Disappointing Zen Stories
2008-07-04 21:30:52 UTC
___ Blocked, In 2006 ____________________
For more than a year from late 2004 to early 2006, there were
successive efforts to block my postings (posted by hand, not by bot)
on Google Groups, previously known as Usenet. This blocking was
specifically to the alt.* groups which are unmoderated and supposedly,
according to the beginnings of Usenet, were originally meant to be the
Wild West of network computing.
For those who do not know, Usenet preceded the invention of the
Internet by many years, for some groups back to the ARPANet of the
early 1980s. So it would be preposterous for Google to attempt to own
Usenet or attempt to regulate the alt.* groups.
My postings to the alt.* groups related to Buddhism are a sore point,
especially to those inclined towards the distortions of Buddhism
collectively known as "Zen" or "Nichiren Shoshu".
After many attempts and around about June of 2006, finally, I was
blocked completely. My usual tricks to get around the blockage by
dynamism (trying to get past the blocking code by various means, like
multiple signons and IP addresses) no longer worked.
I was stumped and chanted a lot about it. I chant Nam-Myoho-Renge-Kyo
which is the practice of the Lotus Sutra, which sounds like ...
Then I figured it out, it was so simple ... the size limits on posting
had been arbitrarily dropped from 100K bytes down to about 20K bytes,
which selected against most of my posts.
I post big files, so that I don't have to break them up into a lot of
little pieces in 5 or 10 part postings that would multiply my posting
footprint by a large factor. I don't post that often, so this would
mean a far more intrusive presence (and much more busy work for me).
Faced with the scale of the editing task, I went to post a complaint
about the blockage in Google Groups around November of 2006, and lo
and behold, the user interface had changed, and the 20K size limit was
back to 100K.
So I continued my previously-disrupted posting, and I posted a retro-
complaint entitled: "Previously Silenced By Google Groups ...", which
is located at... (http://groups.google.com/group/alt.zen/msg/
____ Blocked Again, In 2008 _____________
And on I went for another year and a half until last month (May 2008)
when I noticed the user interface had changed yet again. Now, there
was a 'Captcha', which is a little entry-box for an image of a barely
readable text image, which defeats the 'bots', but not the humanoids
However, many of my larger posts were blocked again. This was
different, though. Bigger posts were blocked, but size was not the
filter, content was the filter. Like in the case of Google China.
To analyze the content filtering, I took my blocked post entitled
"Makiguchi's Pacifism Confirmed, and Putative Scholar Brian Victoria's
Inaccurate Cherry-Picking Research Exposed, for Jon Petry -". This
post (pre-blocking) was originally located at:
That post describes a concerted and intense cowardly Zen attack
(supported by Nichiren Shoshu and other temples) on the deceased
founder of the SGI, Tsunesaburo Makiguchi. (If I describe it too much
here, this post will be blocked, too. So, go read it there.)
After removing and re-posting successively cut-down versions of this
post, I finally came down to a 1.6 KB version (which could have been
reduced further, but diminishing returns had been reached). Among
other text items in the remainder, the names of the three founding
High Priests of the Fuji school, the supreme object of worship and the
first two Presidents of the SGI were present. I am sure this is not a
complete list, since many other of my posts are also blocked.
Ultimately, this means that no one can talk about these topics in
Google Groups, otherwise, I would be able to get around these filters,
Content-filtering is done by having a list of scored text items, which
if present raise the total score for a posting or writing or web page.
When a certain total score is reached, that scored post becomes
This is a pain to do, unless you have intelligent tools: First you put
a list of all the posts, writings and web pages to be blocked on the
one hand (my posts, in this case), and the preferred posts, writings
and web pages to be allowed to be expressed freely, on the other hand
(all the rest of the posts of the Zen believers in alt.zen, in this
case). Then you run these two sets against topological text sorting
tools which arrive at two sets of possible expressions, and histograms
of those two sets. After a complicated kind of subtraction of the
combined preferred histogram from the blocked histogram, what you have
as the remainder is the filter (roughly). You test the filter by
running it against both sets and getting a zero result from the
preferred set and the largest possible filtration result against the
As an example, the famous Chinese human rights activist Huang Qi was
arrested on June 10th (last week) for possessing state secrets. He was
researching (for his website) the corrupt practices of the contractors
who built the hundreds of schools that collapsed during the school
session on the day of the Sichuan earthquake, and the malfeasance of
the government inspectors who were responsible for guaranteeing the
safety of those thousands of dead children, and the potential
government official cover up which is now attempting to point the
faulty finger of justice against a few selected unlucky scapegoats who
will receive the bullet in the head meant for the true perpetrators of
this holocaust against Chinese children, who did not have enough
political influence with the State to survive the earthquake. So, when
it came time to double the size of schools, they just simply added a
third and fourth story on schools engineered for two stories.
In China, state secrets are whatever the state prosecutors decide they
are ... in this instance - any information relating to this case.
Now, I guarantee you, that the NPR audio report on Huang Qi, any
textual web reports and this posting you are reading will never see
the light of day as a result of a Chinese Google search. Also, by no
accident, those URLs will not be accessible by Chinese Internet.
This, in a nutshell, is the power of Google to stifle free speech
across a society. And it's happening now in the West, as well.
____ Google Aligns With _________________
____ The Forces Of Evil _________________
At the very least, one individual Google employee has been using his
Google benefit of working on things not related to his job function,
by trying to block my posts, for some time, now ('the unseen actor').
At the worst, Google is consciously aligning with the forces of evil
against the Lotus Sutra and this practitioner of the Lotus Sutra
(constituting 'a broad conspiracy' against the Lotus Sutra). By
posting this protest publicly to alt.zen, I am effectively making all
cases equivalent to the latter case. Here's how that's done.
Upon reading this post, alt.zen will not be able to resist trumpeting
this 'triumph' of Zen over my efforts to refute their distortions, and
they will gloat over the public support of Google in this religious
discussion, by Google's suppression of my tiny half of it. That will
reach publicly enough into Google to make them complicit overall with
the 'unseen actor' doing the filtration, creating a 'broad
Google, having taken over Usenet as 'Google Groups', is now an
authority over what has been a public forum of expression since the
early 1980s. At its height, Usenet had many thousands of servers in
government facilities, businesses, schools and libraries all around
the world. Every posting to well-supported groups thus had many
thousands of copies distributed around the world. Usenet had a huge
data replication footprint. Upon the formation of Google Groups,
servers started to disappear as Google BECAME Usenet.
When Google Groups aligns with Zen against the Lotus Sutra and its
practitioner (me), it becomes what Nichiren Daishonin calls the "Third
Powerful Enemy", when a secular authority aligns with evil priests to
persecute the practitioner of the Lotus Sutra. (http://
The mere fact that I may appear to be unworthy is irrelevant. The
Lotus Sutra elevates and transports its practitioners, like a fly upon
the tail of Pegasus. (Even seemingly unworthy, but totally sincere
believers, such as myself.) That is the nature of the Supreme teaching
of the Buddha, the Lotus Sutra.
____ Cause And Effect ___________________
The title of the Lotus Sutra, which is also its practice, is "Nam-
Nam = Devotion of ones life
Myoho = Wonderful Law or Mystic Law
Renge = Lotus or Entity of Cause and Effect
Kyo = Sutra or Teaching or Sound
The Law is causal, in that it contains cause and effect. Offend the
Law at the heart of the Lotus Sutra and you get effects which are
completely different from those of secular crimes.
Secular crimes affect this one provisional life and their effects will
mostly only plague you for this one provisional life, or into the next
one. Offenses against the Law manifest the truth of your life, of
uncountable existences from the beginningless past into the remotest
Slander of the Law, such as Zen practice, or aligning with Zen in
attacking the Lotus Sutra and its practitioners, effectively rings the
eternity of your life, like a bell.
Nichiren Daishonin stated that there were two kinds of people who vex
the practitioners of the Lotus Sutra:
From the "Letter to Horen" Writings of Nichiren Daishonin, p. 522 ...
. There are two types of people who show animosity toward
. the Lotus Sutra. The first are people who cultivated the
. roots of goodness in former existences, who in their
. present existence are searching for some connection with
. Buddhism, who conceive a desire for enlightenment and
. are capable of attaining Buddhahood. It is these people
. whose mouths are stopped or whose heads split apart.
. The other type are people who have slandered the correct
. teaching in their previous existences, slander it in
. their present existence, and for existence after
. existence go on creating karma that will condemn them to
. the hell of incessant suffering. These people, even
. though they may curse, will not have their mouths
. stopped. They are like men who have already been
. sentenced to execution and are awaiting their turn in
. prison. While they are in prison, regardless of what
. evil acts they may commit, they will receive no further
. punishment other than the death sentence already passed
. upon them. However, with regard to people who are
. eventually to be released, if they commit evil acts in
. prison, then they will receive warnings.
So, the first kind who actively seek enlightenment make a reverse
relationship with the Lotus Sutra, and receive the punishment of
having their heads split into seven pieces, and they eventually
recover and attain the way quickly.
The second kind appear to commit one mayhem after another against the
Lotus Sutra and its practitioners without receiving any effect
(creating widespread doubt in the veracity of the teachings and the
worthiness of its practitioners), and are only finally brought down in
the end. They are called 'icchantika' (Hindu) or 'issendai' (Japanese)
which both mean 'incorrigible disbelievers'. Like the original
Devadatta, Shakyamuni Buddha's cousin, who attempted to murder the
Buddha and take his place and distort his teachings, Zen and its
accomplices will eventually suffer a similar fate.
Which one is Google? Only time will tell.
The first time I was blocked in mid-2006, Google acquired Youtube for
more than a billion, whose IP had no technology that Google couldn't
do better, and only fleeting brand value (as Ebay, Myspace and
Facebook have discovered, once you peak it's hard to keep it up for
long. People are just not loyal, when it comes to their whimsy.)
Rapidly evolved buyer's remorse will never be revealed by the board,
but its presence is obvious by their subsequent caution. "Married in
haste, we may repent at leisure" - William Congreve (1693).
And now Mark Cuban is saying "Hulu is kicking Youtube's Ass." This is
by using Youtube's losing business model against them, calling it
"Youtube Arbitrage". Here's his blog posting ...
However all that proceeds from the cause made in mid-2006, the evil
subsided quickly in November 2006, when Google Groups made a new user
interface with the old limits of 100KB for posts, and I went merrily
on posting as before. So, the corporate motto of "Do no evil.",
became, effectively, "Do no evil, and if you do, cease rapidly."
As those in fault-tolerance know, an average MTBF can be made into an
excellent availability by a very short MTR.
Why do nasty people like Mark Cuban and Bill Gates keep kicking the
ass off of Silicon Valley High Tech firms? I'll tell you a history and
give you the current status of the undermining of Silicon Valley by
____ High-Tech Zen Stories ______________
1. Oracle's Zen founder Larry Ellison had a house built in Woodside by
a Zen priest Paul Discoe to look like a Zendo. The original house was
a historic treasure designed by architect Julia Morgan in 1913. She
also designed Hearst Castle, but Larry 'deconstructed' that beauty and
consigned its works to the dustbin, choosing instead cheesy
modernized-'authentic' Japanese styling a la Epcot (a wooden flute
plays in the background).
It cost him $60M+ and was scheduled to be complete in 2001 ... and may
not have actually been completed when he put it on the market for sale
in 2005, asking for $25M. Nice discount, has it sold? Who would buy
this cursed, evil castle?
His office in the emerald city towers looks like the martial-arts-
bakufu fantasies of a self-styled Shogun. Next thing you know, he'll
be taking hostages like Nobunaga and Hideyoshi did ... well, actually
his acquisitions are kind of like that. Maybe now there will be Sankin
Kotai-styled processions to his shrine in Woodside, like the Ieyasu-
cult worshiping at the Nikko shrine (just North of Edo) during
Tokugawa Iemitsu's reign.
Oracle continually loses the juicy, high-visibility, critical
computing accounts, because of this Zen connection. It is an all-
marketing outfit with perpetually flawed technology (for some
reason ...). The licensing contract with the users absolutely
prohibits any performance testing, to prevent any comparisons to
better performance, and to prevent the dissemination of info on how
Oracle performs if you don't do benchmarketing cheats. Because of
this, when users do compare serializable and reliable operations
platform to platform, Oracle doesn't look like such a prize. Some
stories on this ...
_ _ Oracle Zen Stories _ _ _ _ _ _ _ _ _
- - - - - - - - - - - - - - - - - - - - -
[a.] When Target's (Compaq Nonstop) and Mervyn's (Oracle) CRM
databases were compared by Dayton-Hudson (who owned both companies),
they found that the Oracle systems were far more expensive to operate
because an army of operators had to continually build indexes to
support sales queries, because the Oracle query engine did not perform
well against the base tables. Not only that, but it took an hour to
build these indexes, so that queries were always against a fuzzy one-
hour-plus old dataset. Like any mainframe, the Nonstop SQL queries
went against the base tables and the results were never older than
when the query was started. The rumor was that Mervyn's CRM database
was ultimately recoded away from Oracle.
[b.] In 2001, EBay's IT leadership begged a prominent transaction
performance conference for help, because their Oracle database
required an hour to rebuild indexes, and so users could only find the
auctions they were seeking based on data that was between one hour and
two hours old. This had a deleterious effect on finding those auctions
based on last minute activity. The rumor was that IBM's mainframe DB2
guys came in and saved them from this cognitive dissonance.
[c.] When Citibank wanted to build a Worldwide Funds Transfer system,
a competition was held between three competitors: IBM Z/OS Sysplex
Mainframe DB2, HP 9000 Superdomes using Oracle, and Compaq Nonstop.
Within minutes of the beginning of the competition, the HP/Oracle team
was called in, and after a brief Q&A they were summarily dismissed
from the competition. No one outside of that office knows precisely
what was asked and answered. Was it something about having to bring
Oracle down to take database and catalog dumps before and after
changing the schema DDL (or doing just about anything?). Was it
something about the fact that after a crash, database integrity was
always lost, because they didn't do things correctly while they were
still in the steady state of 'up for business'? Was it that they
occasionally threw a redo record into the log to allow them to
summarily drop a pesky lock before commit, as a performance cheat?
Only Oracle knows - which piece of Zen was at fault.
[d.] While banks, exchanges and critical computing have been largely
smart enough to learn from their mistakes in selecting Oracle, run-of-
the-mill Fortune-500s and the government are not so smart. So you get
the churning of failed systems development with Oracle at the center
of it, irregardless of all the other variables: agency, contractor or
type of system. The pattern is that if Oracle is involved, tax dollars
go up in smoke with drastically reduced expectations in the end. The
FBI and the Virtual Case File system (VCF), the IRS and Tax Systems
Modernization (TSM) and many others have crashed and burned using
Oracle systems to the tune of billions wasted.
"The FBI spent $170 million on a 'Virtual Case File' system that does
not operate as required. After three years of work under a cost-plus
contract failed to produce a functional system, the FBI scrapped the
program and began work on the new 'Sentinel' Case File System...."
"In the mid-'90s, you couldn't swing an embattled CFO without hitting
a public company that blamed its disappointing financials on an ERP
implementation gone bad, but vendors SAP (NYSE: SAP) and Oracle (NSDQ:
ORCL) were often made the scapegoats, not the companies' own IT
And that is true, the IT organizations WERE at fault in their
selection of Oracle Zen.
[e.] There ARE exceptions to this pattern: some critical computing
customers ARE ALSO not so smart. Bear Stearns WAS a top corporate
customer of Oracle's, a real shining star. Now Bear Stearns has become
a near miss with utter catastrophe for the economy, a source of
massive litigation and prosecution, and the poster child for the
subprime failure converting hidden chains of Collateralized Debt
Obligations (CDOs) into fully revealed chains of margin calls.
[f.] Financials like Bear Stearns, that lost their shirts, but
survived, have switched from betting the business on subprimes to
gambling on oil futures. This will finish off those rank and file
American families whose homes are now worth less or are in the red
against their sagging bank loan. Now these beleaguered people can no
longer afford to eat or go and take the kids to the doctor, because
the speculation on oil has meant that the entire family budget now
goes into the car's gas tank or the house's heating oil tank.
Amazingly, one of the two exchanges which determines oil futures is
the ICE Futures Europe (London), which is the only major exchange in
the world (that I know of) based on ... you guessed it ... Oracle. I
sense a coming peripeteia (reversal of fortune) for the oil gamblers
who believe they cannot possibly lose using Oracle technology.
"... leaders of the world's two leading oil exchanges, the New York
Mercantile Exchange and the ICE Futures Europe, based in New York and
[g.] Waste Management had Oracle build their legacy systems, then had
enough of Oracle and switched their apps over to SAP's TomorrowNow
subsidiary. TomorrowNow, it turned out, had used stolen logons and
passwords from BEAR STEARNS to extract information from Oracle
accounts to steal a large portion of the Oracle customer base. Oracle
sued, the truth was revealed and SAP distanced themselves from
TomorrowNow at warp speed. The moral of the story is: you don't want
to be an Oracle customer, you don't want to maintain Oracle legacy,
and you don't even want to steal Oracle customers and maintain Oracle
legacy. Toxic Zen shock !!!
"SAP said more than 200 companies were taking advantage of its Safe
Passage program to move off apps acquired by Oracle." May 22, 2006
"... Oracle's allegation that SAP employees pretended to be Oracle
customers to log on to its Web site and copy proprietary technical and
customer-support data so it could provide lower-cost services to those
customers, with the eventual goal of luring them over to competing SAP
products." Nov 20, 2007
"While Waste Management didn't provide details of its legacy systems,
it appears some were developed by Oracle (NSDQ: ORCL)-owned companies.
Waste Management used the services of SAP's TomorrowNow, a company
that provides third-party support for Oracle software such as Siebel
and JD Edwards, and is the subject of a lawsuit Oracle filed against
SAP last year." March 27, 2008
[h.] Some years ago, a company called Innobase, with a product called
Innodb, did some very cute performance work based on MySQL. By
creating a new version of the storage/query engine for MySQL, which
did vertical partitioning, this meant that the index and database
blocks only contained info for one field (as opposed to the entire
record) against the primary record key, seriously reducing the numbers
of blocks fetched to do a partial scan for selects or nested loop
joins. Typically, database query plans are composed of sequential
table and index scans, selects, projects, sorts and joins and hash
equijoins on indexed keys (there are many variations on these) ... but
the partial table and index scans tend to precede the others in the
order, so this means something fundamental to the start-up time before
parallelism kicks in. You pay a price getting the pipelines
flowing !!! Cutting that fixed start up price substantially makes
shorter work of the whole business. For small queries this becomes an
even bigger deal. At the ends of query plans, the dataset sizes are
smaller, so they are equivalent to small queries and benefit from this
As soon as Innodb started to look to have potential, Oracle snapped
them up in 2005.
Amazingly, they recently compared their database against Oracle (not
listed as 'other'), PostgreSQL, and Microsoft (listed as 'other',
because they also do not allow users to print benchmark results).
(some of this is at http://www.innodb.com/innodb/benchmarks/)
Basically, Innodb killed the competition!!! But wait, it seems that
they ran the other databases without consulting the other vendors
(listing Microsoft as 'other' implies that). Oh, and the Innodb index
and data blocks were compressed (according to a presentation by J.D.
Duncan). So, you ask what about updates? It turns out the engine
somehow doesn't support them with compression, but they say that's
coming in a later release. (Hmmm.)
At that point, presumably, the update operation will decompress the
block, then update a row and recompress the block? Or maybe the
compression is only for part of the block. Since you need to read the
first part to find the things you need for a scan, or sometimes you
need the last part (offsets or pointers) to do a binary search, what
part would you compress? When you are doing a scan, and you fetch the
leading part of the block into the initial 2nd level cache line, then
you want to pull out the fields in it while cache prefetch is putting
the rest of the block into the following 2nd level cache lines, you DO
NOT want to stall for the VERY LAST 2nd level cache line in the block
to get filled so that you can decompress the block to look at that
initial 2nd level cache line you got ages ago. Finally, fast databases
use bigger block sizes (because you pay the I/O cost for the first 500
bytes and the rest is pretty much free, DMA-wise and 2nd level cache
prefetch wise), thus increasing the overhead and stall-time for
decompressing/recompressing the deluxe-sized index and data blocks.
Usually, naive users that deploy databases that support compression
for index and data blocks will turn on compression for both, then
immediately turn off index block compression due to the performance
hit, and somewhat later on, turn off data block compression. That is,
if they actually USE the database for something and want to USE the
indexes to find what they just did, which implies a moderate to heavy
update load during peak business hours when response time actually
matters. So block compression is anathema for the real-time or even
just online databases that people actually use and want to actually
see the results of their work, now.
So Innodb's block compression performance numbers are a crock.
But what about the improvements in the insert rate in that one
database test against PostgreSQL? Note that when compared to MyIsam of
MySQL, inserts on primary key were only a half a percent better. So,
no magic is involved. Basic MySQL beats PostgreSQL in some tests, too.
It's all in what tests you run, and how you set up the competition as
a straw man to beat the crap out of.
In a completely proctored TPC-C run (proctored by someone that
everyone could trust such as Chris Date, who used to proctor some of
these in the early days), there is a step where the fully running test
configuration, without ANY configuration or switches being changed, is
crashed and then brought back up to see if the database integrity is
correct to the last committed transaction before the crash. Doing as
the Innodb article suggests, and running some 'TPC-A-like' tests, non-
proctored, without the TPC-C crash test, means that any kind of funny
business can be happening.
In the old days, there were a multitude of rumors about this kind of
funny business from Innobase's new daddy, Oracle. Oracle has redo logs
which add redo records for every transaction, only to be used in
recovery after commit, and undo logs which add undo records only to be
used for uncommitted transaction recovery. Never the twain shall
meet ... mostly.
In fact, if there were hotspots in the database during a benchmark,
Oracle (as rumor has it) would summarily drop the lock most
responsible for critical transaction rate slowdown (like the bank
branch balance in a TPC-C test) and add a redo record (something like
"no integrity for this record"), and sometime way later get around to
writing the transaction commit record, after which the pre-released
locks should have actually been released.
This violates the third letter of the ACID word, which is 'I' for
Isolation. It also violates the 'Log Serializability' part of
'C' (Consistency), in that you cannot sort the log by transaction
timestamp, rerun it and end up with the same database, which is a
constraint known as 'Sortable Transaction Histories' from Jim Gray/
Andreas Reuter's transaction processing book. (My German colleagues
call this book 'de berbel'.)
Why does the log matter? Because the log IS the database, the database
disk itself is never correct, and is mostly more than 5 minutes old,
due to the '5 minute rule' from the same book. And because (1) the
serializability of the log is broken from the lock release before the
commit record is written, and (2) the database disk never being right,
then the potential recovery of the database as a whole is never right:
a crash only reveals the truth of this. In a benchmark, you can take
measures in the application to make up for the broken reliability of
the RDBMS. In real life, application writers can rarely keep it
together to make these tricks work. This is