2020-11-03
total formal verification is not possible in 2020
'20.11.3: co.quora/adds/cyb/sec/total formal verification is not possible in 2020:
current efforts at formal verification
cannot yet find all unknown vulnerabilities in software:
. as pointed out by the SeL4 team, in
“What is Proved and What is Assumed”
who used formal verification to debug a microkernel,
they are still vulnerable to inaccuracies of the hardware model;
ie, assuming the hardware acts a particular way
when under special conditions it actually does something else.
. some CPU chip instructions are reprogrammable by the manufacturer
with on-chip microcode; eg Intel Microcode
(https://en.wikipedia.org/wiki/Intel_Microcode);
this could either make the CPU more correct
or make it vulnerable to malicious modification?
. also, there was some of the SeL4 software
that was not practically verifiable yet:
"there may still be faults remaining in
specific low-level parts of the kernel
(TLB, cache handling, handwritten assembly, boot code)."
2020-07-31
the first automation was written language
2020.7.28..31: adds/cs/the first automation was written language:
here is what humans have over other animals:
. our language uses symbols that can entirely describe
our understanding of the world;
and our understanding of the world is so vast
that we can survive without adapting our genome
by engineering environmental adaptations.
. also,
language in its written form is
more important than speech (the verbal form);
because it represents a form of automation of our memory;
giving us not only communication
but a virtually infinite memory size.
. the first reliable computations involved
writing the steps of an algorithm
so that we could automate the part of our mind
that remembered each step and their ordering.
here is what humans have over other animals:
. our language uses symbols that can entirely describe
our understanding of the world;
and our understanding of the world is so vast
that we can survive without adapting our genome
by engineering environmental adaptations.
. also,
language in its written form is
more important than speech (the verbal form);
because it represents a form of automation of our memory;
giving us not only communication
but a virtually infinite memory size.
. the first reliable computations involved
writing the steps of an algorithm
so that we could automate the part of our mind
that remembered each step and their ordering.
2019-12-07
chromebook accommodating visual impairments
19.12.4: cyb/chrome/accommodating visual impairments:
. here are chromebook features for
when the eyes have trouble reading.
. select-to-speak can say selected text,
at least in the browser.
. if you can see only large text,
the docked magnifier works well;
also handy is the (shift control +)
screen resolution
(system menu/setting/device/display/
internal display/display size).
. here are chromebook features for
when the eyes have trouble reading.
. select-to-speak can say selected text,
at least in the browser.
. if you can see only large text,
the docked magnifier works well;
also handy is the (shift control +)
screen resolution
(system menu/setting/device/display/
internal display/display size).
2019-11-22
12 p.m. is noon
19.11.17: engl/12pm is noon:
. the way to see that 12 p.m. is noon
rather than midnight;
is to realize that 12 on the clock
is actually the clock's name for 0;
and, 0pm is the start of pm (afternoon),
so 12pm is noon.
. the way to see that 12 p.m. is noon
rather than midnight;
is to realize that 12 on the clock
is actually the clock's name for 0;
and, 0pm is the start of pm (afternoon),
so 12pm is noon.
2018-02-19
Python library for robotic simulation using the MuJoCo engine
17.11.6: news.cs/robotics/
Python library for robotic simulation using the MuJoCo engine:
OpenAI 2017.6:
open-sourcing a high-performance Python library
for robotic simulation using the MuJoCo engine,
developed over our past year of robotics research.
This library is one of our core tools for
deep learning robotics research,
which we’ve now released as a major version of mujoco-py,
our Python 3 bindings for MuJoCo.
. mujoco-py uses data parallelism through OpenMP
and direct-access memory management through
Cython [C translator] and NumPy [math pkg]
to make batched simulation more efficient.
Python library for robotic simulation using the MuJoCo engine:
OpenAI 2017.6:
open-sourcing a high-performance Python library
for robotic simulation using the MuJoCo engine,
developed over our past year of robotics research.
This library is one of our core tools for
deep learning robotics research,
which we’ve now released as a major version of mujoco-py,
our Python 3 bindings for MuJoCo.
. mujoco-py uses data parallelism through OpenMP
and direct-access memory management through
Cython [C translator] and NumPy [math pkg]
to make batched simulation more efficient.
2017-11-24
Intel's lack of documentation stalled MINIX
11.21: news.cyb/adds/doc/Intel's lack of documentation stalled MINIX:
. in an IEEE Computer interview,
[ieeeComputerSociety 2014]
Tanenbaum revealed he designed MINIX when
AT&T closed the source for Unix Version 7
and they could no longer use it
when teaching about the design of an OS.
. but it kept crashing until he heard a rumor
(nothing in the documentation mentioned it)
that the Intel chip when overheating
was causing an interrupt 15.
If his student Robert hadn't revealed the interrupt,
"there would have been no MINIX" he said.
. in an IEEE Computer interview,
[ieeeComputerSociety 2014]
Tanenbaum revealed he designed MINIX when
AT&T closed the source for Unix Version 7
and they could no longer use it
when teaching about the design of an OS.
. but it kept crashing until he heard a rumor
(nothing in the documentation mentioned it)
that the Intel chip when overheating
was causing an interrupt 15.
If his student Robert hadn't revealed the interrupt,
"there would have been no MINIX" he said.
2017-11-23
common core math's subtraction
10.20: web.math/common core math/has new subtraction:
businessinsider:
. the new way is realizing a subtraction problem
is asking you to measure the distance between 2 numbers;
You do that, in turn, by measuring the distance between
landmarks (easy, round numbers).
phys.org 2015:
we all do arithmetic like this in our heads all the time.
Say you are buying a scone at a bakery for breakfast
and the total price is US$2.60.
You hand the cashier a $10 bill. How much change do you get?
Now, you do not perform the standard algorithm in your head.
You first note that you'd need another
40 cents to get to the next dollar, making $3,
and then you'd need $7 to get up to $10,
so your change is $7.40.
called the counting up subtraction method.
from the smaller number
count up to the nearest 10, 100, etc;
to that add the largest number's smaller digits.
businessinsider:
. the new way is realizing a subtraction problem
is asking you to measure the distance between 2 numbers;
You do that, in turn, by measuring the distance between
landmarks (easy, round numbers).
phys.org 2015:
we all do arithmetic like this in our heads all the time.
Say you are buying a scone at a bakery for breakfast
and the total price is US$2.60.
You hand the cashier a $10 bill. How much change do you get?
Now, you do not perform the standard algorithm in your head.
You first note that you'd need another
40 cents to get to the next dollar, making $3,
and then you'd need $7 to get up to $10,
so your change is $7.40.
called the counting up subtraction method.
from the smaller number
count up to the nearest 10, 100, etc;
to that add the largest number's smaller digits.
2017-10-25
I, robot (2004)
10.24: tv.adds/robotics/I, robot (2004):
summary:
. the movie "I, robot" had me realizing
how complicated good programming is.
[full synopsis]
summary:
. the movie "I, robot" had me realizing
how complicated good programming is.
[full synopsis]
2017-10-02
2017-08-08
Dr. Daniel Julius Bernstein
7.25: web.cs/sec/Dr. Daniel Julius Bernstein:
. Dr. Daniel Julius Bernstein
is an authority on computer security;
he's a big fan of extreme sandboxing
(where the app is cooperative)
and safe programming languages.
djb's main page, his papers
(and who they are cited by).
. Dr. Daniel Julius Bernstein
is an authority on computer security;
he's a big fan of extreme sandboxing
(where the app is cooperative)
and safe programming languages.
djb's main page, his papers
(and who they are cited by).
2017-03-01
git uses SHA-1 deprecated by NIST in 2011
2.25: news.cyb/dev/sec/git uses SHA-1 deprecated by NIST in 2011:
3.1: summary:
. git allows teams to concurrently work on software;
it uses SHA-1 hashing of versions,
to tell when files of a version have been modified
to help it merge versions of the software.
. SHA-1 has been cracked so you can modify a file
and yet have it result in the same SHA-1 hash,
thereby hiding the fact that it has been modified.
. the leader of git would like to replace SHA-1
with a more secure hash using more bits,
but would like to use a truncated version of that hash
so that git would only have to store and compare
the same number of bits as SHA-1.
. git is assumed by the leader of git, Linus Torvalds,
to be less vulnerable to the SHA-1 attack
because it includes not just the hash of a file
but also its size;
he gives no proof other than appeal to intuition:
can you imagine a way to add working malware to a file
while also keeping both the hash and the size the same?
3.1: summary:
. git allows teams to concurrently work on software;
it uses SHA-1 hashing of versions,
to tell when files of a version have been modified
to help it merge versions of the software.
. SHA-1 has been cracked so you can modify a file
and yet have it result in the same SHA-1 hash,
thereby hiding the fact that it has been modified.
. the leader of git would like to replace SHA-1
with a more secure hash using more bits,
but would like to use a truncated version of that hash
so that git would only have to store and compare
the same number of bits as SHA-1.
. git is assumed by the leader of git, Linus Torvalds,
to be less vulnerable to the SHA-1 attack
because it includes not just the hash of a file
but also its size;
he gives no proof other than appeal to intuition:
can you imagine a way to add working malware to a file
while also keeping both the hash and the size the same?
2017-02-28
#webassembly is starting beta
2.28: news.cyb/dev.net/webassembly.org is starting beta:
Mozilla, Microsoft, Google, and Apple have teamed up
to make the web more efficient.
Mozilla, Microsoft, Google, and Apple have teamed up
to make the web more efficient.
2016-12-14
SciStarter.com and citizenscience.org
12.12: news.adds/democratizing science/
SciStarter.com and citizenscience.org:
Citizen Science Association and SciStarter
are about making science more accessible
and crowd-sourcing science projects.
SciStarter.com and citizenscience.org:
Citizen Science Association and SciStarter
are about making science more accessible
and crowd-sourcing science projects.
2016-09-17
Milk for openMP
9.15: news.adda/lang/co/Milk for openMP:
917: summary:
. Milk language optimizes openMP,
to avoid having to rewrite code.
. if you are starting from scratch,
better to avoid openMP.
917: summary:
. Milk language optimizes openMP,
to avoid having to rewrite code.
. if you are starting from scratch,
better to avoid openMP.
2016-03-26
reputation-based trust management
2.12: news.cyb/sec/reputation-based trust management:
2.14: summary:
. even if the NSA keeps vulnerabilities in place
so that they may continue accessing their backdoors;
they do not want you pawned by other nation states.
. they recommended "reputation management",
which I believe refers to
reputation-based trust management.
NSA`Tailored Access Operations`Rob Joyce:
. admins need to lock things down as far as possible;
whitelisting apps, locking down permissions,
patching as soon as possible,
and using reputation management.
. when up against a new piece of malware
it will be missed by Signature-based antivirus
but could still be caught by reputation.
2.14: summary:
. even if the NSA keeps vulnerabilities in place
so that they may continue accessing their backdoors;
they do not want you pawned by other nation states.
. they recommended "reputation management",
which I believe refers to
reputation-based trust management.
NSA`Tailored Access Operations`Rob Joyce:
. admins need to lock things down as far as possible;
whitelisting apps, locking down permissions,
patching as soon as possible,
and using reputation management.
. when up against a new piece of malware
it will be missed by Signature-based antivirus
but could still be caught by reputation.
2016-01-11
#robotics #AI heaven-or-hell-its-your-choice.com
1.10..11: news.cs/robotics/heaven-or-hell-its-your-choice.com:
summary:
. Alan Keeling` Heaven or Hell It's Your Choice;
is a book warning us about the coming robotics age.
. his idea of machine learning dangers is delusional
but he's onto something when he warns that
the current political systems will create killer robotics.
. robotics is a weaponizable technology
and it needs to be tightly controlled by a global government
in order to keep various military powers
from programming the robots to kill each other.
. the reason for capitalism is a sort of fascism
where good people take resources from the others
and good is defined by who's most profitable.
. let the free market decide who makes the money
and thus who can support unlimited breeding rates.
. it's civil war as each culture fights to expand;
and, free markets are the battleground;
but, robotics can be used to cheat capitalism
by sabotaging or killing competitors.
. we need to insist that all robotics are open sourced;
and prove that any robot unleashed in the real world
is following a constitution that hurts no humans.
summary:
. Alan Keeling` Heaven or Hell It's Your Choice;
is a book warning us about the coming robotics age.
. his idea of machine learning dangers is delusional
but he's onto something when he warns that
the current political systems will create killer robotics.
. robotics is a weaponizable technology
and it needs to be tightly controlled by a global government
in order to keep various military powers
from programming the robots to kill each other.
. the reason for capitalism is a sort of fascism
where good people take resources from the others
and good is defined by who's most profitable.
. let the free market decide who makes the money
and thus who can support unlimited breeding rates.
. it's civil war as each culture fights to expand;
and, free markets are the battleground;
but, robotics can be used to cheat capitalism
by sabotaging or killing competitors.
. we need to insist that all robotics are open sourced;
and prove that any robot unleashed in the real world
is following a constitution that hurts no humans.
2016-01-01
0.999... is a hyperreal not equal to one
12.8: co.quora/math/
here's why I have a problem with (0.999...) = 1:
. if the number (0.999...) = (1 - 1/infinity) = 1;
then the set [0,1) = {0, ... 1- 1/infinity} = {0, ... 1} = [0..1];
but then we have [0,1) = [0,1] so did we want to mean that?
12.17: wiki:
The equality of 0.999... and 1 is closely related to
the absence of nonzero infinitesimals in the real number system,
the most commonly used system in mathematical analysis.
Some alternative number systems, such as the hyperreals,
do contain nonzero infinitesimals;
and then the symbol "0.999..." admits the interpretation
of falling infinitesimally short of 1.
The equality 0.999... = 1 has long been accepted by mathematicians
because they are concerned with real numbers not hyperreals.
12.27: me:
0.999... is not a real number; it's a hyperreal;
because it is equal to 1 -1/infinity (the infinitesimal);
making it infinitely close to 1 but not real;
that's why 0.999... can't be equal to 1;
1 is a real; 0.999... is a hyperreal.
here's why I have a problem with (0.999...) = 1:
. if the number (0.999...) = (1 - 1/infinity) = 1;
then the set [0,1) = {0, ... 1- 1/infinity} = {0, ... 1} = [0..1];
but then we have [0,1) = [0,1] so did we want to mean that?
12.17: wiki:
The equality of 0.999... and 1 is closely related to
the absence of nonzero infinitesimals in the real number system,
the most commonly used system in mathematical analysis.
Some alternative number systems, such as the hyperreals,
do contain nonzero infinitesimals;
and then the symbol "0.999..." admits the interpretation
of falling infinitesimally short of 1.
The equality 0.999... = 1 has long been accepted by mathematicians
because they are concerned with real numbers not hyperreals.
12.27: me:
0.999... is not a real number; it's a hyperreal;
because it is equal to 1 -1/infinity (the infinitesimal);
making it infinitely close to 1 but not real;
that's why 0.999... can't be equal to 1;
1 is a real; 0.999... is a hyperreal.
2015-12-27
Raffi Khatchadourian`doomsday invention #robotics
12.14: cs/ai/Raffi Khatchadourian`doomsday invention:
. an article in the New Yorker, nov 23, p64,
Raffi Khatchadourian "the doomsday invention:
will artificial intelligence destroy us?":
. Nick Bostrom wrote in "superintelligence:
paths, dangers, strategies"
that humans risk being killed off when the robotic network
"gains the ability to improve itself,
and in short order exceeds the intellectual potential
of the human brain by many orders of magnitude."
. an article in the New Yorker, nov 23, p64,
Raffi Khatchadourian "the doomsday invention:
will artificial intelligence destroy us?":
. Nick Bostrom wrote in "superintelligence:
paths, dangers, strategies"
that humans risk being killed off when the robotic network
"gains the ability to improve itself,
and in short order exceeds the intellectual potential
of the human brain by many orders of magnitude."
2015-08-03
#robotics beware offensive autonomous weapons
8.1: news.cs/robotics/beware offensive autonomous weapons:
8.3: summary:
. we are getting close to robotics that are
so talented they may serve as soldiers;
but scientists warn we should avoid
such a robotics arms race.
. supercomputers will be used to design
robotic soldiers we don't fully understand.
. robots designed for mutually assured destruction
could cause the extinction of the human race
much more effectively than nuclear weapons.
8.3: summary:
. we are getting close to robotics that are
so talented they may serve as soldiers;
but scientists warn we should avoid
such a robotics arms race.
. supercomputers will be used to design
robotic soldiers we don't fully understand.
. robots designed for mutually assured destruction
could cause the extinction of the human race
much more effectively than nuclear weapons.
2015-05-30
Linus Torvalds`thoughts on #Python
2.19: co.quora/adda/lang/python/Linus Torvalds`thoughts on #Python:
co.quora.com/ What-does-Linus-Torvalds-think-of-Python:
co.quora.com/ What-does-Linus-Torvalds-think-of-Python:
eternal civilization foundations
2.18: news.adds/eternal civilization foundations:
openculture introduces longnow.org:
. longnow's Manual for Civilization
aims to select 3,500 books deemed to be
most likely to sustain or rebuild civilization.
openculture introduces longnow.org:
. longnow's Manual for Civilization
aims to select 3,500 books deemed to be
most likely to sustain or rebuild civilization.
2015-02-20
Linus Torvalds`use asciidoc
2.19: web.adda/markup/Linus Torvalds`use asciidoc:
summary:
. my programming language, adda,
includes a markup language
so that simple ascii documentation
can be converted to html with advanced features
like hyperlinks, and formatting .
. xml and html code is written in ascii
but it's not as readable by humans
since it's designed primarily for machines .
. Linus Torvalds was famous for
condemning xml and supporting asciidoc
so I became interested in studying asciidoc
and incorporating parts of that into adda markup .
. there is an asciidoc renderer for Chrome OS .
summary:
. my programming language, adda,
includes a markup language
so that simple ascii documentation
can be converted to html with advanced features
like hyperlinks, and formatting .
. xml and html code is written in ascii
but it's not as readable by humans
since it's designed primarily for machines .
. Linus Torvalds was famous for
condemning xml and supporting asciidoc
so I became interested in studying asciidoc
and incorporating parts of that into adda markup .
. there is an asciidoc renderer for Chrome OS .
2015-01-09
spie.org #phy #library
2014.12.12: lib.phy/spie.org/samples:
SPIE the international society for optics and photonics,
is advancing an interdisciplinary approach
to the science and application of light.
SPIE publishes the SPIE Digital Library,
containing more than 400,000 research papers
from the Proceedings of SPIE and the Society's
10 scholarly journals with around 18,000 new papers added each year,
and more than 195 eBooks from the SPIE Press catalog.
The SPIE Press publishes print monographs, tutorial texts,
Field Guides, and reference books.
SPIE also publishes a wide variety of open access content.
SPIE the international society for optics and photonics,
is advancing an interdisciplinary approach
to the science and application of light.
SPIE publishes the SPIE Digital Library,
containing more than 400,000 research papers
from the Proceedings of SPIE and the Society's
10 scholarly journals with around 18,000 new papers added each year,
and more than 195 eBooks from the SPIE Press catalog.
The SPIE Press publishes print monographs, tutorial texts,
Field Guides, and reference books.
SPIE also publishes a wide variety of open access content.
2015-01-04
magnetic fields affect the gravitational field #phy
2014.11.16: news.phy/aether/Dr.Judy Wood`
magnetic fields affect the gravitational field:
Dr. Judy Wood (2hr:43min):
magnetic fields affect the gravitational field:
Dr. Judy Wood (2hr:43min):
see Boyd Bushman experiments:
. when magnets are pushing against eachother
they are less prone to the pull of gravity
they fall slower than a non-magnetic control;
. 2 magnets arranged so that they are attracting each other
fall faster than the control:
being attracted by magnetism
makes them more prone to gravity's pull .
Subscribe to:
Posts (Atom)
