The RISKS Digest
Volume 30 Issue 91

Tuesday, 6th November 2018

Forum on Risks to the Public in Computers and Related Systems

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

Please try the URL privacy information feature enabled by clicking the flashlight icon above. This will reveal two icons after each link the body of the digest. The shield takes you to a breakdown of Terms of Service for the site - however only a small number of sites are covered at the moment. The flashlight take you to an analysis of the various trackers etc. that the linked site delivers. Please let the website maintainer know if you find this useful or not. As a RISKS reader, you will probably not be surprised by what is revealed…

Contents

Like clockwork: How daylight saving time stumps hospital record keeping
Sydney Lupkin
Daylight Savings results in hospital records shutdown
New Yorker
How Daylight Saving Time Messes With Hospitals
Fortune
File-Sharing Software on State Election Servers Could Expose Them to Intruders
ProPublica
Your brain: The next hacking frontier
TechBeacon
Selfie attempt results in damage to artwork by Dali and Goya
CNN
Facebook adding extra CGI parameters to other people's links
ycombinator
What it's like to use Tesla's newest self-driving tech
Gabe Goldberg
Why Big Tech pays poor Kenyans to programme self-driving cars
bbc.com
EU border `lie detector' system criticised as pseudoscience
The Guardian
Credit Card Chips Have Failed to Halt Fraud, Survey Shows
Fortune
Check this out: Radisson Hotel Group 'fesses up to `security incident'
The Register
A new study finds potentially manipulative ads in apps for preschoolers
WashPost
Cisco Adaptive Security Appliance Software and Cisco Firepower Threat Defense Software Denial of Service Vulnerability
Cisco
The D in Systemd stands for 'Dammmmit!' A nasty DHCPv6 packet can pwn a vulnerable Linux box.
The Register
T Wi-Fi kit bit by TI chip slip: Wireless gateways open to hijacking via BleedingBit chipset vulnerability
The Register
ISP pissed at Elsevier Takedowns/blocks, so...
danny burstein
Re: Ethics of whom to kill
Wol
Re: Explainable AI Simulation for AVs
Richard Stein
Erling Kristiansen
Re: Toward Human-Understandable, Explainable AI
John Beattie
Re: Driverless cars: Who should die in a crash?
John Beattie
Re: The spreading scourge of broken SSL implementation
Sergio Gelato
Julian Bradfield
Jury duty, recidivus
Rob Slade
Info on RISKS (comp.risks)

Like clockwork: How daylight saving time stumps hospital record keeping (Sydney Lupkin)

Jim Reisert AD1C <jjreisert@alum.mit.edu>
Sun, 4 Nov 2018 07:45:25 -0700
Sydney Lupkin, Kaiser Health News, 3 Nov 2018

https://www.usatoday.com/story/news/health/2018/11/03/daylight-saving-time-hospital-electronic-medical-records-emergency-fall-back/1864579002/

  Modern technology has helped medical professionals perform robot-assisted
  surgeries and sequence whole genomes. But hospital software still can't
  handle daylight saving time.

  Epic Systems, one of the most popular electronic health records software
  systems used by hospitals, can delete records or require cumbersome
  workarounds when clocks are set back for an hour—prompting many
  hospitals to opt for paper records for part of the night shift.

  And it happens every year.

  "It's mind-boggling," said Dr. Mark Friedberg, a senior physician policy
  researcher at RAND. In 2018, he said, "we expect electronics to handle
  something as simple as a time change."

  "Nobody is surprised by daylight savings time. They have years to prep.
  Only, surprise, it hasn't been fixed."


Daylight Savings results in hospital records shutdown (New Yorker)

Steve Golson <sgolson@trilobyte.com>
Tue, 6 Nov 2018 12:05:29 -0500
Problems with new electronic medical records system:

  Last fall, the night before daylight-saving time ended, an all-user e-mail
  alert went out. The system did not have a way to record information when
  the hour from 1 a.m. to 1:59 a.m. repeated in the night.  This was, for the
  system, a surprise event. The only solution was to shut down the lab
  systems during the repeated hour. Data from integrated biomedical devices
  (such as monitoring equipment for patients' vital signs) would be
  unavailable and would have to be recorded by hand. Fetal monitors in the
  obstetrics unit would have to be manually switched off and on at the top
  of the repeated hour.

The whole article is well worth reading:

https://www.newyorker.com/magazine/2018/11/12/why-doctors-hate-their-computers


How Daylight Saving Time Messes With Hospitals (Fortune)

Gabe Goldberg <gabe@gabegold.com>
Mon, 5 Nov 2018 17:29:20 -0500
The clocks went back one hour in (almost all) U.S. counties and states at 2
A.M. on Sunday, marking the `fall back' that signals
the end of Daylight Saving.
<https://click.email.fortune.com/%3Fqs%3D4781bb52c80c7dabf45d7dda982bf7332691a035149964328f3b1d2019da6f49e5165b3ed7f92087ee25bd7ca483ee0d75e810ac628562c1
And, as a report from Kaiser Health News highlights
<https://click.email.fortune.com/%3Fqs%3D4781bb52c80c7dabc89230e068d424c4153d8f87150658730d6df86797ea708c88a662f84d5a93723c93f10be314c25b6b57d510611fb2f4
that brings with it a whole bunch of technical headaches for hospital
systems and their electronic record keeping systems.

Modern medical innovations include the ability to transform human immune
cells into cancer-destroying mercenaries. And yet, a one-hour shifting of
clocks can force hospitals to temporarily switch from ostensibly newfangled
(and expensive) electronic health records to old-fashioned paperwork. In
fact, popularly used systems like Epic Systems software can delete records
or require cumbersome workarounds when clocks are set back for an hour,
according to KHN. (Epic, for its part, told the publication that, Daylight
savings time is inherently nuanced for healthcare organizations, which is
why we work closely with customers to provide guidance on how to most
effectively use their system to care for their patients during this time
period.)

One hour may not seem like a whole lot of time. But it can make a big
difference when it comes to keeping tabs on patients vitals or whether or
not they need scheduled medication.

https://view.email.fortune.com/%3Fqs%3D161a5916fd2cfcbc55f8fc149eae8b7ab098b460bce21e7b1922c7d87a7a9e9b37e1e68aa5fc8fc8eaddf8badad6ff68c28abd39418efffbeb08875c11c8ffbf5d76f3898e08b242

It's not just health IT that notices; databases, security systems, anything
logging events has to deal with a missing hour in spring and a duplicated
hour in fall.


File-Sharing Software on State Election Servers Could Expose Them to Intruders (ProPublica)

"Peter G. Neumann" <neumann@csl.sri.com>
Fri, 2 Nov 2018 19:37:27 PDT
ProPublica analysis found election computer servers in Wisconsin and
Kentucky could be susceptible to hacking by anonymous FTP.  Wisconsin shut
down its service after complaints.

https://www.propublica.org/article/file-sharing-software-on-state-election-servers-could-expose-them-to-intruders


Your brain: The next hacking frontier (TechBeacon)

Gabe Goldberg <gabe@gabegold.com>
Thu, 1 Nov 2018 20:51:13 -0400
This week, researchers unveiled worrying results about how easy it is to
hack medical implants, such as brain stimulators.

The claim is that hackers are a decade or two away from being able to mess
with our memories—the very essence of who we are. But neuro-modulation is
a promising branch of medical science, so it would be a shame if these
worries were overblown, right?

Sci-fi it's not, they claim. In this week's Security Blogwatch, we're
even more scared than we were yesterday.
<https://techbeacon.com/contributors/richi-jennings>

Your humble blogwatcher <http://richi.uk/> curated these bloggy bits for
your entertainment.  Not to mention: Thought-provoking stuff about nitrogen.

https://techbeacon.com/your-brain-next-hacking-frontier


Selfie attempt results in damage to artwork by Salvador Dali and Francisco Goya (CNN)

Jim Reisert AD1C <jjreisert@alum.mit.edu>
Sat, 3 Nov 2018 19:54:54 -0600
Not again......

Amir Vera, Jennifer Hauser and Alla Eshchenko, CNN

https://www.cnn.com/2018/11/03/europe/russian-women-damage-artwork/

8:13 PM ET, Sat November 3, 2018

A young woman trying to take a selfie knocked over two works of art at a
gallery in Yekaterinburg, Russia, on October 27, 2018.  A picture is worth a
thousand words, but what about a selfie?

A group of women in Yekaterinburg, Russia, may find out soon after one of
them tried to take a selfie on October 27 and accidentally knocked over a
structure at the International Arts Center Main Avenue. The structure was
carrying two works of art, according to the Russian Ministry of Internal
Affairs (MIA) and state-run news agency TASS.

The damaged artworks, according to TASS, include a Francisco Goya etching
from the Los Caprichos series and Salvador Dali's interpretation of
it. Goya's work was also part of the gallery owner's private collection.


Facebook adding extra CGI parameters to other people's links (ycombinator)

Eli the Bearded <*@eli.users.panix.com>
Sat, 3 Nov 2018 04:10:34 -0400
https://news.ycombinator.com/item%3Fid%3D18275061
https://community.cloudflare.com/t/facebook-now-adds-fbclid-query-string-to-urls-busting-cloudflares-cache/40355

In some apparent attempt to better track user clicks, Facebook has
started adding an extra parameter to links. This will break many
mechanisms for caching dynamic content, as the Cloudflare discussion
illustrates. In the case of my site it turns a URL like this:

  http://abc.def/ghijklmno.cgi%3Fpqrs%3Dtuvw
Into this:

  http://abc.def/ghijklmno.cgi%3Fpqrs%3Dtuvw%26fbclid%3DxyzR0bBzJRwc-q1btq_wHCtliXasz-C66UzxCc6DuqIBAYu9setNAg-IJ1nY8

(Censored to not advertise) Note how the parameter is *longer* than the
whole original URL. And this is not something I get any benefit from, I
do not use Facebook at all.

Besides breaking caching, it will destroy any CGI already using a fbclid
query parameter, has been breaking some links as reported in the
ycombinator piece, and it is also likely to seriously pollute other
people's log summaries.

I have decided that is a good thing, and have configured my site to now
generate 4xx errors in response to unexpected fbclid parameters. I don't
want people to think they can willy-nilly add extra things to CGI requests.
This needs to be coordinated with the target sites.

Unfortunately many people will decide they do need Facebook and will
rollover for this.


What it's like to use Tesla's newest self-driving tech

Gabe Goldberg <gabe@gabegold.com>
Fri, 2 Nov 2018 16:02:23 -0400
Autopass. If the person in front of you is driving too slowly—45 in a 55
mph zone, for example—what would you do? Why, you'd pass them.

Now, the Tesla can do that, too. If it notices that you're being blocked,
and that there's room in the next lane, a notification appears on your
screen. It informs you that if you put on your turn signal, Autopilot will
take it from there. It does the passing maneuver smoothly and
gracefully. (It doesn't actually return to your original lane, however --
just changes into a faster lane, passing the slowpoke, and stays there.)

How aggressive is it? That's up to you. In the onscreen settings, you can
adjust how impatient your car is. The options are Disabled (off), Mild,
Average, and Mad Max. In Mad Max mode, the Tesla will suggest passing if the
guy in front of you is going even a couple of mph below the speed limit.

(The Mad Max setting is characteristic of the Musk-esque sense of humor
that's baked in to Teslas. The acceleration options on the Model S are
labeled Chill, Standard, Sport, Insane, and Ludicrous.)

https://finance.yahoo.com/news/tesla-now-self-drivingest-car-road-063800677.html

Mad Max passing and Ludicrous acceleration. Just what the world needs.


Why Big Tech pays poor Kenyans to programme self-driving cars (bbc.com)

Richard Stein <rmstein@ieee.org>
Mon, 5 Nov 2018 12:10:37 +0800
https://www.bbc.com/news/technology-46055595

No need to build an explainable AI simulator when there's an army of
carbon-based trainers assisting AV neural network/image recognition learning
processes.

To their credit and initiative, Samasource's staffing model remotely and
inexpensively empowers Kenyan women. They construct the training images
applied to condition AV reactions/behavior.

"Brenda loads up an image, and then uses the mouse to trace around just
about everything. People, cars, road signs, lane markings—even the sky,
specifying whether it's cloudy or bright. Ingesting millions of these images
into an artificial intelligence system means a self-driving car, to use one
example, can begin to 'recognise' those objects in the real world. The more
data, the supposedly smarter the machine.

"She and her colleagues sit close—often too close—to their monitors,
zooming in on the images to make sure not a single pixel is tagged
incorrectly. Their work will be checked by a superior, who will send it back
if it's not up to scratch. For the fastest, most accurate trainers, the
honor of having your name up on one of the many TV screens around the
office. And the most popular perk of all: shopping vouchers."

Driver social skills, per https://catless.ncl.ac.uk/Risks/30/90%23subj12
(Shaprio), are neither integrated nor accountable. Training data set
localized bias may influence AV obstacle reaction.

A preference would be to apply training datasets that demonstrate courteous
v. aggressive driving, professional v. amateur, or reckless v.
cautious. Possibly based on US driving habits per Boston, Los Angeles, New
York, Miami, Philadelphia, Sydney AU, Beijing or Shanghai PRC, etc.  Use
real-time sequences (~50-100Hz) as training input. Clearly a very
challenging problem.

Risk: AV training strategy using discrete images discount localized
carbon-based driver intent and precursor conditions.

On 02OCT2018, the NHTSA published "A Framework for Automated Driving System
Testable Cases and Scenarios," retrieved on 04NOV2018 from
https://www.nhtsa.gov/document/framework-automated-driving-system-testable-cases-and-scenarios.
This document details a range of test scenarios for automated driving system
(ADS) response intervals from 0.1 to ~15 seconds (see document pg. 12 for
ADS task decomposition hierarchy).

This document does not establish or mandate compliance. Unclear if AV
manufacturers will be required to disclose ADS test results based on the
document and attach to the "car window sticker."


EU border `lie detector' system criticised as pseudoscience (The Guardian)

Jose Maria Mateos <chema@rinzewind.org>
Fri, 02 Nov 2018 07:44:21 -0400
https://www.theguardian.com/world/2018/nov/02/eu-border-lie-detection-system-criticised-as-pseudoscience

The EU has been accused of promoting pseudoscience after announcing plans
for a `smart lie-detection system' at its busiest borders in an attempt to
identify illegal migrants.

The lie detector, to be tried in Hungary, Greece and Latvia, involves the
use of a computer animation of a border guard, personalised to the
traveler's gender, ethnicity and language, asking questions via a webcam.

The deception-detection system will analyse the micro-expressions of those
seeking to enter EU territory to see if they are being truthful about their
personal background and intentions. Those arriving at the border will be
required to have uploaded pictures of their passport, visa and proof of
funds.


Credit Card Chips Have Failed to Halt Fraud, Survey Shows (Fortune)

Gabe Goldberg <gabe@gabegold.com>
Mon, 5 Nov 2018 17:33:38 -0500
New chip-enabled credit cards, which were rolled out to U.S. consumers
starting in 2015, were supposed to put an end to rampant credit card fraud.

So much for that.

A new report from the research firm Gemini Advisory has found that, of more
than 60 million cases of credit card theft in the last 12 months, a whopping
93% of the stolen cards had the new chip technology.

This represents a major setback for the technology, known as the EMV
standard, which is named after the companies (Europay, Mastercard and Visa)
that created it.

45.8 million records [were] likely compromised through card-sniffing and
point-of-sale (POS) breaches of businesses such as Saks, Lord & Taylor,
Jason's Deli, Cheddar's Scratch Kitchen, Forever 21, and Whole Foods. To
break it down even further, 90% or 41.6 million of those records were EMV
chip-enabled, states the report.

In theory, EMV should reduce fraud because every card transaction requires
an encrypted connection between the chip card and the merchant's
point-of-sale terminal. EMV is meant to replace conventional swipe
transactions that rely on magnetic strips, which contain data that is
relatively easy for criminals to intercept and then copy on to a new card.

But while the EMV standard is supposed to ensure the card data cannot be
captured, many merchants are failing to properly configure their systems,
according to a Gemini Advisory executive who spoke with Fortune. (Fortune
has also reached out to the payment processors for comment and will update
this article accordingly.) The upshot is that criminals have been able to
insert themselves into the transaction data steam, either by hacking into
merchant networks or installing skimmer devices in order to capture card
information.

The stolen data is typically sold on the so-called dark web, which is where
Gemini Advisory compiled the data for its report.

http://fortune.com/2018/11/05/credit-card-chips-fail-to-halt-fraud-survey-says/


Check this out: Radisson Hotel Group 'fesses up to `security incident' (The Register)

Monty Solomon <monty@roscom.com>
Fri, 2 Nov 2018 01:20:34 -0400
Loyalty card members deets exposed
https://www.theregister.co.uk/2018/10/31/radisson_hotel_group_fesses_up_to_security_incident/


A new study finds potentially manipulative ads in apps for preschoolers (WashPost)

Jose Maria Mateos <chema@rinzewind.org>
Fri, 02 Nov 2018 07:54:23 -0400
https://www.washingtonpost.com/amphtml/technology/2018/10/30/new-study-finds-potentially-manipulative-ads-apps-preschoolers/

Apps marketed to children 5 and younger deploy potentially manipulating
tactics to deliver ads to children, raising questions about the ethics of
child software design and consumer protection, according to a new study.

Researchers from the University of Michigan C.S. Mott Children's Hospital
looked at more than 100 apps, mostly from the Google Play app store, and
found that nearly all of them had at least one type of ad, often interwoven
into the apps' activities and games. The apps, according to the researchers,
used a variety of methods to deliver ads to children, including commercial
characters, pop-up ads, in-app purchases, and, in some cases, distracting
ads, hidden ads or ads that were posed as gameplay items.

The authors suggest that the deceptive and persuasive nature of the ads
leaves children susceptible to them, because of their lack of mental
development in controlling their impulses and attention.


Cisco Adaptive Security Appliance Software and Cisco Firepower Threat Defense Software Denial of Service Vulnerability (Cisco)

Monty Solomon <monty@roscom.com>
Fri, 2 Nov 2018 01:46:21 -0400
A vulnerability in the Session Initiation Protocol (SIP) inspection engine
of Cisco Adaptive Security Appliance (ASA) Software and Cisco Firepower
Threat Defense (FTD) Software could allow an unauthenticated, remote
attacker to cause an affected device to reload or trigger high CPU,
resulting in a denial of service (DoS) condition.

The vulnerability is due to improper handling of SIP traffic. An attacker
could exploit this vulnerability by sending SIP requests designed to
specifically trigger this issue at a high rate across an affected device.
Software updates that address this vulnerability are not yet
available. There are no workarounds that address this
vulnerability. Mitigation options that address this vulnerability are
available.

This advisory is available at the following link:
https://tools.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-20181031-asaftd-sip-dos

https://tools.cisco.com/security/center/downloadPDF.pdf


The D in Systemd stands for 'Dammmmit!' A nasty DHCPv6 packet can pwn a vulnerable Linux box. (The Register)

Monty Solomon <monty@roscom.com>
Fri, 2 Nov 2018 01:32:41 -0400
Hole opens up remote-code execution to miscreants “ or a crash,
if you're lucky

https://www.theregister.co.uk/2018/10/26/systemd_dhcpv6_rce/


T Wi-Fi kit bit by TI chip slip: Wireless gateways open to hijacking via BleedingBit chipset vulnerability (The Register)

Monty Solomon <monty@roscom.com>
Fri, 2 Nov 2018 01:35:50 -0400
Firmware security patches hit to fix critical holes in enterprise network
access points

https://www.theregister.co.uk/2018/11/01/it_bit_by_ti_chip_slipup_dubbed_bleedingbit/


ISP pissed at Elsevier Takedowns/blocks, so...

danny burstein <dannyb@panix.com>
Sat, 3 Nov 2018 00:37:51 -0400
Lots, make that LOTS, of slippery slopes here...

[Twitter]

Mike Masnick

Whoa. Elsevier forces an ISP to block some websites... so the ISP also
blocks Elsevier's websites, giving everyone who visits an explanation about
the evils of forced censorship...

<https://twitter.com/torrentfreak/status/1058427804637782016

and

[torrentfreak]

Swedish ISP Protests "Site Blocking" by Blocking Rightsholders Website Too

Ernesto on 2 Nov 2018

Bahnhof has suffered a major defeat against publisher Elsevier after a court
ordered the Swedish ISP to block a series of domain names, including
Sci-Hub.  The decision goes against everything the company stands for but it
can't ignore the blocking order. Instead, the ISP has gone on the offensive
by blocking Elsevier's own website and barring the court from visiting
Bahnhof.se.


Re: Ethics of whom to kill (Slade, RISKS-30.90)

Wols Lists <antlists@youngman.org.uk>
Fri, 2 Nov 2018 01:48:19 +0000
Has Rob Slade not heard of "The exception proves the rule"? Yes I know this
saying is horribly mis-used, but it almost certainly comes from the fact
that it only takes ONE inconvenient fact to destroy a scientific theory.

It is also an inconvenient fact that people dismiss inconvenient facts
as "oh that's just an anecdote". But it only takes one inconvenient
anecdote to be verifiable, at which point it becomes a data point
capable of destroying your theory and lifetime's work.

If there are a lot of anecdotes out there you cannot just dismiss and
ignore them. That's how the ozone hole was missed by computers ignoring
strange readings, until a scientist actually looked and thought "that's
not right!" You need to look at the anecdotes and explain them away,
otherwise they could well be inconvenient facts that mean you are
completely wrong.


Re: Explainable AI Simulation for AVs (Shapir Response in RISKS-30.90 to Stein, RISKS-30.89)

Richard Stein <rmstein@ieee.org>
Fri, 2 Nov 2018 12:28:28 +0800
In Risks-30.90, Amos Shapir wrote: >Driving is a team effort; it seem likely
that AVs will need to share >the roads with human drivers for quite a long
time, and would have to >be taught some social skills, before they can blend
in safely.

I agree with you. How to telescope carbon-based motorist intent to a robot?
Turn signals and brake lights are not always applied in a timely
fashion. Hand signals are probably a no-op for AV vision recognition and
interpretation. What about spilled coffee, DUI swerving, etc. per
https://catless.ncl.ac.uk/Risks/30/82%23subj23.1 therein, which might compel
a Trolley Problem scenario?

How to construct an "anxiety" algorithm component into an AV operational
control program? Anxiety—anticipatory fear—would play an important
role in silicon-based v. carbon-based vehicle interaction. When an AV
demonstrates safe/defensive driving techniques due to internal distraction
via a BlueTooth or WiFi hack attack, blown tire, collision, bird poo on the
sensors, skunk or chicken crossing the road, low fuel warning, LRU
malfunction, or smokey road conditions due to nearby fires etc., then I'll
believe AI has arrived.

If AV capabilities mature to show benefit via NHTSA statistics, feckless
parallel parking attempts by carbon-based drivers will make "AV Funniest
Videos" highlight reels.

In https://catless.ncl.ac.uk/Risks/30/56%23subj33.1
transition risk arising from AV introduction. Until an AV supreme transport
system materializes, adaptation to a "shared road" model constitutes a
paramount public health and safety objective.

The Pepsi Challenge on public health and safety benefits from AV deployment
has a heavy thumb on the scale tipped against it.


Re: Explainable AI Simulation for AVs (Amos Shapir)

Erling Kristiansen <erling.kristiansen@xs4all.nl>
Fri, 2 Nov 2018 20:41:20 +0100
And let's not forget that there are around 200 countries on our globe.
Traffic rules vary, sometimes significantly, sometimes very subtly, from one
country to another.  Some countries drive on the right, some on the left.
And driver `culture' differs quite a lot. And traffic signs and road
markings are different.

And how about non-standard signs? If a human sees a warning sign with a duck
or a cow, it is immediately obvious what it means, but what will an AV that
was not trained on such non-official signs do? And how about signs
containing text, that are obvious to a human, but likely make no sense to an
AV? And stuff that may resemble a sign, but is not.


Re: Toward Human-Understandable, Explainable AI (RISKS-30.88)

John Beattie <jkb@jkbsc.co.uk>
Fri, 2 Nov 2018 18:21:34 +0000
DJC writes:

  But as a matter of fact—honesty and integrity aside—humans aren't
  very good at knowing the grounds for their important decisions. Daniel
  Kahneman got the Nobel Prize for studying the reality of how people
  decide; cf. his book "Thinking, Fast And Slow".  He and his colleagues did
  many, many experiments to expose the *real* bases for how people make
  decisions; and those bases are often not only unknown to their subjects,
  but impossible for them to know, because they happen in inaccessible
  processes of their cognition.

This is true and not very relevant. An AI making a decision about, for example,
insurance or, say, an application at the local county hall needs to be able to
show the basis for the decision.  An arbitrary decision is not acceptable.


Re: Driverless cars: Who should die in a crash? (bbc.com)

John Beattie <jkb@jkbsc.co.uk>
Fri, 2 Nov 2018 18:26:18 +0000
There is a basic mismatch with reality about all those hypothetical cases
about who dies in a crash, speaking purely on engineering and commercial
grounds.

In practice, the AI will be challenged exactly as a car driver is: why did
you do that, why didn't you do the other. FWIW, the answer will be something
along the lines of, "The car was about to crash I didn't have time to make
fine decisions, I just hit the brakes and turned the steering as best I
could".

No AI in a car will have the extra resources to determine the locations and
motion of all or even some humans in the environment. It will have exactly
and only the resources to drive the car reasonably well in most of the
circumstances it is likely to meet.


Re: The spreading scourge of broken SSL implementation (RISKS-30.90)

Sergio Gelato <Sergio.Gelato@astro.su.se>
Sat, 3 Nov 2018 17:40:48 +0100
[note to moderator: feel free not to run this if other contributors have made
the same point.]

Mark Thorson complains about the growing number of HTTPS webservers that are
incompatible with Safari on his iBook G4, pointing out that some, like
www.google.com

The sites he describes as broken require TLS 1.2. The versions of Safari that
have been released for PowerPC Macs do not support this protocol. Given
https://tools.ietf.org/html/draft-moriarty-tls-oldversions-diediedie-00
the chances of reversing the trend look slim. Interposing proxy software that
performs protocol conversion (and HSTS enforcement, etc.) on the client
seems a better bet.

The RISK here, as I see it, is of making a poor tradeoff between security,
cost of maintenance and backwards compatibility.


Re: The spreading scourge of broken SSL implementation (Thorson, RISKS-30.90)

Julian Bradfield <jcb@inf.ed.ac.uk>
Mon, 5 Nov 2018 15:16:45 +0000
Mark Thorson complained that there is a recent spread of broken SSL
implementations on the Web, as he cannot access some sites from his iBook
G4.

He is partly correct, but not in the way he thinks. What he actually
experiences is that he is using a machine and OS that only supports the
obsolete and now deprecated TLS version 1.0 protocol - a protocol which is
now explicitly forbidden to be supported by any site taking credit card
payments. Therefore his browser is unable to establish a secure connection
to sites that no longer support insecure versions, although some sites such
as Google (and my own institution's academic site) still allow it.

So what is broken? Of the three sites he mentions, one,
https://www.ncahf.org/
behaves correctly and returns a protocol_version alert, so that a
decent browser (whether this includes an ancient Safari predating the
existence of more than one TLS protocol version, I don't know) will
display an appropriate error screen.

The other two sites https://marginalrevolution.com/
https://www.goldmine-elec-products.com/ both break the TLS protocol by
closing the connection without sending an alert message. marginalrevolution
also only supports weak ciphers.


Jury duty, recidivus

Rob Slade <rmslade@shaw.ca>
Sat, 3 Nov 2018 09:55:51 -0800
I'm off the hook for jury duty, so my presentation joke remains intact.

The jury or trial was canceled, almost literally at the last minute.
Selection was to start on Monday, and I got a call yesterday (Friday) late
in the afternoon.  (I almost didn't answer it, since it was on my cell,
which I vaguely recall them asking for when I registered my confirmation.
Almost nobody knows my cell, so I generally know who is calling, and I
didn't recognize the number.)  After the call I realized that the person who
claimed to be from the sheriff's office had given me almost no checkable
information.  (I did later find find an email notifying me of the
cancellation, so that was something.)

But it did put me in mind of a possible form of jury tampering.  Anyone can
call and claim to be from the sheriff's office.  (In Canada sheriffs handle
court security, and some other forms of court administration, such as jury
pool management.)  And, if there is no way for the juror to confirm, then it
would be easy enough to get rid of jurors you don't want.  Just have them
not show up.

Of course, this risk is slight.  To gain access to information about the
jury pool you would have to suborn a member of the sheriff's office and, if
you could do that, there would be a number of other ways to tamper with the
jury.

Just an idle security maven thought ...

  [The risk may seem slight.  However, jury tampering and juror conflicting
  are very old lawyerly arts.  PGN]

Please report problems with the web pages to the maintainer

x
Top