been played.  This may require making many thousands of playlists.

Since this is all based on random numbers (via the pseudorandom number generator built into Tcl, which is fairly good but simple) the number of playlists can vary a lot from one experiment to the next.

So, each such experiment is repeated from scratch all over again for many repetitions and averaged to get stable and presumably accurate results.  I have found that I can get reasonable stable results in most cases with 1000 repetitions, but there is still some "noise" in the graphs, so for the really important tests I do 10,000 repetitions.  I compute statistics such as 95% confidence interval, 82% confidence interval (useful for comparing two numbers) and residual kurtosis.

MakePlaylist and all the companion programs including TestPlaylist and TestPlaylistRange are available as Free Software under a GPL 3+ at SourceForge.  What is described here requires Version 1.1 or greater.

Algorithms

These are simplified from what MakePlaylist actually does, mainly because of the options.

RWR

    Make list of all titles

    Until desired playlist length reached:

        Draw a random number

        Scale number to match size of list

        Index into list and copy title

        Put title into playlist

RWOR

    Make list of all titles

    Until desired playlist length reached

        Shuffle list randomly

        While there are still more items in list and desired playlist length not reached

            Copy first or next title from list

            If it has been played in this epoch

                Continue

            Put title into playlist

            Add title to played history

            Continue

                                    

RESULTS

RWR vs RWOR, 100-1 LP, 1000 reps

I was very pleased that my RWOR selection system correctly selects all the items in the library in precisely 1 epoch every time, 10 playlists of 100 titles each, so long as all items are played.

Playing only 99 files out of a hundred makes RWOR substantially worse, requiring 2 epochs to play all the files.  Still many times better than RWR.  Each additional unplayed file has a lesser effect than the previous one.   Even just playing 1 file out of 100 in the playlist, with a library size of 1000, it was still better, ever so slightly, than RWR.

This is the Lossless behavior.

For RWR, under the test conditions of L=1000 and S=100 it takes an average of an astonishing 7.506 epochs to select all the titles (95% confidence interval is 0.003, but the high kurtosis also means that 0.003 is probably an underestimate of the confidence interval so I can't be sure it's even above 7.5).  Some titles are played over a dozen times before all files are played once.  I knew RWR was bad, but I didn't realize it was this bad.

Redefining the epoch as the number of files played in an epoch of playlists. the RWR stays virtually constant as fewer items are played in each playlist.  It should be this way as each RWR draw is precisely the same, completely independent of all played histories.  In this regards, RWR is an Independent process, always the same regardless of input.

I felt like I had discovered the magic point where the misinformation outweighed the information and felt jubilant that I had discovered something important.  But I knew I needed way more test runs, and a test infrastructure that computed important statistics so I could determine whether the difference from RWR was significant or not.

At that time, I was making actual playlists from folders of files, storing results to the played history, and running the -stats option of MakePlaylist (which also loaded the history and folders of files) to see how complete it was.  I was not able to run more than a small number of the toughest cases.

To make the tests run faster, I made an option to eliminate all playlist and database output.  Everything is done inside the TestPlaylist shell, which textually includes the current MakePlaylist as a procedure.  And instead of selecting actual files in folders, it simply selects from integers numbered 1...<library size>.  These refinements make it possible to create the needed millions of playlists even using a pre-compiled language (Tcl) without any parallelism.  I also added statistics to the output.  This was not all done at once but in a series of separate refinements.

At first, the result with 10000,100,1 being the magic point held up when I did 10 test runs of each.

With more improvements, I was able to increase the number of test runs to 100 the result still held, but with hardly any statistical significance.  But with even more improvements, I was able to increase the number of test runs to 1000, and then the result failed,  RWOR was still better than RWR even when playing only one title out of 100 from a library of 10000.  The 'magic point' disappeared.  The statistical significance of this difference is quite a bit shy of 95% confidence.

RWOR playlists required for 1/100/10000: 96820

RWR playlists required for 1/100/10000: 97774

Actual difference: 954

Difference required for 95% confidence of difference: 1106

The difference required for 95% confidence the means are being sampled from different distributions is the sum of the two 83% confidence intervals.  The calculation is now done by TestPlaylist but these calculations were done by hand.

I was disappointed that 95% confidence was not achieved but rather unwilling to repeat these tests with, say, 10 times more repetitions than the already 1000 repetitions because these runs each took a week and a half already.

 But if we ask the question of whether the true values favor RWR over RWOR, my original conjecture, that is probably greater than 95% at this point, because there are 3 possibilities:

1. RWOR better than RWR

2. They are the same.

3. RWR better than RWOR

#3 is the farthest from the existing mean values from large number of tests.  Given that, #2 is more likely than #3.  #3 is the least likely of all, so it is possible that RWR is worse not better or same is already at a 95% confidence interval.  I don't know how to do those calculations, but I think my original idea that RWR is better specifically at 1/100/10000 is likely disconfirmed near 95% confidence, possibly better.

And things aren't hetter (and likely worse) with a smaller library size of 1000 either:

RWOR playlists required for 1/100/1000: 7307

RWR playlists required for 1/100/1000: 7507

Actual difference: 200

Difference required for 95% confidence of difference: 342

This suggests that the library size either isn't a factor, or goes in the reverse direction from what I conjectured, larger libraries show better results for RWOR down to one 1 unit of valid data out of 100.  So it looks like I don't have to keep trying larger and larger libraries to find the pivot point.  And that's a very good thing because each tenfold increase in the playlist size causes a tenfold increase.  And they were already taking over a week.

BTW, when RWR at 1/100/10000 says 77774 playlists required, that's for each run.  Do that 1000 times and it means 77,774,000 virtual playlists need to be created.  That's why it takes so long.  I think my Tcl code is actually very fast.

So, even with only 1 file played out of 100, RWOR still achieves better results than RWR at every library size I can try.  And there's no indication that even larger libraries would break this.

That is why I am calling my RWOR algorithm Lossless.  The performance with less and less information is monotonically asymptotic to No Information (RWR) so it never gets worse than No Information.  You could in principle work backwards from the number of playlists required to play all the files to the number of items that are being skipped.  You'd just have to create millions and millions of playlists to be sure. 

At this point, the only way I could figure out to makeMakePlaylist Paradoxical was to make it lose some of the information it is supposed to save.  I decided to add a PARADOX <LP> option to MakePlaylist which causes it to fail to record the first LP items in the played history.  This then produces the Paradoxical result that fewer than LP items are played, there is only misinformation and no correct information at all.

The only way I can see to achieve a Truncating result is to replace the first items with RWR truly independent items.   I plan to add a TRUNCATE <TP> option in which the first TP items are randomly chosen with RWR.  By default, history is still recorded for them (ie, there is no information lost) but that's irrelevant for RWR, it is independent of any history.  One could also combine the PARADOX and TRUNCATE options.  I suspect it would still produce truncating results but with lower epoch counts than just TRUNCATE alone.  The virtual epochs the RWOR system 'sees' are shorter, making things repeated more often, but it barely matters to the truncating result which is identical with RWR at TP and below.

Paradox 4, 15-1 LP, 10000 reps vs RWR avg 100000

Calling anti-Zionism fake

 One actually existing form of uncalled for antisemitism is claiming those Jews who call themselves (and meet the standard I am about to describe) anti-Zionist are fake.  Many do this, including Gilad Atzmon, who calls Tony Greenstein a fake anti-Zionist simply because he is Jewish and won't call himself an 'ex-Jew' like Atzmon has.

This canard is suggesting anti-Zionist Jews are all liars, they know they can go to Israel whenever they want (actually they can't if they have a social media history of anti-Zionism), so they're all really closet Zionists.  That is anti-semitism.

I have absolutely no doubt that Tony is sincere, and I feel similar about many organization including Jewish Voice for Peace (JVP).  JVP had not started as an anti-Zionist organization, but when they declared themselves one in 2019 it cost them membership and support.  They were very sincere.

But there is a standard, a standard that people like Noam Chomsky and Norman Finkelstein don't meet (they've never actually said they were anti-Zionist either).  And that standard is this.  If you believe there needs to be a 'Jewish state,' or that it is useless to think past the existence of one, you are not anti-Zionist.

Anti-Zionists do not believe in the Two State Solution, that is a concept from Liberal Zionism.  Chomsky and Finkelstein are best classified as Liberal Zionists.

Zionism was not something all Jews believed in the beginning. From 1890 until around 1960 most Jews in the world were either non-Zionist or anti-Zionist.  The exceptions were more heavily concentrated among the wealthier Jews.  Then because of the Holocaust and a PR blitz which included Fiddler on the Roof, Zionism won the hearts and minds of most Jews.

Jews are not all bankers, media moguls, and top government officials.  Many are workers and Jews were leaders in labor organizing, including the Jewish Bund (which opposed Zionism).  More Jews have been workers than super wealthy people, and they were concerned about the masses including but not exclusively themselves.  Jews have been historically identified with the left in politics (once again, except those at the top).

But inexorably Zionism has dragged many Jews--who are the wealthiest ethnicity overall in the USA--to the right.  That is what happens when you become part of an imperial project.  Richer and righter.

But still, there are once again increasingly more legitimate Jewish anti-Zionists and it is antisemitic to say they are fake.  There are still many Jewish socialists, communists, anarchists, and anti-Zionists.

Neurosymbolic Hybrids

 Gary Marcus argues for neurosymbolic hybrid AI approaches that combine LLM's for pattern recognition with symbolic reasoning for planning and generalization.  I've been on this page (as were most AI researchers then) since the 1980's.

https://garymarcus.substack.com/p/even-more-good-news-for-the-future

But why should I want to aid the supercop predictive thought policing future?

Erase what I just said.  Keep on doing it wrong.

Marcus presents papers showing that traditional approaches work better for very targeted problems, including protein folding and databases.

People forget about databases and think LLM must be the best for all sorts of problems because it can "think" across all boundaries.

Paradoxical Processes and Quantum Physics

 My experiments with random selection without replacement as an information processes reveal that what I call Paradoxical behavior is achieved with internal information loss, because internal and external losses may compound.  Seemingly simpler Truncating behavior seems to require injection of random numbers (as with dither in digital audio processing).

Starting with the hypothesis that Paradoxical processes are behind the strangeness of quantum physics, this suggests that contrary to the laws of quantum physics, in some cases there IS data loss, and that drives the strangeness.

And I see how this could happen.  What we know of as bosons (protons and neutrons) and leptons (electrons) could be complex structures built around and sustaining tiny black holes.  These black holes provide the "gravity" that matter has.  Meanwhile, the structure around the black holes keeps the black holes alive by capturing the Hawking Radiation and forcing it back into the holes.  That overall process is information losing.  Nothing but black holes can have an information losing character (like my Dumb process, information goes in but doesn't come out).  The information lost is the precise time, giving it a granularity, so they don't exist in continous time but leaps ahead, meaning some information states don't exist.  There is less information in the world than we think.  The effects of this are profoundly paradoxical.