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 15

Tuesday 21 February 2017

Contents

German parents told to destroy Cayla dolls over hacking fears
BBC News
Self-driving cars? Only 2578 problems
Gabe Goldberg
The previous owners of used smart cars can still control them via the cars' apps
BoingBoing
"Safety, Security, and Privacy Threats Posed by Accelerating Trends
Helen Wright and Ben Zorn
"Why Humans Distrust Algorithms--and How That Can Change"
Cade Massey and Joseph Simmons
Serious Computer Glitches Can Be Caused By Cosmic Rays
Science
Re: Dutch election will be counted by hand
Erling Kristiansen
Re: Old Intel chips
Martin Ward
Re: Facebook Trending
Wols
Re: Google and Evil
Charles Cazabon
Re: WiReD
John Bechtel
Re: The AI Threat Isn't Skynet. It's the End of the Middle Class
Michael Marking
Info on RISKS (comp.risks)

German parents told to destroy Cayla dolls over hacking fears (BBC News)

Jim Reisert AD1C <jjreisert@alum.mit.edu>
Fri, 17 Feb 2017 11:49:51 -0700
http://www.bbc.com/news/world-europe-39002142

An official watchdog in Germany has told parents to destroy a talking doll
called Cayla because its smart technology can reveal personal data.  The
warning was issued by the Federal Network Agency (Bundesnetzagentur), which
oversees telecommunications.

Researchers say hackers can use an unsecure bluetooth device embedded in the
toy to listen and talk to the child playing with it.

But the UK Toy Retailers Association said Cayla "offers no special risk".
In a statement sent to the BBC, the TRA also said "there is no reason for
alarm".


Self-driving cars? Only 2578 problems

Gabe Goldberg <gabe@gabegold.com>
Thu, 16 Feb 2017 23:57:57 -0500
The 2,578 Problems With Self-Driving Cars

Last year, a self-driving car failed about every 3 hours in California,
according to figures filed with the state's Department of Motor Vehicles.

Every January, carmakers testing self-driving cars in California have to
detail how many times their vehicles malfunctioned during the preceding
year.  These so-called disengagement reports detail every time a human
safety driver had to quickly take control of their car, either due to
hardware or software failure or because the driver suspected a problem.

The reports—detailing 2,578 failures among the nine companies that
carried out road-testing in 2016—give a unique glimpse into how much
testing the different companies are doing, where they are doing it, and what
is going wrong.  None of this year's disengagements resulted in an accident.

http://spectrum.ieee.org/cars-that-think/transportation/self-driving/the-2578-problems-with-self-driving-cars

Alphabet's spin-out company Waymo still has by far the biggest testing
program—its 635,868 miles of testing accounted for over 95 percent of
all miles driven by self-driving cars in California in 2016.  Waymo's fleet
of 60 self-driving cars reported a total of 124 disengagements, 51 of them
due to software problems. That represents a sharp reduction in
disengagements from the previous year, from 0.8 disengagements for every
1,000 miles of autonomous driving down to just 0.2.
<https://medium.com/waymo/accelerating-the-pace-of-learning-36f6bc2ee1d5#.2fldeeluo>


The previous owners of used smart cars can still control them via the cars' apps (BoingBoing)

Lauren Weinstein <lauren@vortex.com>
Mon, 20 Feb 2017 09:40:02 -0800
NNSquad
http://boingboing.net/2017/02/20/the-previous-owners-of-used.html

It's not just that smart cars' Android apps are sloppily designed and thus
horribly insecure; they are also deliberately designed with extremely poor
security choices: even if you factory-reset a car after it is sold as used,
the original owner can still locate it, honk its horn, and unlock its doors.
Again, this is by design: because auto-makers are worried about lockout and
hacks (for example, a valet resetting your car to lock out your app), only
the original dealer can sever the car's connection with the cloud accounts
of the original owner.  Charles Henderson, the leader of IBM's X-Force Red
security division presented on this risk at last week's RSA conference in
San Francisco (you can read his essay on the subject here). His ultimate
recommendation is this counsel of despair: unless you are very
technologically savvy, you should only buy new cars, not used ones.  It's
not just cars, either—the problem extends to smart appliances,
thermostats, and other devices. Renting a house, staying in a hotel room, or
buying a house without replacing its appliances and HVAC systems also
exposes you to risks from the previous users of the devices in it.

    [Arthur T also noted this:
  How Previous Owners Can Potentially Still Access Their Cars
  Long After They've Sold Them
  http://jalopnik.com/how-previous-owners-can-potentially-still-access-their-1792534479
    ]


"Safety, Security, and Privacy Threats Posed by Accelerating Trends in the Internet of Things"

"ACM TechNews" <technews-editor@acm.org>
Fri, 17 Feb 2017 12:25:39 -0500 (EST)
Helen Wright, Ben Zorn, CCC Blog, 16 Feb 2017,
via ACM TechNews, 17 Feb 2017

The Internet of Things (IoT) is having a multi-trillion-dollar impact on a
range of industries, in addition to its significant societal impact on
energy efficiency, health, and productivity.  Although interconnected smart
devices have unforeseen potential benefits, the IoT also comes with
increased risk and potential for abuse.  One major challenge of having a
proliferation of IoT devices is the increased complexity that is required to
operate them safely and securely, which creates new safety, security,
privacy, and usability challenges.  A recent Computing Community Consortium
Computing in the Physical World Task Force report highlights some of the new
challenges created by the IoT, and argues issues related to security,
physical safety, privacy, and usability are interconnected.  However, the
report notes more research is needed to help manage the complexity, and to
connect usability concerns with safety, security, and privacy.
https://orange.hosting.lsoft.com/trk/click?ref=3Dznwrbbrs9_6-12ae0x211101x071860&


"Why Humans Distrust Algorithms--and How That Can Change"

"ACM TechNews" <technews-editor@acm.org>
Fri, 17 Feb 2017 12:25:39 -0500 (EST)
Knowledge@Wharton, 13 Feb 2017
via ACM TechNews, 17 Feb 2017

University of Pennsylvania professors Cade Massey and Joseph Simmons say
their research into humans' distrust of algorithms is rooted in people's
tendency to avoid following consistent, evidence-based rules in favor of
their instincts and intuition.  "People want algorithms to be perfect...even
though what we really want is for them to simply be a little better than the
humans," Simmons notes.  The professors say one way to get people more
comfortable with using algorithms is to give them a measure of control.  "We
can't get people to use algorithms 100 percent, but we can get them to use
algorithms 99%, and that massively improves their judgments," Simmons says.
Massey notes models and algorithms are held to a higher standard than
people, and he and Simmons say their next research focus will be on applying
their theories of humans' aversion to using algorithms to real-world
scenarios.
https://orange.hosting.lsoft.com/trk/click?ref=3Dznwrbbrs9_6-12ae0x211109x071860&


Serious Computer Glitches Can Be Caused By Cosmic Rays

Lauren Weinstein <lauren@vortex.com>
Sun, 19 Feb 2017 17:03:34 -0800
*Science* via NNSquad
https://science.slashdot.org/story/17/02/19/2330251/serious-computer-glitches-can-be-caused-by-cosmic-rays

  A "single-event upset" was also blamed for an electronic voting error in
  Schaerbeekm, Belgium, back in 2003.  A bit flip in the electronic voting
  machine added 4,096 extra votes to one candidate. The issue was noticed
  only because the machine gave the candidate more votes than were possible.
  "This is a really big problem, but it is mostly invisible to the public,"
  said Bharat Bhuva. Bhuva is a member of Vanderbilt University's Radiation
  Effects Research Group, established in 1987 to study the effects of
  radiation on electronic systems.


Re: Dutch election will be counted by hand (The Guardian)

Erling Kristiansen <erling.kristiansen@xs4all.nl>
Sat, 18 Feb 2017 10:40:02 +0100
Voting by paper ballots counted by hand has been
in force for a number of years in The Netherlands. What's new is
that the aggregation of votes into regional and national totals
will now also be done manually. This is not very clear in the
Guardian article.

This may delay the outcome a bit. But it seems to be accepted that a
reliable result is more important than an early one.


Re: Old Intel chips (RISKS-30.14)

Martin Ward <martin@gkc.org.uk>
Sat, 18 Feb 2017 10:02:19 +0000
>  [I am curious about chip flaws being more common than I thought.  Is anyone
>  is a position to knowledgeably comment about this?]

It seems that the designer of a safety-critical system should specify only
Intel chips that are *at least* four years old: so that there is some hope
that the most damaging bugs will have been shaken out!

A chip less than four years old is basically still in "alpha test".


Re: Facebook Trending (RISKS-30.11)

Wols Lists <antlists@youngman.org.uk>
Fri, 17 Feb 2017 22:09:59 +0000
And this will work how? When even the mainstream news agencies are in
the habit of (unintentionally) creating fake news?

I'm reminded of a Radio 4 program a loooong time ago. There was a
particularly juicy story in the news and a journalist decided to do some
digging to find out where it had come from. Something to do with "a crisis
in the Cabinet" with the Prime Minister (John Major, I think) having
difficulties.

A little bit of investigation quickly traced it back. The story originated
in an interview with a Labour Shadow Minister, being interviewed about
something else, and being pressed to provide information he didn't have, so
he speculated. Oh!!!

So the journalist dug into that story. To discover it, likewise, came from
an interview with another, non-Labour, politician being pressed to speculate
about another different story.

To cut a long story short, I think the journalist traced this back through
five or six different speculative stories, before he finally came to the
real story that started it all, that had absolutely nothing to do with the
story currently in the news.


Re: Google and Evil

Charles Cazabon <charlesc@pyropus.ca>
Fri, 17 Feb 2017 15:19:24 -0600
In the previous RISKS, Geoff Kuenning noted Google's creepy
photo-notification, but closes with:

> And IMHO it certainly violates their motto of "Don't be evil."

Google silently dropped the "Don't be evil" mantra many years ago.  Here's an
Article from 2009 noting they had dropped it:
http://www.siliconvalleywatcher.com/mt/archives/2009/04/google_quietly.php

Google has been quite evil for quite a long time.


Re: WiReD

John Bechtel <john@bechtel.me.uk>
Mon, 20 Feb 2017 12:21:50 +0000
Tiresome as it may be that some sites have ad-blocker-blockers, they have to
pay their bills.  Articles are what Wired sells—why should they give
content away?  If you don't like their requirements, don't view their
content.  I don't see asking me to view ads or pay for access as ransomware,
any more than I think its unfair that my supermarket charges me for
groceries.

In RISKS there's [sometimes] a note about a paywall—this is essentially
the same thing.


Re: The AI Threat Isn't Skynet. It's the End of the Middle Class (WiReD, RISKS-30.14)

Michael Marking <marking@tatanka.com>
Mon, 20 Feb 2017 18:21:30 +0000
  [via Dave Farber's IP]

The WiReD article [URL only, in RISKS-30.14] is misleading and incomplete,
and to a great extent incorrectly identifies the problem, for several
reasons.

(1) The problem isn't AI, or other forms of automation, it's the use to
which AI and automation are put and the basic mechanisms for allocating and
deploying resources in our society.  For example, if AI were to be used to
benefit the general population (healthcare, education, or other altruistic
use) without an implication or requirement that anyone need profit by it,
then one implication of the article—that the continued destruction of the
middle class will to a great extent be the fault of AI -“ would vanish.
Automation doesn't have to destroy jobs: we only say that, so we don't have
to make difficult choices. We pretend that there's no choice. We simply
don't want to deal with it. Yes, there are hardships, but as long as those
hardships are someone else's, we don't care.

As it stands, society surrenders to capital most decisions about resource
allocation, and capital naturally acts in its own best interests. That is
the problem.

(2) The assumption that `jobs' are the only way to allocate resources is
false. It derives from another false assumption, that the wealthy are
morally entitled to all profits, interest, and rents. People blindly believe
that the wealthy—who acquired their wealth from the labours of the
working class in the first place—are `trapped' due to economic
circumstances into unfortunate decisions that require job
destruction. Nowadays, in our culture, people believe in the positive
morality of interest (formerly, `usury') and profits and rents as surely as
they once believed in the divine, inherited right of kings to rule over the
lesser folk. These `rights' of the upper class, the owners, and the
privileged have no basis in science or honest reasoning.

There is plenty of work to be done, and there is plenty of wealth to feed,
clothe, and shelter everyone, but with all of our immense knowledge we can
think of only one way to address both issues: jobs offered by capitalists on
the condition that they enable the capitalists to maximize their own
accumulation of surplus value. Capitalists and landlords hold the common
welfare hostage, demanding ransom. Once you let go of that connection, once
you abandon that assumed precondition, many possibilities come to mind,
including ways to deal with workers displaced by automation.

The original economic sin was handing over the surplus to the few, and
society has been in Purgatory and in Hell ever since.

(3) The article wrongly implies that this is a new problem, that it started
to show up in the 1980s. In fact, automation and machinery deployed by the
capitalist system, which limits workers to market-minimized wages (including
`fringe benefits' as part of a compensation package) has been forcing
millions into poverty for centuries. Usually, the worst consequences are
safely out of view, in the hinterlands of the empire or in the slums of our
cities, where we aren't compelled to see them. AI is just another kind of
capital resource, along with machine tools, farmland, workers, and pig
iron. In their quests for profits, firms have been minimizing the costs of
production factors for a long time. Now, however, the threat is closer to
home.

Think of the sewing machine. A sewing machine can make a family more
independent and better off. On the other hand, sewing machines in sweatshops
can provide poverty wages for many, profits for a few, and destroy
handicraft industries in far-away places at the same time. The way we have
employed technology wasn't mandatory: society had choices.  AI and other
automation offer similar opportunities for good and for bad.

Where were these concerns when the lower classes, including those in the
colonies and neo-colonies, were being thrown out of work? Where are those
concerns now, as the poor and working classes sink lower, year after year?
We don't give a damn about the lot of the poor or the working class in the
Mideast, in Latin American, in Africa, or at home, as long as we middling
sorts get our cut of their resources and of the fruits of their labours. We
in the middle work for the rich, on commission: we have our incentives.

Meanwhile, we have been conditioned by our commercial culture into believing
that Bernie Sanders is a socialist, that the Soviet Union was communist, and
that we live in a democracy. (I hope that most of the readers of this list
know that all three beliefs are false.) Society, being willingly and
comfortably ignorant, is thus poorly equipped to discuss this problem, let
alone to `rise up' (to use McAfee's words, quoted in the article) and to
rectify things.

Again, the problem is not AI. Long ago we handed over to the upper class the
power to rule over us middle, working, and lower class subjects. However, as
Aristotle put it in his Politics, “But, although it may be difficult in
theory to know what is just and equal, the practical difficulty of inducing
those to forbear who can, if they like, encroach, is far greater, for the
weaker are always asking for equality and justice, but the stronger care for
none of these things.'' (Part 3, tr. Benjamin Jowett) We traded up, for the
larger size problem.

The problem is real. It is political, moral, social, economic, and ethical.
But it is not technological.

(BTW, the slides, by McAfee, linked from the article, are interesting, and
worth examining, but they are short on detail and I can't draw conclusions
from them alone. He doesn't explain how his plausible conclusion, that
automation of routine tasks, accounts for the phenomenon with any direct or
detailed data on specific job categories; he classifies, in the slides, jobs
merely by high, medium, and low income. Subtext: you have to buy his new
book. His conclusion requires a logical leap I can't bring myself to make. I
hope that he's not equating `medium' income jobs with `middle class' jobs,
that's not the way most researchers label those kinds of things; if that's
what he's doing, it's very misleading. Maybe the misnomer arises in the
WiReD reporting. Nor does he discuss possible additional social and
financial causes for the phenomenon, causes which have been considered
elsewhere. The first chapter of his book, co-authored with Erik
Brynjolfsson, is on-line; it doesn't look too promising with regard to hard
data: he deals more in summaries and conclusions than in raw data. None of
this means that AI and other automation aren't to some extent causing the
`spread' problem, it's merely that he doesn't make the case that the
phenomenon is that simple. I conclude that it's an oversimplification. Kind
of like a lot of TED talks, you can now go on home, being inspired and
informed, knowing that you've met your obligation to go to church this week,
but there's nothing in your own life that you, yourself, need to change, and
nothing that you can do.)

Please report problems with the web pages to the maintainer

Top