The Risks Digest

The RISKS Digest

Forum on Risks to the Public in Computers and Related Systems

ACM Committee on Computers and Public Policy, Peter G. Neumann, moderator

Volume 30 Issue 44

Thursday 31 August 2017

Contents

U.S. National Infrastructures
Henry Petroski via PGN
Taiwan Grid Outage Caused By Human Error
Rob Wilcox
Pacemaker firmware updates
Peter Gregory
Donald Trump's cybersecurity advisers resign, warning of 'insufficient attention to the growing threats'
Chris Baynes
FBI pushes private sector to cut ties with Kaspersky
CyberScoop
WikiLeaks Turned Down Leaks on Russian Government During U.S. Presidential Campaign
Foreign Policy
The Crisis of Connected Cars: When Vulnerabilities Affect the CAN Standard
Trend Micro
Quebec man fights back after dealer remotely disables car over $200 fee
CBC
Yu Pingan arrested for involvement in hacking OPM
Gizmodo
US Voting Machine Supplier Leaks 1.8 Million Chicago Voter Records
Gizmodo
DreamHostStatus.com forgot to use separate nameservers
Dan Jacobson
Cracked screen => cracked security?
Dan Goodin
Identity Thieves Hijack Cellphone Accounts to Go After Virtual Currency
The NYTimes
Google Accidentally broke the Internet throughout Japan
Engadget
Apple, Facebook, Google and others sign brief concerned about warrantless location tracking
Roger Fingas
"Even Artificial Neural Networks Can Have Exploitable 'Backdoors'"
WiReD
`Devil's Ivy' Is Another Wake-Up Call for IoT Security
Threatpost
US Army backing off a bit from its decision regarding sUAS usage
Gary Mortimer
Aero-ease
Aeon
98.5% of unique net neutrality comments oppose Ajit Pai's anti-Title II plan
Ars Techica
Risks of IBAN checksums
Paul van Keep
Ethereum Hack
Bruce Schneier
I knew what you were going to do next: AI learns from pro gamers, then crushes them
The Washington Post
How Peter Thiel's Secretive Data Company Pushed Into Policing
WiReD
From Isaac Asimov to Aimee Mann, 'robophobia' plagues humans
WashPo
Carl Sagan in 1995
Rich Kulawiec
UK Today's Roads Aren't Good Enough for Driverless Cars
Chris Drewe
Uh oh—too easy to confuse self-driving cars
IEEE Spectrum via Gabe Goldberg
Re: "Driverless" van in Virginia
Don Norman
Re: Is LIBOR, Benchmark for Trillions of Dollars in Transactions, a Lie?
Amos Shapir
Re: The Death of Ruby? Developers Should Learn These Languages Instead
Amos Shapir
Re: Botched Firmware Update Bricks Hundreds of Smart Door Locks
Michael Bacon
Re: Microchipping employees
David Randolph
Lindsay Marshall named UK National Teaching Fellow
PGN
Info on RISKS (comp.risks)

U.S. National Infrastructures (Henry Petroski)

"Peter G. Neumann" <neumann@csl.sri.com>
Thu, 31 Aug 2017 7:42:06 PDT
I've been following Henry's work even before the publication of
  To Engineering Is Human: The Role of Failure in Successful Design.

See RISKS-3.25, 9.15, 9.16, 12.51, 18.61, 20.61, 26.80 for previous items.
His latest article is a real blockbuster, and deserves your attention.

  Henry Petroski, The State of Our Infrastructure, *American Scientist*,
  September-October 2017, pp. 274--277
  https://www.americanscientist.org/article/the-state-of-our-infrastructure

This article picks up from his previous article on this topic in that
journal 8 years ago.  His latest take on this subject includes a Report Card
on U.S. intrastructures based mostly on ASCE evaluations from 1998 to 2017.
Essentially every infrastructure sector—Highways, Mass Transit, Aviation,
Water.  Schools, Energy, etc.—had a grade wallowing around in the range
from D+ to D-.  Bridges, Solid Waste and Ports actually achieved a C+ grade
in 2017.  The only notable improvement involved the Rail sector, which had
climbed from a C- to a B.  However, the estimated investment for remediation
in 2013 had risen to $3.6 trillion total by 2020—which Henry notes is
almost as much as the entire current federal budget.  The 2017 estimate is
3.5% of GDP (until 2025).  Considering there has been very little effort to
even begin, we are just kicking the can further down the road.

For greater depth, see Henry's 2016 book, The Road Taken: The History and
Future of America's Infrastructures.

You may wonder why I am putting this item in a forum devoted to
computer-related risks.  There are two primary reasons.  (1) Many of these
infrastructures are monitored by and controlled by computer systems that are
not secure, reliable, or in some cases not sufficiently respectful of needs
for human safety.  In some cases, the shortcomings of the computer systems
may be contributing to the low grades of the infrastructures.  (2) The same
miserable grades could be allocated to the security and integrity of
computer systems and networks.

I have long written on the risks of short-term optimization and the need for
the proactive and holistic long-term thinking that is required to prevent
this sort of pervasive degeneration.  In some sense, the lack of that
thinking is continually making matters worse, and making any remediation
even more difficult (politically, economically, and realistically).  The
same comment also intensifies the potential implications of climate change
on most of these infrastructures.  PGN


Taiwan Grid Outage Caused By Human Error

Rob Wilcox <robwilcoxjr@gmail.com>
Sun, 20 Aug 2017 11:31:55 -0700
I study grid operations. Major blackouts are studied like air disasters.
The cause is almost always human error compounded by a lack of situational
awareness. I would classify that as a user experience UX design failure.

At 16:52 local time August 15, 2017 in the Taiwan grid became unbalanced and
protection systems shut portions down as designed. A routine maintenance
error caused the failure of a six unit natural gas power plant supplying
about 12% of the country's load.

The Tatan power plant is fueled by liquified natural gas. National gas
company maintenance staff was replacing a power supply for a control system
governing the flow of natural gas to the generators. They did not switch the
connected control systems to manual control, leaving them on the automatic
setting.

The connected control systems automatically closed two valves supplying gas
to the generators for several minutes.

The Taiwan grid was operating close to the Summer load peak at the time,
due to hot weather.

Power was fully restored to the country about 4 1/2 hours later.

The Minister of Economics Affairs Chih-kung Lee and the chairman of the
national gas company Chen Chin-de have resigned as a result of the blackout.

The automatic protection systems in the electric grid shut it down quickly
when generation and load become unbalanced.

Usually the grid will divide into working islands and outage islands.

To restart the grid, a "black start," islands of generation have to be
brought up in tandem with islands of load in exact balance while managing
transmission constraints. That is a manual process by generator staff, field
staff and operations control center staff. It is also governed by the
maximum ramping speed of each generator.

Blackouts are rare. There is not much first hand operational experience in
black starts. Each utility will have written restoration plans. The control
center staff trains black starts on simulators.

Bringing up the grid when there is a large air conditioner load is
complicated by motor stall current and voltage excursions.

Yes, the grid is analog, with humans in the loop!

Blackout and Taiwan energy strategy:
https://www.bloomberg.com/news/articles/2017-08-16/taiwan-s-president-apologizes-for-blackout-affecting-millions

Standard Operating Procedure Not Followed (8th time is the charm!):
http://focustaiwan.tw/search/201708160017.aspx?q=blackout

Failure and restoration:
http://focustaiwan.tw/news/aeco/201708150033.aspx

Weather drives August peak load:
http://focustaiwan.tw/news/aeco/201708070018.aspx


Pacemaker firmware updates

Peter Gregory <Peter.Gregory@optiv.com>
Wed, 30 Aug 2017 15:02:20 +0000
The U.S. Food and Drug Administration issued an alert regarding the recall
of network-connected pacemakers from St. Jude Medical, now Abbott
Laboratories. Apparently some 465,000 people are affected.

It's one thing to do a firmware update on one's laptop, tablet, or mobile
device, or for a router, firewall, doorbell, or thermostat.  But what if a
pacemaker is bricked after a user (or their physician) updates the firmware?
My heart flutters at this prospect.  And I dare not think of a ransomware
attack on a pacemaker - how would that work?

https://www.bankinfosecurity.com/medical-device-recall-a-10238

Peter H Gregory | Executive Director - CISO Services
peter.gregory@optiv.com<mailto:peter.gregory@optiv.com>

  [A long-time colleague of mine with close first-hand (and first-heart)
  experience had this response when I shared the above with him:

    For some reason neither of the Canadian hospitals used that remote
    update/sensing feature.  The same was true in Ireland (where some
    doctors did not understand the pacemaker maintenance system at all).
    They both thought that a patient should be in their presence and
    examined personally before doing anything.

  PGN]


Donald Trump's cybersecurity advisers resign, warning of 'insufficient attention to the growing threats' (Chris Baynes)

Shannon McElyea <shannonm@gmail.com>
August 29, 2017 at 8:52:32 PM EDT
Chris Baynes, *The Independent*, 28 Aug 2017, via Dave Farber's IP.

The panel is tasked with advising the US Homeland Security Department on
cybersecurity and the protection of infrastructure.  The eight departing
members accused Trump's administration of failing to be "adequately
attentive to the pressing national security matters" or "responsive to sound
advice received from experts".

  “Your actions have threatened the security of the homeland I took an oath
  to protect,'' said their letter, obtained by IT news website Nextgov.

http://www.independent.co.uk/news/world/americas/us-politics/donald-trump-cyber-security-advisers-resign-growing-threat-charlottesville-a7916496.html?cmpid=3Dfacebook-post

Donald Trump's cyber-security advisers resign warning of 'insufficient
attention to the growing threats'


FBI pushes private sector to cut ties with Kaspersky (Cyberscoop)

Lauren Weinstein <lauren@vortex.com>
Sat, 19 Aug 2017 17:26:41 -0700
  The FBI has been briefing private sector companies on intelligence
  claiming to show that the Moscow-based cybersecurity company Kaspersky Lab
  is an unacceptable threat to national security, current and former senior
  U.S. officials familiar with the matter tell CyberScoop.  The briefings
  are one part of an escalating conflict between the U.S.  government and
  Kaspersky amid long-running suspicions among U.S.  intelligence officials
  that Russian spy agencies use the company as an intelligence-gathering
  tool of global proportions.
https://www.cyberscoop.com/fbi-kaspersky-private-sector-briefings-yarovaya-laws/


WikiLeaks Turned Down Leaks on Russian Government During U.S. Presidential Campaign (Foreign Policy)

Lauren Weinstein <lauren@vortex.com>
August 17, 2017 at 8:51:10 PM EDT
http://foreignpolicy.com/2017/08/17/wikileaks-turned-down-leaks-on-russian-government-during-u-s-presidential-campaign/

  In the summer of 2016, as WikiLeaks was publishing documents from
  Democratic operatives allegedly obtained by Kremlin-directed hackers,
  Julian Assange turned down a large cache of documents related to the
  Russian government, according to chat messages and a source who provided
  the records.  WikiLeaks declined to publish a wide-ranging trove of
  documents—at least 68 gigabytes of data—that came from inside the
  Russian Interior Ministry, according to partial chat logs reviewed by
  Foreign Policy.


The Crisis of Connected Cars: When Vulnerabilities Affect the CAN Standard (Trend Micro)

Martyn Thomas <martyn@thomas-associates.co.uk>
Fri, 25 Aug 2017 12:46:31 +0100
http://blog.trendmicro.com/trendlabs-security-intelligence/connected-car-hack/

" ... what should the security industry's response be when a hack is found
that is not only successful in being able to drastically affect the
performance and function of the car, but is also stealthy and vendor
neutral?  Enter the hack that does just that—one that has been discovered
and proven to be effective ..."


Quebec man fights back after dealer remotely disables car over $200 fee (CBC)

Jose Maria Mateos <chema@rinzewind.org>
Tue, 29 Aug 2017 12:45:45 -0400
http://www.cbc.ca/news/canada/montreal/quebec-man-fights-back-after-dealer-remotely-disables-car-over-200-fee-1.4265588

A car dealership in Sherbrooke, Que., may have broken the law when it used a
GPS device to disable the car of a client who was refusing to pay an extra
$200 fee, say consumer advocates consulted by CBC News.

Bury, Que., resident Daniel Lallier signed a four-year lease for a Kia Forte
LX back in May from Kia Sherbrooke. Two months later, the 20-year-old's
grandmother offered to buy the car outright when he lost his job and
couldn't make his weekly payments.

After settling the balance and paying a $300 penalty, Lallier said, the
dealership told him he would have to pay an additional $200 to remove a GPS
tracker that had been installed on the car.  The device allows dealers to
remotely immobilize a car in case lease payments are in arrears. [...]

After refusing to pay the fee, a mechanic notified Lallier by text message
that his car was being remotely disabled until the dealership recovered the
device and $200 fee.  "I went outside and tested my car, and it wouldn't
work at all.  It wouldn't start period, and I got angry," Lallier said.


Yu Pingan arrested for involvement in hacking OPM (Gizmodo)

"Peter G. Neumann" <neumann@csl.sri.com>
Mon, 28 Aug 2017 10:55:02 PDT
https://gizmodo.com/fbi-arrest-chinese-national-linked-to-opm-data-breach-m-1798411342

"A 36-year-old Chinese national was arrested in Los Angeles this week in
connection with a computer hacking conspiracy involving malware linked to
the 2014 US Office of Personnel Management (OPM) data breach.

Yu Pingan of Shanghai, China, was arrested on Wednesday while traveling at
Los Angeles International Airport. Also identified by the hacker pseudonym
“GoldSun,” Yu has been charged under the Computer Fraud and Abuse Act and is
further accused of conspiracy to commit offense or defraud the United
States."


US Voting Machine Supplier Leaks 1.8 Million Chicago Voter Records (Gizmodo)

Lauren Weinstein <lauren@vortex.com>
Thu, 17 Aug 2017 13:51:29 -0700
via NNSquad
http://gizmodo.com/us-voting-machine-supplier-leaks-1-8-million-chicago-vo-1797947510

  A leading US supplier of voting machines confirmed on Thursday that it
  exposed the personal information of more than 1.8 million Illinois
  residents.  State authorities and the Federal Bureau of Investigation were
  alerted this week to a major data leak exposing the names, addresses,
  dates of birth, partial Social Security numbers, and party affiliations of
  over a million Chicago residents. Some driver's license and state ID
  numbers were also exposed.


DreamHostStatus.com forgot to use separate nameservers

Dan Jacobson <jidanni@jidanni.org>
Sat, 26 Aug 2017 06:45:53 +0800
To ensure users could still see status reports even when DreamHost.com was
down (e.g., during a DDoS attack), the separate DreamHostStatus.com was
established.

Alas, they forgot to also use separate nameservers...


Cracked screen => cracked security? (Dan Goodin)

Henry Baker <hbaker1@pipeline.com>
Sat, 19 Aug 2017 13:55:22 -0700
Dan Goodin - Aug 18, 2017 12:27 pm UTC
https://arstechnica.com/information-technology/2017/08/a-repair-shop-could-completely-hack-your-phone-and-you-wouldnt-know-it/

Secret chips in replacement parts can completely hijack your phone's security
Booby-trapped touchscreens can log passwords, install malicious apps, and more.

People with cracked touch screens or similar smartphone maladies have a new
headache to consider: the possibility the replacement parts installed by
repair shops contain secret hardware that completely hijacks the security of
the device.

The Mafioso of old never allowed repairmen into their homes.  Stories abound
regarding multiplicities of dead washing machines, TV's, etc.

It appears that their fears were justified.

On the other hand, these stories play right into the hands of those trying
to kill "the right to repair" supported by the EFF.

  [Also posted to
http://www.metzdowd.com/mailman/listinfo/cryptography
  PGN]


Identity Thieves Hijack Cellphone Accounts to Go After Virtual Currency (The NYTimes)

"Bob Frankston" <Bob19-0501@bobf.frankston.com>
22 Aug 2017 09:38:11 -0400
The New York Times, 21 Aug 2017
https://www.nytimes.com/2017/08/21/business/dealbook/phone-hack-bitcoin-virtual-currency.html

Yet another reminder of the risks of simplifying assumptions. In this case
assuming that email and phone calls are a secure form of identity when they
are really just creating a focus for attacks. It's also a reminder of the
reason why we money isn't just a technology but part of larger social
systems and why the challenge of establishing trust is so difficult. Let's
not forget how many mechanisms pile on a DNS that doesn't even let you own
your identity.

  [Gabe Goldberg commented on this one as well:
  So-called phone porting attacks are exposing a vulnerability that could be
  exploited against anybody with valuable emails or other digital files.
  PGN]


Google Accidentally broke the Internet throughout Japan (Engadget)

"Dave Farber" <farber@gmail.com>
Mon, 28 Aug 2017 17:21:22 -0400
https://www.engadget.com/2017/08/28/google-accidentally-broke-internet-japan/


Apple, Facebook, Google and others sign brief concerned about warrantless location tracking (Roger Fingas)

geoff goodfellow <geoff@iconia.com>
Tue, 15 Aug 2017 11:32:52 -1000
Roger Fingas, Apple Insider,15 Aug 2017
http://appleinsider.com/articles/17/08/15/apple-facebook-google-others-sign-brief-concerned-about-warrantless-location-tracking

Several high-profile technology companies, including Apple, have submitted a
amicus brief in a key case at the U.S. Supreme Court, expressing concerns
about warrantless police access to cellphone location data.

Other tech firms listed in the brief include Airbnb, Cisco, Dropbox,
Evernote, Facebook, Google, Microsoft, Mozilla, Snap, Twitter, and Verizon.
Collectively, the companies argue that the court should "refine the
application of certain Fourth Amendment doctrines to ensure that the law
realistically engages with Internet-based technologies and with people's
expectations of privacy in their digital data."

The case in question is Timothy Carpenter v. United States. Police obtained
Carpenter's location history without a warrant, leading to his eventual
robbery conviction. At court he's being represented by the American Civil
Liberties Union, which says that the government violated Fourth Amendment
rights against search and seizure...SNIP


"Even Artificial Neural Networks Can Have Exploitable 'Backdoors'"

Stuart Shapiro <s_shapiro@ACM.ORG>
Fri, 25 Aug 2017 14:11:25 -0400
"The stunt demonstrated a potential security headache for engineers working
with machine-learning software.  The researchers showed it's possible to
embed silent, nasty surprises into artificial neural networks, the type of
learning software used for tasks such as recognizing speech or understanding
photos.

For their part, the NYU researchers are thinking about how to make tools
that would let coders peer inside a neural network from a third party and
spot any hidden behavior. Meanwhile? Buyer beware."

https://www.wired.com/story/machine-learning-backdoors

That last bit could have relevance for validation and testing more generally.


`Devil's Ivy' Is Another Wake-Up Call for IoT Security (Threatpost)

Gabe Goldberg <gabe@gabegold.com>
Tue, 15 Aug 2017 19:17:28 -0400
“In the case of this camera, in order to exploit the vulnerability you
would need to send a malicious payload to port 80,'' M Carlton, Senrio's
vice president of research, told the website Threatpost.
<https://threatpost.com/bad-code-library-triggers-devils-ivy-vulnerability-in-millions-of-iot-devices/126913/>
“The camera then processes the data using the vulnerable library. The
attacker then sends the specially crafted payload that triggers the buffer
stack overflow which leads to custom code execution.''

With the Axis cameras, after exploiting the vulnerability, Senrio
researchers could reboot a device and change settings to block access to the
video feed. More disturbingly, a device could also be reset to factory
defaults, which would cause it to issue a prompt to change the user name and
password, after which attackers would have complete control of the device.
In other words, tech savvy thieves could use this exploit to turn off
security cameras before pulling off a heist, and security personnel wouldn't
be able to quickly get the cameras back up and running.

http://windowsitpro.com/internet-things-iot/devils-ivy-another-wake-call-iot-security


US Army backing off a bit from its decision regarding sUAS usage (Gary Mortimer)

john hight <johnhight@gmail.com>
Tue, 15 Aug 2017 18:39:27 -0700
Gary Mortimer, sUAS News: Aug 2018

An exception to policy with recommendations from the asymmetric warfare
group that will permit the use of DJI kit once some conditions have been
met.  The Android Tactical Assault Kit will become the ground-control
station (GCS) of choice when a DJI plugin has passed OPSEC (Operational
Security) scrutiny.  It was developed by the Air Force Research Lab (AFRL),
Army Research Laboratory (ARL) and the Defense Advanced Research Projects
Agency (DARPA).

https://www.suasnews.com/2017/08/us-army-walks-back-dji-decision-slightly/


Aero-ease (Aeon)

Mark Boolootian <booloo@ucsc.edu>
Thu, 24 Aug 2017 12:07:15 -0700
A wonderful article on the parlance of pilots:

https://aeon.co/essays/the-language-of-the-cockpit-is-technical-obscure-and-irresistibly-romantic


98.5% of unique net neutrality comments oppose Ajit Pai's anti-Title II plan (Ars Techica)

Lauren Weinstein <lauren@vortex.com>
Wed, 30 Aug 2017 09:23:02 -0700
via NNSquad
https://arstechnica.com/tech-policy/2017/08/isp-funded-study-finds-huge-support-for-keeping-current-net-neutrality-rules/

  A study funded by Internet service providers has found something that
  Internet service providers really won't like.  The overwhelming majority
  of people who wrote unique comments to the Federal Communications
  Commission want the FCC to keep its current net neutrality rules and
  classification of ISPs as common carriers under Title II of the
  Communications Act, according to the study released today.  The study
  (available here) was conducted by consulting firm Emprata and funded by
  Broadband for America, whose members include AT&T, CenturyLink, Charter,
  CTIA-The Wireless Association, Comcast, NCTA-The Internet & Television
  Association, the Telecommunications Industry Association (TIA), and
  USTelecom.


Risks of IBAN checksums

Paul van Keep <paul@vankeep.com>
Wed, 16 Aug 2017 11:32:23 +0200
Most bank account numbering systems incorporate some sort of checksum into
their numbering scheme to avoid simple transcription mistakes.  Dutch bank
accounts rely on the eleven-test (elfproef:
https://nl.wikipedia.org/wiki/Elfproef) and the European successor IBAN uses
the 97 check (see:
https://en.wikipedia.org/wiki/International_Bank_Account_Number).
Surprisingly, even with both checks combined, these safeguards can fail to
do their job quite easily as I found out last month.  At the beginning of
July I was supposed to get quite a substantial payment from a financial
institution.  But when the money failed to show up in my account after a few
days I called the company.  The error was then quickly uncovered.  The
account I supplied to them on my contract, in my handwriting, ended in
719.  The person who entered the details into their system interpreted the 7
as a 9 and then the 9 as a 3.  That resulted in a valid account number for
the eleven-test (7*3+1*2+9*1 = 32 and 9*3+1*2+3*1 = 32), but also produced
the exact same checksum for the IBAN 97 check (719 / 97 = 7 remainder 40 and
913 / 97 = 9 remainder 40).

So, even though the two checksum systems look very different, it turns out
that it's really easy to produce a hash collision with just a two digit
change.  In this case I did get my money a few days later and I assume the
initial recipient didn't get to enjoy his or her new found wealth for very
long.

The risks: relying on two checksums to validate manual input isn't enough
(and my handwriting is illegible).

Paul van Keep

  [The "check" is in the "fail"!]


Ethereum Hack

Bruce Schneier <schneier@schneier.com>
Tue, 15 Aug 2017 00:01:12 -0500
CRYPTO-GRAM, August 15, 2017 [PGN-excerpted for RISKS]
Bruce Schneier, CTO, IBM Resilient, schneier@schneier.com
https://www.schneier.com
<https://www.schneier.com/crypto-gram/archives/2017/0815.html>

The press is reporting a $32M theft of the cryptocurrency Ethereum. Like all
such thefts, they're not a result of a cryptographic failure in the
currencies, but instead a software vulnerability in the software surrounding
the currency—in this case, digital wallets. This is the second Ethereum
hack this week. The first tricked people in sending their Ethereum to
another address.

This is my concern about digital cash. The cryptography can be bulletproof,
but the computer security will always be an issue.

https://motherboard.vice.com/en_us/article/zmvkke/this-is-not-a-drill-a-hacker-allegedly-stole-dollar32-million-in-ethereum
https://www.cryptocoinsnews.com/hackers-seize-32-million-in-parity-wallet-breach/
30-million-worth-of-ethereum/

The first hack:
https://www.bleepingcomputer.com/news/security/hacker-steals-7-million-worth-of-ethereum-from-coindash-platform/
https://motherboard.vice.com/en_us/article/zmvg58/hacker-allegedly-steals-dollar74-million-in-ethereum-with-incredibly-simple-trick


I knew what you were going to do next: AI learns from pro gamers, then crushes them (WashPo)

Monty Solomon <monty@roscom.com>
Tue, 15 Aug 2017 23:14:24 -0400
`It knew what you were going to do next': AI learns from pro gamers — then
crushes them.  It only took the bot a few weeks to go from novice to world
class.

https://www.washingtonpost.com/news/innovations/wp/2017/08/15/it-knew-what-you-were-going-to-do-next-ai-learns-from-pro-gamers-then-crushes-them/


How Peter Thiel's Secretive Data Company Pushed Into Policing (WiReD)

Lauren Weinstein <lauren@vortex.com>
Sun, 20 Aug 2017 09:55:36 -0700
via NNSquad
https://www.wired.com/story/how-peter-thiels-secretive-data-company-pushed-into-policing/

  The scale of Palantir's implementation, the type, quantity and persistence
  of the data it processes, and the unprecedented access that many thousands
  of people have to that data all raise significant concerns about privacy,
  equity, racial justice, and civil rights. But until now, we haven't known
  very much about how the system works, who is using it, and what their
  problems are. And neither Palantir nor many of the police departments that
  use it are willing to talk about it.


From Isaac Asimov to Aimee Mann, 'robophobia' plagues humans (WashPo)

Lauren Weinstein <lauren@vortex.com>
Tue, 15 Aug 2017 21:21:33 -0700
WashPo via NNSquad
https://www.washingtonpost.com/national/from-isaac-asimov-to-aimee-mann-robophobia-plagues-humans/2017/08/16/25b21096-8239-11e7-9e7a-20fa8d7a0db6_story.html

  Robots are secretly plotting to kill us. Or enslave us. Or, at best, they
  will take our jobs, one by one.  From science fiction written by Isaac
  Asimov eight decades ago to "Dilbert" cartoons today, the relationship
  between robots and humans has long fascinated—and worried—people.
  There's even a term, "robophobia," for an irrational anxiety about robots
  and other advanced automation machines.

Positronically!


Carl Sagan in 1995 (Rich Kulawiec)

Rich Kulawiec <rsk@gsp.org>
August 14, 2017 at 6:39:08 PM EDT
  “I have a foreboding of an America in my children's or grandchildren's
  time—when the United States is a service and information economy; when
  nearly all the key manufacturing industries have slipped away to other
  countries; when awesome technological powers are in the hands of a very
  few, and no one representing the public interest can even grasp the
  issues; when the people have lost the ability to set their own agendas or
  knowledgeably question those in authority; when, clutching our crystals
  and nervously consulting our horoscopes, our critical faculties decline,
  unable to distinguish between what feels good and what's true, we slide,
  almost without noticing, back into superstition and darkness.  The dumb
  down of America is most evident in the slow decay of substantive content
  in the enormously influential media, the 30-second sound bites (now down
  to 10 seconds or less), lowest common denominator programming, credulous
  presentations on pseudoscience and superstition, but especially a kind of
  celebration of ignorance.''

  Carl Sagan, "The Demon-Haunted World: Science as a Candle in the Dark",
  1995


UK Today's Roads Aren't Good Enough for Driverless Cars

Chris Drewe <e767pmk@yahoo.co.uk>
Tue, 15 Aug 2017 21:35:12 +0100
A short article in this Saturday's newspaper's cars section features various
industry commentators on whether today's roads are good enough (at least in
the UK) for driverless cars.  One problem is that the cars can follow white
road markings easily, but on side streets or country lanes these are often
poor quality or just not there; reportedly the cars' cameras are only
black-and-white so it's difficult to see where the edge of the road is.
Heavy rain can defeat radar sensors, and rainwater on the road surface at
night can obscure lane markings, if there are any.  Then there's snow...
One commentator suggested that segregated dedicated driverless lanes may be
needed in cities.

Something that's intrigued me is road works.  In the UK, motorways are
usually three traffic lanes + hard shoulder (emergency lane) in each
direction; when major repairs are needed (happens a lot), often there's a
temporary crossover built into the central reservation (median strip) so
that traffic uses the shoulder and adjacent lane in one direction and the
other two lanes the other way (this is 'contraflow').  There are plenty of
signs, cones, reflective studs, temporary lane markings, etc. separating the
lanes, which humans can follow without too much difficulty (though I've
unintentionally taken an exit more than once—it feels like the Cresta
Run, driving through a canyon of cones!), but how would a driverless car
manage?  A possible fix would be to have an electronic map of the area which
could be transmitted to cars as they approach so that they can guide
themselves through, though somebody would have to take responsibility for
setting up the map and updating it as the works progress.


Uh oh—too easy to confuse self-driving cars

Gabe Goldberg <gabe@gabegold.com>
Thu, 17 Aug 2017 00:11:52 -0400
Slight Street Sign Modifications Can Completely Fool Machine Learning
Algorithms

It's very difficult, if not impossible, for us humans to understand how
robots see the world. Their cameras work like our eyes do, but the space
between the image that a camera captures and actionable information about
that image is filled with a black box of machine learning algorithms that
are trying to translate patterns of features into something that they're
familiar with. Training these algorithms usually involves showing them a set
of different pictures of something (like a stop sign), and then seeing if
they can extract enough common features from those pictures to reliably
identify stop signs that aren't in their training set.

This works pretty well, but the common features that machine learning
algorithms come up with generally are not “red octagons with the letters
S-T-O-P on them.” Rather, they're looking features that all stop signs
share, but would not be in the least bit comprehensible to a human looking
at them. If this seems hard to visualize, that's because it reflects a
fundamental disconnect between the way our brains and artificial neural
networks interpret the world.

http://spectrum.ieee.org/cars-that-think/transportation/sensors/slight-street-sign-modifications-can-fool-machine-learning-algorithms


Re: "Driverless" van in Virginia

Don Norman <dnorman@ucsd.edu>
Thu, 17 Aug 2017 17:48:44 -0700
One of the many risks of the RISKS digest is that uninformed people use it
to make fun of legitimate research.  Why did our esteemed moderator let this
one in?  I have a theory that it was a honey pot, intended to lure me into
making a response. Well, Peter, if that is so, then it worked.

In RISKS 30.43, a reader responded to a news article about people at
Virginia Tech dressed in car suits. "What is the Risk?" he asked.  "Is it a
study to see if people freak out at the sight of a "driverless" van?"

Wendy Ju, a research scientist at Stanford University studying autonomous
vehicles invented the clever trick of studying driverless cars by hiding a
real driver inside a suit made of the same upholstery used for the car
seat. The driver is not visible through the window of the car. However, the
driver can see out through the loosely woven fabric and can readily control
the car. (A closer look reveals that the driver's seat is thicker than the
passenger's, but in our studies, nobody has ever noticed that.)

Why? Consider the communication between driverless vehicles and road users,
where road users are cars with drivers, motorcycles, bicycles,
skateboarders, pedestrians, etc.  Why?  well, suppose you want to cross the
street populated by truly driverless vehicles.  How do you know if they see
you? How do you know if you can cross? How would you wave them on? How would
they wave you on? (What if you waved one vehicle on but the others didn't
notice, so they continue moving?)

At the Design Lab at UC San Diego, we also constructed a car seat and are
testing solutions to these situations. Virginia Tech is doing the same and
we know of other groups as well. I can also assure you that Stanford, UC San
Diego, and VA Tech all have IRB (Institutional Review Board) approval to do
this work.

It is easy to think of solutions, when there is only one driverless car and
one road user, but what if there were many such cars and many road users?

The communication problem between driverless vehicles and road users is a
serious issue. Moreover, it requires standardization: if every automobile
company used their own signaling methods, the result would be chaos.

Together with the Nissan Research Center-Silicon Valley and the Toyota
Research Institute, the UCSD Design Lab recently co-sponsored a full day
standards meeting in San Francisco with multiple OEMs, relevant government
agencies, representatives of standards groups (ISO) and university research
labs, both from the US and Europe.  The preliminary results of all the
research were extremely useful.

I know it is easy and fun to joke about the notion of a driver in a car seat
suit, but it is legitimate, important research that has the potential to
save lives.

Moral: Don't make fun of an idea unless you know the whole story.

Rothenb=C3=BCcher, D., Li, J., Sirkin, D., Mok, B., & Ju, W. (2015). *Ghost
driver: a platform for investigating interactions between pedestrians and
driverless vehicles*. Paper presented at the Adjunct Proceedings of the 7th
International Conference on Automotive User Interfaces and Interactive
Vehicular Applications. from
http://dl.acm.org/citation.cfm?doid=3D2809730.2809755

Emmenegger, C., Risto, M., Bergen, B., Norman, D., & Hollan*,
J. (2016). *The Critical Importance of Standards for the Communication
Between Autonomous Vehicles and Humans*. Paper presented at the Automobile
Vehicle Systems conference.

Don Norman, Prof. and Director, DesignLab, UC San Diego
dnorman@ucsd.edu designlab.ucsd.edu/  www.jnd.org  <http://www.jnd.org/>


Re: Is LIBOR, Benchmark for Trillions of Dollars in Transactions, a Lie?

Amos Shapir <amos083@gmail.com>
Fri, 18 Aug 2017 14:11:06 +0300
When a bank owes $100 million it might fail, but if it owes $100 *billion*,
it's "too big to fail".  Extending that logic, a $350 *trillion *problem is
no problem at all, precisely because treating it as a problem means "the end
of economy as we know it".

The "revelation" about LIBOR is just as if in the 1960's, when every pound
Sterling banknote in the UK still included the statement "I promise to pay
the bearer the sum of 1 pound sterling silver", someone would have
"discovered" that the B of E does not really hand out bars of silver at
all...
Eventually that statement was unceremoniously removed, and about the same
time the USA had abandoned the gold base, and nothing really happened.

I assume that this is what's going to happen with LIBOR:  Banks would just
find another justification to the way it's value is determined, and
everyone would keep using that value in the same way.  As long as the value
"feels right" and everyone agrees to use it, it will remain useful.

This just underlines the simple fact that all money in any form --
including pure hard gold—is actually virtual, and has always been since
its invention.


Re: The Death of Ruby? Developers Should Learn These Languages Instead (Manning, RISKS-30.43)

Amos Shapir <amos083@gmail.com>
Fri, 18 Aug 2017 14:17:50 +0300
> ... Arthur C. Smith is shown teaching Electrical Engineering to a class of
> engineers sent to MIT by their employers for a year of upgrading.

But today, are there any employers willing to pay for a year of an
engineer's re-education?

They'd rather fire Sr. and hire Jr. instead for half the salary.


Re: Botched Firmware Update Bricks Hundreds of Smart Door Locks (Bacon, RISKS-30.43)

Michael Bacon - Grimbaldus <michael.bacon@grimbaldus.com>
Wed, 16 Aug 2017 19:36:39 +0100
Following my contribution of fishy puns, that were then battered [sea what I
did there?] by PGN, I must express a little surprise at his omission of the
obvious pun on the bricking of an IoT domestic portal access control device
... "DEAD LOCK".  [MB]

    [That would be very appropriate if you were in a deadly embrace with a
    giant squid.  Also, browse on "lock fish" and "fish lock"—with quite
    different meanings.  However, DEAD LOX would be tautologous.  PGN]


Re: Microchipping employees (RISKS-30.40)

"David Randolph" <dave@prairietrail.com>
Tue, 22 Aug 2017 17:54:12 -0500
The reports of a company putting microchips into their employees show that
they have fallen into the basic identification technology trap: that we can
build a technology that will uniquely and permanently identify someone.

Microchips work today because they are being used for pets. Once we use them
for identifying people for the purposes of moving money or goods and
services, people will figure out a way to fake them. It will be trivial to
design a microchip that not only reports the current id, but can be
reprogrammed to a new id from a simple device. Secondly, it will be fairly
easy to build a scanner that picks up the ids of anyone nearby. Quick scan
and reprogram and I am a new person with your credit limit.

David Randolph, Prairie Trail Software, Inc., Plano, TX


Lindsay Marshall named UK National Teaching Fellow

Peter Neumann <Neumann@csl.sri.com>
Thu, 31 Aug 2017 9:13:05 PDT
You should all know that Lindsay has single-handedly built the searchable
RISKS repository at Newcastle   <http://catless.ncl.ac.uk/Risks/>, and
has been maintaining it for lo these many years.  I am eternally indebted
to him, and wish to congratulate him on this award.

Brian Randell just informed me that Lindsay has been named a National
Teaching Fellow in the U.K.
  http://www.ncl.ac.uk/press/news/2017/08/nationalteachingfellow/

Lindsay is quoted:

  “Naturally, there have been many changes in teaching approaches and
  attitudes and this is particularly apparent in computing where the pace of
  change means it is essential to stay current, both in subject knowledge
  and teaching technique.''

  “Teaching and inspiring the next generation is both a pleasure and a
  privilege and I feel very honoured to have been nominated for this award.''

HEA Chief Executive, Professor Stephanie Marshall, said: “A National
Teaching Fellowship is the most prestigious individual award for excellence
in teaching in higher education. These awards represent a fantastic
achievement by all 55 new NTFs, and I am sure the whole sector joins me in
applauding them in their success.''

Please report problems with the web pages to the maintainer

Top