RISKS-LIST: Risks-Forum Digest Saturday 4 May 2019 Volume 31 : Issue 22
ACM FORUM ON RISKS TO THE PUBLIC IN COMPUTERS AND RELATED SYSTEMS (comp.risks)
Peter G. Neumann, moderator, chmn ACM Committee on Computers and Public Policy
***** See last item for further information, disclaimers, caveats, etc. *****
This issue is archived at <http://www.risks.org> as
<http://catless.ncl.ac.uk/Risks/31.22>
The current issue can also be found at
<http://www.csl.sri.com/users/risko/risks.txt&...
Contents:
World's Top Internet User Taps Fake News Busters for Elections
(Bloomberg)
Wells Fargo and Post Office Horizon (Lindsay Marshall)
Database Exposes Medical Info, PII Data of 137k People in U.S.
(Bleeping Computer)
Ladders Data Leak: Over 13M User Records Exposed Due To Cloud
Misconfiguration (IBTimes)
How angry pilots got the Navy to stop dismissing UFO sightings; UFO
information not expected to go to general public, Navy says (Wash Post)
This $1,650 pill will tell your doctors whether you've taken it.
Is it the future of medicine? (WashPost)
"Telecom giants battle bill which bans Internet service throttling for
firefighters in emergencies" (ZDNet)
UK Police Have a Message for Crime Victims- Hand Over Your Private Data
(NYTimes)
NSA Reports 75% Increase in Unmasking U.S. Identities... (WSJ)
New Documents Reveal DHS Asserting Broad, Unconstitutional Authority to
Search Travelers' Phones and Laptops (EFF)
Zero-day attackers deliver a double dose of ransomware -- no clicking
required? (Ars Technica)
Electronic Health Records and Doctor Burnout (Scientific American)
Hertz, Accenture, and the blame game (Browser London)
Monster screwup on dividends (Korea Herald)
NSA-inspired vulnerability found in Huawei laptops (Bruce Schneier)
Vodafone found hidden backdoors in Huawei equipment (Bloomberg)
Vodafone denies Huawei Italy security risk (BBC)
Re: Huawei's code is a steaming pile... (Keith Thompson, Dmitri Maziuk,
phil colbourn)
Re: Should AI be used to catch shoplifters? (Richard Stein)
Re: A video showed a parked Tesla Model S exploding in Shanghai
(Roger Bell-West)
Re: A 'Blockchain Bandit' Is Guessing Private Keys and Scoring Millions
(Dan Jacobson)
Re: An Interesting Juxtaposition (Gene Wirchenko)
Re: Gregory Travis' article on the 737 MAX (Gregory Travis)
Digital health ... (Rob Slade)
Re: Is curing patients, a sustainable business model? (Toby Douglass)
"Bernie Sanders wants you to expose your friends, Facebook-style" (ZDNet)
Abridged info on RISKS (comp.risks)
----------------------------------------------------------------------
Date: Sat, 4 May 2019 10:00:56 -1000
From: the keyboard of geoff goodfellow <geoff@iconia.com>
Subject: World's Top Internet User Taps Fake News Busters for Elections
(Bloomberg)
** Philippines' elections body cracks down on misleading posts*
** Media, academe team up to fact check election-related news*
EXCERPT:
In the Philippines -- where 76 million Internet users stay online the
longest in the world -- just a handful of people spend a few hours each day
to fight fake news about the upcoming midterm elections.
The Commission on Elections has formed a team of 10 government workers to
spot and report misleading online posts to Facebook Inc., with whom the
poll body has an agreement to quickly take down false information. Weeks
before the May 13 elections, the group has already identified hundreds of
fake news posts -- mostly those claiming ballots have been tampered with,
or that the poll results are predetermined.
``What we're trying to do is to institutionalize this reporting process in a
way that Facebook will not have any other recourse but to act on it,''
Election Commission spokesman James Jimenez said in an interview. ``Fake
news could affect how people see the credibility of the elections and the
mandate of the winner.''
Read more: What Happens When the Government Uses Facebook as a Weapon?
https://www.bloomberg.com/news/features/2017-...
acebook-into-a-weapon-with-a-little-help-from-facebook
With more voters using social media now, the election body expects fake news
to spread faster this time compared to the 2016 vote, when President Rodrigo
Duterte won. Still, Jimenez said the team formed to fight fake news is not
enough to adequately combat disinformation...
https://www.bloomberg.com/news/articles/2019-...
ps-fake-news-busters-for-elections
------------------------------
Date: Fri, 3 May 2019 14:13:55 +0000
From: Lindsay Marshall <Lindsay.Marshall@newcastle.ac.uk>
Subject: Wells Fargo and Post Office Horizon
I was recently asked by the BBC to comment on two `computer glitches', and,
naturally, I turned to RISKS to get more information. I found to my surprise
that neither seemed to have been mentioned. Here are links for the cases:
https://en.wikipedia.org/wiki/Horizon_(IT_sys...
https://edition.cnn.com/2018/12/12/business/w...
ndex.html
Note that neither of these seem to be even remotely ┤glitches'.
------------------------------
Date: Fri, 3 May 2019 21:25:56 -0400
From: Monty Solomon <monty@roscom.com>
Subject: Database Exposes Medical Info, PII Data of 137k People in U.S.
(Bleeping Computer)
https://www.bleepingcomputer.com/news/securit...
i-data-of-137k-people-in-us/
------------------------------
Date: Fri, 3 May 2019 21:27:20 -0400
From: Monty Solomon <monty@roscom.com>
Subject: Ladders Data Leak: Over 13M User Records Exposed Due To Cloud
Misconfiguration (IBTimes)
https://www.ibtimes.com/ladders-data-leak-ove...
ud-misconfiguration-2789394
------------------------------
Date: Thu, 2 May 2019 15:18:03 -1000
From: geoff goodfellow <geoff@iconia.com>
Subject: How angry pilots got the Navy to stop dismissing UFO sightings;
UFO information not expected to go to general public, Navy says (Wash Post)
http://www.washingtonpost.com/national-securi...
-navy-stop-dismissing-ufo-sightings/
[AND]
https://www.washingtonpost.com/world/national...
nformation-to-the-general-public-expected/2019/05/01/25ef6426-6b82-11e9-9d56-1c
0cf2c7ac04_story.html
https://www.sfgate.com/news/article/UFO-infor...
al-13810876.php
------------------------------
Date: Wed, 1 May 2019 14:26:17 -1000
From: geoff goodfellow <geoff@iconia.com>
Subject: This $1,650 pill will tell your doctors whether you've taken it.
Is it the future of medicine? (WashPost)
When the Food and Drug Administration approved in late 2017 a schizophrenia
pill that sends a signal to a patient's doctor when ingested, it was
seen not only as a major step forward for the disease but as a new frontier
of Internet-connected medicine.
Patients who have schizophrenia often stop taking their medicine, triggering
psychotic episodes that can have severe consequences. So the pill, a
16-year-old medication combined with a tiny microchip, would help doctors
intervene before a patient went dangerously off course.
Seventeen months later, few patients use the medication, known as Abilify
MyCite. Doctors and insurance companies say it is a case in which real-world
limitations, as well as costs, outweigh the innovations that the medical
industry can produce.
In the case of schizophrenia patients, some doctors warn that Abilify
MyCite could exacerbate the very delusions that the medication is designed
to prevent.
``Patients who have a lot of paranoia might be uncomfortable with the idea
of a medicine that is transmitting signals. The patient may be afraid to
take it,'' said Richmond psychiatrist James Levenson. ``The science of this
one is kind of ahead of the data.''
The debate over Abilify MyCite underscores a dilemma American health care
will increasingly face as the medical industry and Silicon Valley try to
promote innovation. For decades, medicine has been effectively delivered
through a few simple mechanisms: a pill, a cream, a nose spray, a needle.
But in the hopes of improving outcomes further, the industry is turning to
an array of new technologies against one of the biggest, and most human,
challenges in treating disease: getting people to take their medicine in a
consistent way.
Companies are producing apps for substance abuse treatment, diabetes
management, and heart and blood pressure monitoring at a rapid clip.
Studies are underway for more digital pills to treat cancer, cardiovascular
conditions and infectious disease.
And while many of these may pass regulatory hurdles that show they're safe
-- especially at a time when the Trump administration has been leaning into
medical innovation and pushing back against excessive regulation -- doctors
and insurers are not convinced that the technologies will so easily make
the difference that the pharmaceutical industry is betting billions on.
``I think that these technologies have a lot of potential benefits, but it's
going to be a question of evidence -- that they can demonstrate value to
patients and payers,'' said Scott Gottlieb, who stepped down this month as
FDA commissioner, a job in which he made approval of leading technology a
hallmark.
The first digital therapy to win FDA market clearance, Abilify MyCite's
sensor-embedded pill remains off the market because of physician and
insurance industry reservations.
Now Maryland-based Otsuka Pharmaceutical, which makes the medication, may be
able to jump-start its acceptance by offering it to mentally ill people who
qualify for low-income government health insurance. Otsuka won approval from
Virginia Medicaid authorities last month to begin coverage. The company also
is starting a pilot program in Florida and is considering another in
Oklahoma.
Otsuka considers itself a pioneer. Abilify is an older brand-name drug
marketed by the company to treat schizophrenia and other serious mental
illnesses. Abilify MyCite adds the electronic tracking component and, at
$1,650 a month, costs almost 30 times as much as a 30-day supply of generic
Abilify at a Costco pharmacy.
Otsuka developed the treatment with Proteus Digital Health, a Silicon
Valley company that markets the digital component. Proteus is pioneering
its use in other therapies including cancer patients taking chemotherapy
drugs.
After the daily antipsychotic pill is swallowed, a digital sensor the size
of a grain of sand (and made of copper, magnesium and silicon, which Proteus
says are all found in food) transmits a signal when it comes into contact
with stomach acid. The signal is captured by a patch worn on the patient's
torso. The patch sends a signal to an app on the patient's smartphone. The
app uploads data to a secure website for viewing by doctors. Otsuka has won
special federal approval to provide smartphones ``with highly limited
functionality'' to people who can't afford them.
The goal is to solve a vexing problem: Schizophrenia patients often stop
taking their medicine, triggering psychotic episodes that can have severe
consequences. Abilify MyCite is supposed to help doctors keep track of
which patients are staying on their medication. The app also allows
patients to enter information about their mood.
The approval led to debate among psychiatrists about the ethics of invasive
monitoring for patients whose mental competency at times may be borderline.
They raised questions about patients' autonomy, data privacy and ability to
navigate the technical challenges of the system.
But proponents say the medical need is so great that Abilify MyCite
deserves a close look.
Virginia state Sen. R. Creigh Deeds (D-Bath), who chairs a special mental
health committee in the legislature, said he had not heard of the therapy
until contacted by The Washington Post. But he said in an interview that he
was intrigued by a technology that could help people like his mentally ill
son, Austin `Gus' Deeds, 24, who slashed Deeds on the face in 2013 before
taking his own life. Deeds said his son had stopped taking medication nearly
a year beforehand. ``There is a need for people who are caregivers to make
sure the person's taking the medicine, The other side of it is the civil
liberty issue for the person who is sick.''
Gus Deeds thought his medications ``made him less of who he was. It dumbed
down his personality,'' Deeds said. But, he added, ``a person does not have
the right to destroy their life, or the life of others.''
He said he did not have an opinion on whether Virginia Medicaid should add
Abilify MyCite to its list of approved prescription drugs.
Otsuka emphasizes that no patient will be asked to use Abilify MyCite
without showing a clear desire to do so. Schizophrenia patients who have
paranoid feelings about ingesting a digital pill are unlikely candidates for
the drug, the company said.
``It's unlike a pharmaceutical launch where you proactively blitz all the
states. We're not doing that,'' said John Bardi, Otsuka's vice president
for public affairs and digital business development. ``It's really about
patients who want to improve their treatment goals. If they have any
concerns, it's probably not the right solution for them.'' ...
https://www.washingtonpost.com/business/econo...
octors-whether-youve-taken-it-is-it-the-future-of-medicine/2019/04/28/393281b2-
4c10-11e9-b79a-961983b7e0cd_story.html
------------------------------
Date: Wed, 01 May 2019 10:15:03 -0700
From: Gene Wirchenko <gene@shaw.ca>
Subject: "Telecom giants battle bill which bans Internet service throttling
for firefighters in emergencies" (ZDnet)
[What a PR blunder by the telecom industry!]
Charlie Osborne for Between the Lines | 26 Apr 2019
The industry faced backlash following last year's wildfires and
firefighter service throttling.
https://www.zdnet.com/article/telecom-firms-b...
ice-demand-for-firefighters-in-emergencies/
selected text:
Internet service providers (ISPs) and telecom firms are fighting a bill
which would force them to provide unfettered broadband services and prevent
them from throttling data use in emergency situations.
The proposed legislation is due to voted upon by California's Communications
and Conveyance Committee next week.
As reported by StateScoop, the bill -- introduced in February -- aims to
prevent a repeat of what happened in summer 2018 during the Mendocino
Complex Fire, one of the largest wildfires recorded in California's history.
As firefighters from the Santa Clara County Central Fire Protection District
fought to contain the fires, they found their Internet service drastically
reduced, having been throttled in what Verizon Wireless later called a
"customer support mistake."
Such connectivity can be crucial in emergency situations to coordinate
rescue and firefighting efforts. The fire department had an "unlimited" plan
with Verizon, but Ars Technica reports this service was throttled to speeds
of either 200kbps or 600kbps once 25GB -- the monthly cap -- was surpassed.
Verizon said at the time that the company has an internal policy to remove
"data speed restrictions when contacted in emergency situations," but this
did not happen during the wildfires.
To lift the throttling, instead, Verizon told the department to upgrade to a
more expensive plan.
------------------------------
Date: Wed, 1 May 2019 14:31:01 -1000
From: geoff goodfellow <geoff@iconia.com>
Subject: UK Police Have a Message for Crime Victims- Hand Over Your
Private Data (NYTimes)
The British police delivered a striking warning to crime victims on Monday:
If you want the case to be pursued, be prepared to turn over personal data
from your mobile phone, laptop, tablet or smart watches.
``Police have a duty to pursue all reasonable lines of enquiry,'' Assistant
Commissioner Nick Ephgrave, the National Police Chiefs' Council lead for
criminal justice, said in a statement. ``Those now frequently extend into
the devices of victims and witnesses as well as suspects -- particularly in
cases where suspects and victims know each other.''
But the new policy raised concerns about potential invasions of privacy and
the risk of discouraging people from reporting crimes, particularly
offenses like sexual assault that are already underreported because victims
fear being treated like the guilty ones.
In many cases, the police already search digital trails, which can produce
evidence that either backs up an accusation or casts doubt on it. Privacy
advocates say that police departments often improperly download cellphone
data from people they detain, without their knowledge or consent.
Under the new approach, victims and witnesses will routinely be asked to
sign a form saying that they consent to the police extracting data from
their electronic devices, which can mean text messages, emails, contacts,
social media records, Internet browsing history and more. Otherwise, the
case might not proceed...
https://www.nytimes.com/2019/04/29/world/euro...
l
------------------------------
Date: Wed, 1 May 2019 14:29:09 -1000
From: the keyboard of geoff goodfellow <geoff@iconia.com>
Subject: NSA Reports 75% Increase in Unmasking U.S. Identities... (WSJ)
*The National Security Agency, responsible for electronic eavesdropping,
disclosed the identities of people or entities that are normally redacted
in intelligence reports*
EXCERPT:
The National Security Agency revealed to federal agencies the identities of
almost 17,000 U.S. residents or corporations whose information was
collected under a foreign surveillance law in 2018, registering about a 75%
increase in unmaskings over the previous year, according to an annual
transparency report released Tuesday.
The NSA, responsible for electronic eavesdropping, disclosed the identities
of people or entities that are normally redacted in intelligence reports --
in response to specific requests from other government agencies to reveal
the identities, a process known as unmasking.
In 2018, NSA said it unmasked 16,721 U.S. identities caught up in
intelligence intercepts produced by a foreign intelligence law, the report
said. It unmasked 9,529 in 2017 and 9,217 in a 12-month period across 2015
and 2016.
The surge in the number of unmaskings last year was fueled in part by an
effort to determine the identities of victims of cyberattacks from foreign
intelligence agencies, according to Alex Joel, head of civil liberties and
transparency at the Office of the Director of National Intelligence which
released Tuesday's report.
Mr. Joel, in a call with reporters, said there were a number of varied
factors -- including world events and evolving threats--that could result in
statistical fluctuations in a given year for a certain type of surveillance.
Unmasking is a term used when the identity of a U.S. citizen, lawful
resident, or corporate entity is revealed in classified intelligence
reports. Unmasking is designed to be only used for national-security
reasons, such as helping officials assess intelligence by providing the
identity of someone two foreign spies may be discussing on a call. But the
process is governed by strict rules across the U.S. intelligence apparatus
that make it illegal to use unmaskings for political purposes or to leak
classified information...
[...]
https://www.wsj.com/articles/nsa-reports-75-i...
s-under-foreign-surveillance-law-in-2018-11556641509
https://www.washingtonpost.com/world/national...
ntities-likely-to-warn-victims-of-foreign-spying-new-report-suggests/2019/04/30
/35739e80-6b50-11e9-9d56-1c0cf2c7ac04 story.html
------------------------------
Date: Wed, 1 May 2019 14:32:01 -1000
From: geoff goodfellow <geoff@iconia.com>
Subject: New Documents Reveal DHS Asserting Broad, Unconstitutional
Authority to Search Travelers' Phones and Laptops (EFF)
*EFF, ACLU Move for Summary Judgment to Block Warrantless Searches of
Electronic Devices at Airports, U.S. Ports of Entry*
BOSTON--The Electronic Frontier Foundation (EFF) and the ACLU today asked a
federal court to rule without trial that the Department of Homeland Security
violates the First and Fourth Amendments by searching travelers' smartphones
and laptops at airports and other U.S. ports of entry without a warrant.
The request for summary judgment
https://www.eff.org/document/alasaad-motion-s... comes
after the groups obtained documents and deposition testimony revealing that
U.S. Customs and Border Protection and U.S. Immigration and Customs
Enforcement authorize border officials to search travelers' phones and
laptops for general law enforcement purposes, and consider requests from
other government agencies when deciding whether to conduct such warrantless
searches.
EFF Senior Staff Attorney Adam Schwartz: ``The evidence we have presented
the court shows that the scope of ICE and CBP border searches is
unconstitutionally broad. ICE and CBP policies and practices allow
unfettered, warrantless searches of travelers' digital devices, and empower
officers to dodge the Fourth Amendment when rifling through highly personal
information contained on laptops and phones.''
The previously undisclosed government information was obtained as part of a
lawsuit, Alasaad v. McAleenan
https://www.eff.org/cases/alasaad-v-duke
EFF, ACLU, and ACLU of Massachusetts filed in September 2017 on behalf of
11 travelers--10 U.S. citizens and one lawful permanent resident=94whose
smartphones and laptops were searched without warrants at U.S. ports of
entry.
Esha Bhandari, staff attorney with the ACLU's Speech, Privacy, and
Technology Project: ``This new evidence reveals that government agencies are
using the pretext of the border to make an end run around the First and
Fourth Amendments, The border is not a lawless place, ICE and CBP are not
exempt from the Constitution, and the information on our electronic devices
is not devoid of Fourth Amendment protections. We're asking the court to
stop these unlawful searches and require the government to get a warrant.''
The government documents and testimony, portions of which were publicly
filed in court today, reveal CBP and ICE are asserting broad and
unconstitutional authority to search and seize travelers' devices. The
evidence includes ICE and CBP policies and practices that authorize border
officers to conduct warrantless and suspicionless device searches for
purposes beyond the enforcement of immigration and customs laws. Officials
can search devices for general law enforcement purposes, such as enforcing
bankruptcy, environmental, and consumer protection laws, and for
intelligence gathering or to advance pre-existing investigations. Officers
also consider requests from other government agencies to search devices. In
addition, the agencies assert the authority to search electronic devices
when the subject of interest is someone other than the traveler -- such as
when the traveler is a journalist or scholar with foreign sources who are of
interest to the U.S. government, or even when the traveler is the business
partner of someone under investigation. Both agencies further allow officers
to retain information from travelers' electronic devices and share it with
other government entities, including state, local, and foreign law
enforcement agencies.
The plaintiffs are asking the court to rule that the government must have a
warrant based on probable cause before conducting searches of electronic
devices, which contain highly detailed personal information about people's
lives. The plaintiffs, which include a limousine driver, a military veteran,
journalists, students, an artist, a NASA engineer, and a business owner, are
also requesting the court to hold that the government must have probable
cause to confiscate a traveler's device.
The district court previously rejected the government's motion to dismiss the
lawsuit.
https://www.eff.org/deeplinks/2018/05/victory...
r
The number of electronic device searches at the border has increased
dramatically in the last few years. Last year, CBP conducted more than
33,000 border device searches, almost four times the number from just three
years prior. CBP and ICE policies allow border officers to manually search
anyone's smartphone with no suspicion at all, and to conduct a forensic
search with reasonable suspicion of wrongdoing. CBP also allows
suspicionless device searches for a `national security concern'.
[PGN-pruned for RISKS ...]
<https://www.cbp.gov/newsroom/national-media-r...
ectronic-device-searches-0>
For more information about this case:
https://www.eff.org/cases/alasaad-v-duke
https://www.eff.org/press/releases/new-docume...
onstitutional-authority-search-travelers
------------------------------
Date: Thu, 2 May 2019 15:16:07 -1000
From: geoff goodfellow <geoff@iconia.com>
Subject: Zero-day attackers deliver a double dose of ransomware -- no
clicking required? (Ars Technica)
*High-severity hole in Oracle WebLogic under active exploit for 9 days.
Patch now.*
EXCERPT:
Attackers have been actively exploiting a critical zero-day vulnerability
in the widely used Oracle WebLogic server to install ransomware, with no
clicking or other interaction necessary on the part of end users,
researchers from Cisco Talos said on Tuesday.
The vulnerability and working exploit code first became public two weeks
ago on the Chinese National Vulnerability Database, according to
researchers from the security educational group SANS ISC, who warned that
the vulnerability was under active attack. The vulnerability is easy to
exploit and gives attackers the ability to execute code of their choice on
cloud servers. Because of their power, bandwidth, and use in high-security
cloud environments, these servers are considered high-value targets. The
disclosure prompted Oracle to release an emergency patch on Friday.
On Tuesday, researchers with Cisco Talos said CVE-2019-2725, as the
vulnerability has been indexed, has been under active exploit since at least
April 21. Starting last Thursday -- a day before Oracle patched the zero-day
vulnerability, attackers started using the exploits in a campaign to install
`Sodinokibi', a new piece of ransomware. In addition to encrypting valuable
data on infected computers, the malicious program attempts to destroy shadow
copy backups to prevent targets from simply restoring the lost data. Oddly
enough, about eight hours after infection, the attackers exploited the same
vulnerability to install a different piece of ransomware known as GandCrab.
No interaction required...
https://arstechnica.com/information-technolog...
r-a-double-dose-of-ransomware-no-clicking-required/
------------------------------
Date: Fri, 3 May 2019 21:23:06 +0800
From: Richard Stein <rmstein@ieee.org>
Subject: Electronic Health Records and Doctor Burnout (Scientific American)
[Beware of Dr. Burnout. He is notoriously unready. PGN]
https://blogs.scientificamerican.com/observat...
-doctor-burnout/
The essay cites numerous factors contributing to physician burnout, the the
Agency for Healthcare Research and Quality (AHRQ) identifies: "family
responsibilities, time pressure, chaotic environment, low control of pace,
and the electronic health record."
A few cherry-picked items from the essay follow. Attributed to the EHR, the
author writes:
"In 2013 the Journal of Emergency Medicine reported that, over the course of
a 10-hour shift, resident physicians in a busy emergency room spent 28
percent of their work time with patients and 43 percent on data entry,
during which they made 4,000 keystrokes."
These input keystrokes trace to patient outcome/care/administration metrics:
"159 publicly available measures of outpatient care and that physicians
spent 2.6 hours and staff 12.5 hours per week attending to them. Insurers
and government massaged clinical and billing data with over 500 insurer and
1,700 government standards."
"No matter how good your intentions, if you just keep piling onto a harried
clinician's workday more stuff to do and more data to collect, you run the
risk of actually making care worse, angering patients and alienating
providers. Time pressure, chaotic environment, and low control of pace are
all exacerbated by overzealous oversight via the EHR."
The author suggests one technological fix to lighten clinicians' manual data
entry load: "To date, no maker of an electronic health record has figured
out how to do adequate justice to [patient] stories without sacrificing
data. Automated transcription of dictated notes is a start. Artificial
intelligence that can parse sentences and paragraphs into data should help a
lot."
Certain speech-to-text (STT) platforms advertise transcription success rates
at 99% for certain vocabularies and contexts, with medical specialties of
particular focus.
https://en.wikipedia.org/wiki/Speech_recognit...
"Error rates increase as the vocabulary size grows: e.g. the 10 digits
'zero' to 'nine' can be recognized essentially perfectly, but vocabulary
sizes of 200, 5000 or 100000 may have error rates of 3%, 7% or 45%
respectively."
Single word error rate and command success rate are two key metrics which
are influenced by numerous usage/capability attributes:
"Vocabulary size and confusability, speaker dependence versus independence,
isolated, discontinuous or continuous speech, task and language constraints,
read versus spontaneous speech, and adverse conditions."
https://www.nejm.org/doi/abs/10.1056/NEJMp091... on early voice
recognition/transcription. There are numerous commercial blogs that offer
automated voice transcription systems. See
https://blog.speech.com/2019/01/03/voice-reco...
record
for example.
Risks: Patient outcome benefit by replacing manual data entry with
speech-to-text (STT) transcription. Physician burnout reduction attributed
to STT deployment v. manual data entry.
Why not hire more physicians to unburden their clinical load? $, probably.
------------------------------
Date: Thu, 2 May 2019 23:52:26 -0400
From: Gabe Goldberg <gabe@gabegold.com>
Subject: Hertz, Accenture, and the blame game (Browser London)
The author says:
Either way, much of the reporting I've seen on this story has focused on the
sheer cost of the works and made many excellent points suggesting that the
business model of companies such as Accenture deliberately works to inflate
fees once the client is already heavily committed. Beyond $7 million for
the initial discovery work
https://www.browserlondon.com/services/resear... doesn't
say what the agreed contract fee was, but it does detail how -- once tied in
-- Hertz was continually billed by Accenture for fixes or new technology of
dubious value.
What stands out to me, however, is the other aspect of this situation. How
did the amount spent by Hertz balloon up to $32 million before a stop was
called to the work?
This highlights to me the fundamental issue many businesses seem to
encounter when embarking on large projects that are not within their own
core competency - namely their engagement with the day to day running of the
project. After all, it wasn't until Hertz executive asked about progress on
tablet views that the penny dropped that Accenture simply hadn't done many
of the things Hertz has asked of it.
I've read anecdotal evidence
https://news.ycombinator.com/item%3Fid%3D1974... on this project with
Accenture, Hertz, in fact, fired much of its internal digital and
developmental talent, handing over full control to Accenture. This, in my
opinion, is its first (if not biggest) mistake.
https://www.browserlondon.com/blog/2019/04/30...
------------------------------
Date: Tue, 30 Apr 2019 00:30:34 -0400
From: Gabe Goldberg <gabe@gabegold.com>
Subject: Monster screwup on dividends (Korea Herald)
But someone screwed up. Instead of issuing a KRW1,000 per share dividend, the
person in charge of hitting that button issued a 1,000 share per share
dividend. As the Korea Herald reported, dividends offered to employees due
to the `fat-finger' slip-up came to 112.6 trillion won (about $100
million), over 40,000 times the intended value and 33 times greater than the
company's market cap. Suffice it to say that, if the company couldn't
reverse the error, the company would cease to exist once these 200 or so
employees sold these phantom shares.
http://www.koreaherald.com/view.php%3Fud%3D20...
http://nowiknow.com/why-you-shouldnt-take-adv...
------------------------------
Date: Mon, 15 Apr 2019 06:51:56 +0000
From: Bruce Schneier <schneier@schneier.com>
Subject: NSA-inspired vulnerability found in Huawei laptops
CRYPTO-GRAM, April 15, 2019
This is an interesting story of a serious vulnerability in a Huawei driver
that Microsoft found. The vulnerability is similar in style to the NSA's
DOUBLEPULSAR that was leaked by the Shadow Brokers -- believed to be the
Russian government -- and it's obvious that this attack copied that
technique.
What is less clear is whether the vulnerability -- which has been fixed --
was put into the Huwei driver accidentally or on purpose.
https://arstechnica.com/gadgets/2019/03/how-m...
t-opened-systems-up-to-attack/
https://www.schneier.com/blog/archives/2019/0...
------------------------------
Date: Tue, 30 Apr 2019 15:24:55 -0700
From: "Peter G. Neumann" <neumann@CSL.SRI.COM>
Subject: Vodafone found hidden backdoors in Huawei equipment
For more than a decade, executives, intelligence agencies and conspiracy
theorists have been warning about the dangers of equipment from China's
Huawei Technologies Co.
And for almost as long, Huawei has denied that its telecommunications
products pose any kind of security threat.
The West has finally found its smoking gun. Yet it may not be enough
to sway those on either side of the debate.
As far back as 2009, Vodafone Group Plc -- one of the world's most powerful
and far-reaching telecom companies -- found hidden backdoors that could have
given Huawei access to its fixed-line network in Italy, Bloomberg News's
Daniele Lepido reported Tuesday, citing security briefing documents from the
London-based company.
https://www.bloomberg.com/opinion/articles/20...
-vodafone-are-a-smoking-gun
------------------------------
Date: Tue, 30 Apr 2019 11:53:53 -1000
From: geoff goodfellow <geoff@iconia.com>
Subject: Vodafone denies Huawei Italy security risk (BBC)
Vodafone has denied a report saying issues found in equipment supplied to it
by Huawei in Italy in 2011 and 2012 could have allowed unauthorised access
to its fixed-line network there.
A Bloomberg report said that Vodafone spotted security flaws in software
that could have given Huawei unauthorised access to Italian homes and
businesses.
The US refuses to use Huawei equipment for security reasons.
However, reports suggest the UK may let the firm help build its 5G network.
This is despite the US wanting the UK and its other allies in the "Five
Eyes" intelligence grouping -- Canada, Australia and New Zealand -- to
exclude the company.
Australia and New Zealand have already blocked telecoms companies from using
Huawei equipment in 5G networks, while Canada is reviewing its relationship
with the Chinese telecoms firm.
https://www.bloomberg.com/news/articles/2019-...
ors-in-huawei-equipment
------------------------------
Date: Mon, 29 Apr 2019 18:53:09 -0700
From: Keith Thompson <keithsthompson@gmail.com>
Subject: Re: Huawei's code is a steaming pile... (Shapir, RISKS-31.21)
Amos Shapir <amos083@gmail.com> writes:
> C does not force anyone to use strcpy() etc., it had always provided also
> similar length-limiting functions strncpy() etc.
strncpy() is not a "safer" version of strcpy(), as I've discussed here:
https://the-flat-trantor-society.blogspot.com...
trcpy.html
Even a length-limiting string copy function would not necessarily be
"safe". Consider a copying operation that silently truncates
"rm -rf /home/username/tmpdir"
to
"rm -rf /home/user/name"
------------------------------
Date: Tue, 30 Apr 2019 13:51:04 -0500
From: Dimitri Maziuk <dmaziuk@bmrb.wisc.edu>
Subject: Re: Huawei's code is a steaming pile ... (Ward, RISKS-31.21)
First, nobody's *forcing* anyone to juggle chainsaws.
Second, short answer is no, longer one is "define 'better'". Programming
language is a tool just like a hammer: you can make one that won't hurt your
thumb when you hit it. There will be a trade-off, though. Those trying to
drive in nails might even call that trade-off "undesirable".
(There is in fact a whole "c-minus" argument along the lines that modern
C has already gone too far in the "thumb safety" direction.)
Third, and on another tangent, the idea that computer programs are not aware
of the larger context seems to a recurring motif in RISKS lately.
The problem with "unsafe foo()-like functions" is whether the tool that
classified it "unsafe" based on the context in which the function is
invoked; if not, it may well be a false positive. Without knowing the
specificity and sensitivity of the "safety" test, assertion that "22% of
foo() invocations are unsafe" isn't really worth much, and if lack of
context awareness is a systemic problem, it likely isn't.
------------------------------
Date: Fri, 3 May 2019 14:01:17 +1000
From: phil colbourn <philcolbourn@gmail.com>
Subject: Re: Huawei's code is a steaming pile... (RISKS 31.16)
If Cisco is correct (see
https://blogs.cisco.com/news/huawei-and-cisco...
d
then Huawei's code may still be Cisco's code (or based on it).
Comparing Cisco STRCMP and Huawei's [CODE]: ``It must be concluded that
Huawei misappropriated this code.''
``Because of the many functional choices available to the Huawei developers
(including three of their own routines), the fact that they made the same
functional choice as Cisco would suggest access to the Cisco code even if
the routines had implementation differences. The exactness of the comments
and spacing not only indicate that Huawei has access to the Cisco code but
that the Cisco code was electronically copied and inserted into [Huawei's]
[CODE].''
``The nearly identical STRCMP routines are beyond coincidence. The Huawei
[CODE] routine was copied from the strcmp routine in Cisco strcmp.c file.''
Therefore, HCSEC [Huawei Cyber Security Evaluation Centre] should consider
reviewing code of other manufacturer's equipment used in UK critical
national infrastructure.
If Cisco is correct, then Huawei's code may still be Cisco.
https://blogs.cisco.com/news/huawei-and-cisco...
d
------------------------------
Date: Tue, 30 Apr 2019 18:54:34 +0800
From: Richard Stein <rmstein@ieee.org>
Subject: Re: Should AI be used to catch shoplifters? (cnn.com, R 31 20))
Busted! That is, I have been busted for expressing highly cynical and
condescending, even snarky, remarks about AI deployment as a crime deterrent
mechanism.
A software stack that can accurately and consistently detect larceny or
discriminate larcenous intent from a random customer pool, and then alert
authorities, would be astonishing.
https://edition.cnn.com/2019/04/18/business/a...
The article mentions:
1) The "VaakEye" algorithm was trained against 100K hours of
store-captured surveillance video;
2) A 77% reduction in shoplifting across 50 stores in Japan;
3) Global retail shoplifting losses accrued to $34 billion in 2017.
I will be convinced of VaakEye's product efficacy when/if statistics are
published that confirm accuracy and consistency of larcenous detection, and
show a sufficient reliability guarantee of false positive/negative findings.
Sufficient means 3+ nines, preferably 4+ nines, of accurate and consistent
theft detection.
Until then, a big warning sign should be posted at the shop entrance that
states something like:
"These premises deploy automated shoplifting surveillance technology to
deter stock theft. The surveillance captures and analyzes your shopping
habits, including hand/arm motion between the stock items and your clothes
and/or shopping cart/toke bag. Your facial profile is automatically
constructed and mapped to improve future theft detection capabilities. We
hope your shopping experience is pleasant. Come back again soon!"
------------------------------
Date: Tue, 30 Apr 2019 09:14:17 +0100
From: Roger Bell-West <roger@nospam.firedrake.org>
Subject: Re: A video showed a parked Tesla Model S exploding in Shanghai
(Stein, RISKS-31.21)
But the energy density of petrol (gasoline) is over ten times as much
(46.7MJ/kg), which is what makes it such a good fuel in the first place;
and yet, somehow, parked conventional cars rarely catch fire.
------------------------------
Date: Tue, 30 Apr 2019 19:27:25 +0800
From: Dan Jacobson <jidanni@jidanni.org>
Subject: Re: A 'Blockchain Bandit' Is Guessing Private Keys and Scoring
Millions (WiReD via Meacham)
>>>>> "BM" -- Bill Meacham <bmeacham98@yahoo.com> writes:
BM> ... the odds of guessing a randomly generated Ethereum private key is 1
BM> in
BM> 115 quattuorvigintillion. (Or, as a fraction: 1/2256.) That denominator
BM> is
BM> very roughly around the number of atoms in the universe. ... But as he
I just see "1/2256" above. One in two thousand.
------------------------------
Date: Tue, 30 Apr 2019 18:57:24 -0700
From: Gene Wirchenko <gene@shaw.ca>
Subject: Re: An Interesting Juxtaposition (Wol, RISKS-31.21)
"I think Gene should be blaming the expensive GPS's, not the cheap ones!
Many of my colleagues use Google Maps or Waze because they're so much
better."
How about I blame them all?
Google Maps has some, ah, interesting quirks.
------------------------------
Date: Sat, 4 May 2019 00:23:39 -0400
From: Gregory Travis <greg@littlebear.com>
Subject: Re: Gregory Travis' article on the 737 MAX
First, I am delighted to once again be a part of the RISKS community. Some
may remember postings I made in the (very) early 1990s here, including a
(humorous) sendup of the A320.
Second, the point of my article was to convey to the lay public:
1. Unlike previous 737 models, Boeing's 737 MAX 8 airframe could (and does)
not meet the pitch stability and control force requirements of FAR part 25.
2. Boeing realized this fairly early in the development process with wind
tunnel and computer simulations.
3. Boeing determined that a fairly simple bit of software would make the
problem "go away." Namely programming that took AOA input from a single
(AOA) sensor and used that input to determine whether or not to drive the
horizontal stabilizer trim.
4. Later, during actual flight tests, it was determined that the pitch
instability and control force problems of the airframe were far more
serious than the early wind tunnel and simulations indicated (this is
somewhat common in the industry).
5. Conversely, the software was changed to MUCH more aggressively trim the
horizontal stabilizer. In fact, it could drive the stabilizer to its
mechanical stops in roughly 20-30 seconds.
And:
1. There is an inherent and deep engineering problem in any system that
relies on a single sensor as input without any data validation,
particular a system that can use that data to drive very large flight
surfaces to their mechanical stops in seconds (I am sure some pedant will
complain that the electric motor running the jackscrew has a different
set of stops than the mechanical trim wheel. I am tired of responding to
such irrelevant nonsense).
2. What is often not mentioned is that Boeing explicitly changed the trim
disconnect function for this system. It will not stop if the pilot exerts
countering control force. This is a nonintuitive behavior for any pilot who
are used to autopilots and electric trim automatic disconnects if the pilots
exert a control force contrary to the direction of trim.
3. Aerodynamic loads on the horizontal stabilizer can exceed a human's
ability to move the stabilizer trim manually. Boeing has known this for
nearly thirty years, yet they suggested a fix to the problem was to
disconnect the electric trim (use the cutoff switches) and manually trim.
As the Ethiopian Air pilots found out, that is impossible. Boeing knew
this.
And:
1. Boeing intentionally hid the existence of this system (so that pilot
training would not be required) not only from the line pilots flying
revenue, but from its own test pilots.
2. For example, the Master Minimum Equipment List (MMEL) for the 737 MAX
makes no mention of the system. Although there are cockpit failure
indications for the yaw damper, the speed trim system, the mach trim
system, etc. there is no failure indication for MCAS.
3. Angle of attack sensor failure is common, contrary to assertions
otherwise. The service difficulty database has about 200 entries and that
typically represents 5% of the real-world situation., Frozen water
(heater failure) in the system is a very common failure cause.
4. The 737 MAX MMEL allows the 737 MAX to take off with all angle of attack
sensor heaters inoperative.even though Boeing knew that a single angle of
attack sensor failure could render the aircraft uncontrollable with this
system.
5. In contrast, the MMEL for the A320 requires that at least two of the
three angle of attack sensor heaters be operational before flight.
And:
1. All of this can be traced back to a change in Boeing's corporate culture
that began with the McDonnell Douglas takeover of Boeing in 1997 (where they
used Boeing's own money).
|
BBS Software Chatter 
