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 27 Issue 08

Sunday 11 November 2012


Elections and Hurricanes: After the Aftermath of the Math
Summary of my experiences on the election
Douglas W Jones
My election day reports
Jeremy Epstein
Virginia city's ballot listing Obama as republican, Romney as democrat
Jeremy Epstein
Unusual risk for US voting machines: a spider
Valdis Kletnieks
Covington anomaly: mistaken attribution
Another misguided call for online voting
Lauren Weinstein
"Estonia gets to vote online. Why can't America?"
Lauren Weinstein
Security Researchers Warn New Jersey's Emergency E-mail Voting Could Be An Insecure, Illegal Nightmare
Matt Blaze via LW
Another article on evoting
Ezra Klein via LW
Government Services in Clouds
Chris Drewe
BGP error in Indonesia blocks Google in other areas
Lauren Weinstein
Did Skype Give a Private Company Data on Teen WikiLeaks Supporter Without a Warrant?
Ryan Gallagher via Monty Solomon
Creative Disruption: Sandy Tells Us, *Let's Start Over*
John F. McMullen
Sandy: NYU hospital power outage... may have been from safety sensors
Danny Burstein
Re: Verizon FIOS phone service
Bill Hopkins
Re: When your fuel pumps are below sea level...
Simson Garfinkel
Info on RISKS (comp.risks)

Elections and Hurricanes: After the Aftermath of the Math

"Peter G. Neumann" <>
Sun, 11 Nov 2012 11:19:08 PST
One of the main goals for the conduct of elections should be to provide
sufficient assurance throughout the entire process such that every loser can
justifiably believe that he or she actually was not the winner—that is,
that there were no events, circumstances, or externalities, accountable or
otherwise, that might have altered the results.

From the perspective of RISKS and our long-standing discussions of factors
relating to election integrity, one of the most interesting aspects in last
week's election was that the results of the Presidential race were
definitive enough that they did not depend on the outcomes in larger states
such as Ohio and Florida (where the results were apparently not known
officially until yesterday).  If the results for Obama vs Romney had been
very close, I suspect that we would have seen prolonged law suits from both
parties leading to the Supreme Court—irrespective of the perceived
initial outcome.  Overall for the election for the President and other
offices and ballot issues as well, numerous issues arose during the
campaigning and the voting process—for example, relating to voter
registration, voter disenfranchisement, voter authentication, restrictions
on early voting, shortages of voting machines and trained election officials
that resulted in huge lines in certain precincts, unsanctioned and
unsupervised last-minute changes to proprietary election software, reported
cases of vote flipping on touch screens, inconsistent party affiliations
with unclear implications for straight-party voting, irregularities in
issuing, validating, and counting provisional ballots, cases in which more
votes were reported counted than ballots issued, disappearing ballots,
inconsistencies in announcements of policies, deceptive practices, poorly
defined policies for reviewing and recounting close races, last-minute
attempts to create opportunities for Internet and e-mail voting in response
to the disruptions of Hurricane Sandy (typically without adequate
appreciation for the wide range of potential problems with which RISKS
readers are familiar), along with many other factors such as the perception
of even less visibility, accountability, and oversight for other than top
races.  As I began to note in RISKS-27.06, much greater accountability,
contingency planning, and objective oversight are needed—along with
considerably greater even-handedness—to ensure that future elections will
be able to avoid these problems and others.

  [Some examples of these problems are included in subsequent reportage in
  this issue.  PGN]

Summary of my experiences on the election

"Jones, Douglas W" <>
Thu, 8 Nov 2012 23:10:23 +0000
I spent election day 2012 monitoring incident reports from polling places
around the country.  In doing this, I observed a number of patterns that
seem worthy of note:

In Florida, the root cause of many of the problem reports lies squarely in
the lap of the state legislature.  It was the legislature that created the
laws that led to ballots that were 10 pages long.  There were several
distinct issues that combined to force such long ballots—the sheer number
of constitutional amendments and referenda on the ballot, the decision to
print bilingual and trilingual ballots instead of limiting the number of
languages on each piece of paper, and the printing of the entire text of
each measure on the ballot.  The net result was that some ballot pages
contained just one yes-no choice, and on the Miami sample ballot I tried to
read, page 7 was solid text with no choices at all.

These long ballots slowed down the voting process.  In many precincts,
I saw reports of inadequate numbers of voting booths.  No change to voting
technology could cure this.  Gigantic ballots simply take a long time to
read, regardless of the technology.

These long ballots choked the tabulating machines.  Today's precinct-count
tabulators are easily able to handle a few thousand sheets of paper on
election day, but each page scanned is likely to deposit a few paper fibers
on the scan head, and by the end of an election, the machine needs the
dust blown out of the paper path to prevent misreads and paper jams.  Give
each voter a 10-page ballot instead of the usual one or two page ballot,
and the machine really will need preventative maintenance several times
during election day.  It is no surprise, therefore, that Florida suffered
many scanner failures.

All ballot boxes for precinct-count scanners that I've examined contain
multiple compartments, one of which is an "emergency compartment" for use
when the scanner fails.  In Florida, when scanners failed, the sheer volume
of paper was enough to fill these emergency compartments to capacity,
forcing pollworkers to improvise.

And finally, most precinct-count scanners move the paper fairly slowly, at
speeds comparable to the paper-feed speed of a typical FAX machine.  When
feeding one-page ballots, this does not cause significant problems, but hand
feeding successive pages of a 10-page ballot can take a substantial fraction
of a minute, especially if the scanner is programmed to warn voters of any
blank pages in the ballot in order to protect against inadvertent
undervotes.  In polling places with enough voting booths, the speed at which
ballots could be fed into the machine was reported to be a bottleneck.

In my opinion, Florida's legislature can make several changes to address
these problems: They ought to require that each ballot measure have a long
form and a short form, with only the short form printed on the ballot.  The
short form should be required to be composed by those proposing the ballot
measure, so that all debate about the measure can be informed by both the
full text and the short text from the start.  They should also consider
capping the number of ballot measures in any election.

A second cause for long lines was apparent in Virginia, where I saw numerous
reports of equipment failures.  When the polls opened, significant numbers
of polling places had problems getting things working.  In some cases,
polling places were unable to open on time, and in other cases, polling
places open= ed with only a few functional voting machines.

I can only speculate about the cause of the machine failures, but I note
that in many cases, the machines involved appear to have been purchased
close to a decade ago and appear to have been built using laptop computer
technology.  Election officials are used to voting machines that last
decades.  Mechanical voting machines certainly lasted that long, and many of
the first generation of precinct-count scanners and direct-recording
electronic voting machines have proven to be almost as durable (I have seen
numerous documented cases of lifetimes over 20 years).  Unfortunately, the
technology we use in laptop computers is not generally that durable.  Liquid
crystal displays and touch screens appear to have a useful lifetime measured
in years, not decades.

A second problem has to do with polling place procedures in the event of
failure.  I saw too many reports of voters being turned away or made to wait
for hours until voting equipment could be repaired.  In many states, it is
illegal to turn away a legal voter on election day merely because the voting
machines are broken.  Voters must be allowed to vote, and the mere fact that
the machines are broken is no excuse.

The typical procedure for meeting this requirement requires the pollworkers
to issue emergency paper ballots in the event that the machines fail.  Any
paper can be used, but the supplies packet for a polling place should
include standard generic ballot forms, along with instructions directing the
pollworkers to give a blank emergency ballot and a sample ballot to each
voter, instructing them to write their choice for each ballot question on
the emergency form.  At the end of the day, these emergency ballots must be
hand counted.

Yet a third cause of long lines was directly attributable to the Help
America Vote Act of 2002.  This act required the use of state-wide voter
registration databases.  Putting these databases in place was not trivial,
and this election was, to a significant extent, the first full-scale test of
the new system.  In polling places across the country, on-line tools were
used for voter check-in.  In some cases, these tools included scanners to
read ID information off of drivers' licenses, greatly speeding the check-in
process.  Problem reports with the new voter registration systems fell into
several categories:

In some cases, the electronic pollbook mechanisms simply failed.  As with
voting machines, such failures do not justify closing polling places, and
there must be a way to allow voters to vote when this occurs.  An obvious
fallback measure is to equip each polling place with a paper list of all
voters registered in that polling place and train the workers how to use
that register in the event that their machines (or communication lines)

In other cases, there were simply not enough electronic pollbooks.  Many
election officials appear to have underestimated the time it takes to look
up a voter, possibly because they misunderstood the trial and error nature
of looking up a person in as database.  Am I Doug Jones, Douglas Jones,
Douglas W. Jones or Douglas Warren Jones?  Do I live at 816 Park, 816 Park
Rd. or 816 W. Park Rd.?  There are 16 permutations of the above, and if you
use any of those permutations on a letter, it will arrive in my mailbox
with no problem.  Unfortunately, many statewide voter databases do not
have search tools that are as intelligent as my postman, and diligent
attempts at voter identification by pollworkers can bog down.  It does not
help that the problem can only be solved using significant local context.
E. Park Rd. exists, but has no postal addresses.  Park Pl. exists, but has
no addresses in the 800 block.  There are many people named Douglas Jones
in town, and even multiple people named Douglas W. Jones, but only one at
816 Park.

And in yet other cases, the combination of database and pollworkers could
not correctly match registered voters with their database entry, resulting
in consequences identical to striking a legitimate voter from the voter
rolls, an illegal act.  I cannot tell from the incident reports I saw
whether the voters who were effected were struck because the pollworkers
were insufficiently diligent dealing with alternative name and address
spellings, or whether the fault was in poorly constructed database search

Alert readers will already have seen numerous media reports of vote
flipping, and I certainly saw many reports of that on election day.  Some of
them were actually misinterpretations of something else—what election
insiders refer to as fleeing voters.  With many electronic voting machines,
the resulting incident report runs roughly as follows: "When I into the
voting booth, a bunch of candidates were pre-selected on the face of the
machine."  This report is then misinterpreted as evidence of some kind of
machine rigging when it is really the result of the previous voter fleeing
the machine without taking the final step of casting their ballot.

There were also reports that were most likely caused by miscalibrated touch
screens, leading to reports such as "when I tried to vote for Obama, Romney
lit up" (or visa versa).  Touch screen calibration is an annoying necessity
on resistive elastomeric membrane touch screens, and it is easy, but
pollworkers don't always know how to do it.

In just one case, however, I saw a report (in the blogosphere) of a touch
screen voting system where the evidence suggests something far more
sinister.  A voter who was aware of the calibration issue actually went into
the machine and noticed the problem, and then set about a careful program of
careful diagnosis, repeatedly selecting and deselecting candidates in order
to measure the dimensions of the sensitive area of the screen for each
candidate.  If that report is to be trusted, the machine in question should
be impounded for forensic analysis, because the conclusion was that the
border of the sensitive area between Obama and Romney had been moved,
shrinking one while enlarging the other without changing the dimensions of
the sensitive areas for other candidates.  That could be evidence of genuine
fraud, and it does not fit the symptoms I associate with miscalibration.

One category of incident reports was actually comforting.  These reports
typcally reported a real failure of some kind, for example a broken machine,
and then went on to report, with alarm, that the pollworkers had instituted
some kind of ad-hoc procedure to deal with the failure.  The encouraging
thing I saw was that these "ad-hoc procedures" were almost always, in fact,
the solutions that were required by the local rules.  Many pollworkers did
correctly open the emergency ballot compartments on scanners when those
scanners broke.  They did scan those ballots later, when the scanners were
repaired, and they did issue emergency paper ballots when they ran out of
official ballots or when the electronic voting machines broke.  In short,
when done competently, pollworker training does work.

The biggest problem with spending election day monitoring incident reports
is that all I saw, all day, was evidence of things going wrong.  As a
result, when I finally got a chance to see the media reporting after the
election was largely decided, I was surprised to see people saying that the
day went surprisingly smoothly with only occasional reports of trouble.

A final comment: Sadly, these incident reports are not fed back into the
system.  The Democratic and Republican parties each manage their own
incident reporting databases, but as far as I know, those databases are
routinely destroyed after each election.  The Election Protection folks at
866-Our-Vote maintain a public database, but it is largely ignored by
officialdom.  Sadly, when I have been in a position to look at multiple
incident reporting systems, I have rarely noticed the same incident being
reported more than once.  This makes me suspect that the three databases
I've mentioned above contain very little overlap.  It would be wonderful if
they could all be published, merged and subject to a careful analysis, but I
have no idea how to make this happen.

  [The usual disclaimer: All of the opinions I expressed above are my own
  and do not necessarily reflect the opinion of any agency or organization,
  be it public or private.  I wish, of course, that they would slavishly
  follow my lead except when I am wrong.]

My election day reports

Jeremy Epstein <jeremy.j.epstein@GMAIL.COM>
Fri, 9 Nov 2012 09:52:27 -0500
I wrote two reports about what I saw from a command center on election day.

  [I eschew summarizing, and urge you to read Jeremy's experiences. PGN]

Virginia city's ballot listing Obama as republican, Romney as democrat

Jeremy Epstein <jeremy.j.epstein@GMAIL.COM>
Tue, 6 Nov 2012 14:39:10 -0500
In one Virginia locality, the electronic voting machines were programmed to
show Obama as Republican, Romney as Democrat.  The machines were removed
from service and they're now using paper.  But this part made me really
nervous: "All votes that were cast Tuesday morning will be counted
properly."  What do they mean by "properly" - was a vote for Obama a vote
for the Democratic electors or the Republican electors?  Since some people
vote by name and others by party, you can't tell what voters intended.  I
have no idea how I'd count those votes if I were the judge!

This is really important if Virginia is a close race.  It's not a technical
problem - it's a ballot setup problem.,0,5763810.story

Unusual risk for US voting machines: a spider

Valdis Kletnieks <>
Thu, 08 Nov 2012 20:40:16 -0500
It wasn't voter fraud that delayed the election count in one U.S. town - it
was a spider.  Rehoboth Town Clerk Kathleen Conti says one of the
Massachusetts town's aging voting machines malfunctioned Tuesday.  Ms. Conti
tells *The Sun Chronicle* of Attleboro that she called a technician, who
said a spider web apparently prevented the machine's scanner from counting
ballots. The vote count wasn't completed until Wednesday afternoon.
Rehoboth voters favoured Republican presidential challenger Mitt Romney, who
lives in Massachusetts.  Ms. Conti says she has been pressing to have the
voting machines replaced for several years.  [AP item]

  [Rob Slade commented on this item as well:
    Yet *another* reason to distrust voting machines: Arachnophobia

Covington anomaly: mistaken attribution

"Peter G. Neumann" <>
Tue, 6 Nov 2012 11:06:52 PST
Interesting conundrum. The political affiliation of the Obama-Biden ticket
on voting machines in Covington, VA is listed as Republican on the voting
machines. So question: does a vote for Obama-Biden transfer to a vote for an
elector chosen by the Republicans or the Democrats?

“City of Covington moves to paper ballots after voting machine issue
Mistake made while voting machines were set up COVINGTON, Va.''  We're
learning more about voting errors in the City of Covington.

All voters in Covington will use have to use paper ballots.

There was an error Tuesday morning with the voting machines. If you voted
for President Obama, the machine would list the Obama-Biden ticket as
Republican. All votes that were cast Tuesday morning will be counted

However, election officials decided to switch the City of Covington to paper
ballots to avoid confusion.",0,5763810.story

Another misguided call for online voting

Lauren Weinstein <>
Sun, 11 Nov 2012 08:27:15 -0800  (*The New York Times* via NNSquad)

  "So at a time when we can see video shot by a robot on Mars, when there
  are cars that can drive themselves, and when we can deposit checks on our
  smartphones without going to a bank, why do most people still have to go
  to a polling place to vote?"

I understand why people would love to vote online.  But when [almost] every
recognized expert in the field tells you it would be a disaster, and
fundamentals of computer security agree with them, you have to make a
choice.  Go hi-tech with voting and turn the elections over to hackers,
coercion, and worse, or admit that there are still a few things in life that
are better done the old-fashioned way—if we care about democracy, that

"Estonia gets to vote online. Why can't America?"

Lauren Weinstein <>
Thu, 8 Nov 2012 16:51:34 -0800  (*The Washington Post* via NNSquad)

  "What's more, Estonia has a proportional representation voting system,
  rather than a winner-take-all system like the United States. According to
  Hall, research has found that electoral fraud seems to pop up more
  frequently in winner-take-all systems - since there's more at stake for
  the candidates."

 - - -

Online Voting: Just Say No!

Security Researchers Warn New Jersey's Emergency E-mail Voting Could Be An Insecure, Illegal Nightmare

Lauren Weinstein <>
Mon, 5 Nov 2012 16:24:34 -0800  (*Forbes* via NNSquad)

  It took less than 24 hours for Matt Blaze, a computer science professor at
  the University of Pennsylvania who audited voting systems for California
  and Ohio in 2007, to start pointing out the problems with that workaround:
  Unencrypted e-mail can be spoofed or tampered with. The computers used to
  send the e-mail, many of which will be in public places like libraries or
  shelters, could be compromised to change or block voters' choices. And the
  computer that receives the e-mail may be just as vulnerable to
  sabotage-given that voters will be sending their ballots as attached
  files, the receiving PC will need to open attachments sent by unknown
  users, one of the most common practices leading to malware infections.

Another article on evoting

Lauren Weinstein <>
Thu, 8 Nov 2012 16:41:06 -0800

Government Services in Clouds

"Chris Drewe" <>
Tue, 06 Nov 2012 21:16:16 +0000
The UK 'Daily Telegraph' newspaper has a comment article today (6 Nov 2012)
about Government proposals to make all of its services "digital by default",
partly for easier accessibility, and partly for reduced operating costs; all
fine and dandy, but as the article says, there appear to be at least three
RISK areas here (please be aware that in the UK, almost everyone has some
dealings with the welfare or tax authorities):

* The people most in need of welfare, mainly senior citizens, are least
  likely to be Internet users (as probably remarked before in RISKS).

* All of this data needs to be stored yet readily accessible to authorised
  users in a secure way of course, so lots of RISKS there, plus the
  proposals include "cloud computing"—this is our personal details...

* It's a hotly-contested field, but the UK allegedly has the most Byzantine
  taxation and welfare systems in the world, and the Government is planning
  to start this "digital by default" programme with the Universal Credit
  scheme, a major change to welfare provision, due to start in April 2013,
  so that's two big changes at once.

  [Article by Philip Johnston, 5 Nov 2012, omitted.  PGN]

BGP error in Indonesia blocks Google in other areas

Lauren Weinstein <>
Tue, 6 Nov 2012 10:56:53 -0800  (*CloudFlare* via NNSquad)

  “The case today was similar. Someone at Moratel likely `fat-fingered' an
  Internet route. PCCW, who was Moratel's upstream provider, trusted the
  routes Moratel was sending to them. And, quickly, the bad routes
  spread. It is unlikely this was malicious, but rather a misconfiguration
  or an error evidencing some of the failings in the BGP Trust model."

Did Skype Give a Private Company Data on Teen WikiLeaks Supporter Without a Warrant? (Ryan Gallagher)

Monty Solomon <>
Sun, 11 Nov 2012 16:10:30 -0500
Ryan Gallagher, 9 Nov 2012

Skype's privacy credentials took a hit in July over a refusal to comment on
whether it could eavesdrop on conversations. Now the Internet chat service
is facing another privacy-related backlash-after allegedly handing over user
data without a warrant to a private security firm investigating
pro-WikiLeaks activists.

The explosive details were contained in a report by Dutch investigative
journalist Brenno de Winter, published on earlier this week. Citing an
internal police file detailing an investigation called "Operation Talang,"
Winter wrote that PayPal was attempting to track down activists affiliated
with the hacker collective Anonymous.  The hackers had attacked the PayPal
website following the company's controversial decision to block payments to
WikiLeaks in December 2010. ...

Creative Disruption: Sandy Tells Us, *Let's Start Over*

"John F. McMullen" <>
Wed, 7 Nov 2012 18:36:21 -0500
John F. McMullen, Sandy—My 37th Column for the Westchester Guardian

I'm sitting in a Barnes and Noble in Mohegan Lake, NY—and it is like a
refugee camp *because no homes in the surrounding upper Westchester / Putman
counties in NY have power* due to Hurricane Sandy and, thus, Internet
connection is non-existent in the homes, so people flock to public Wi-Fi
sites. Unfortunately, this Barnes and Noble has very few public access
electric outlets and seven to fifteen people are gathered around the ones
that are available with multiple electric strips "daisy-chained" for laptop
and tablet connection.

Because of the multi-hundred people here (with at least half trying to
connect), at least as many as the bookstore gets in a week, Internet
connection is "iffy" and, even once connected, it is commonplace to be
dropped and have to roll the dice all over again to try to connect. The
Barnes and Noble free connection is based on an AT&T service and is usually
fairly reliable but is obviously overwhelmed today. If one is a CableVision
customer and is lucky enough to find one of the few seats near the window in
the coffee area, the Optimum Wi-Fi service is reachable but those seats
are few.

As recently as five years ago, hurricanes would have kept us in our house
-- but times have changed. It's not even enough now to have just the phone
capability and e-mail access that most smartphones provides provide. Now
the bookstore is filled with students doing papers and assignments;
business people entering orders and checking systems; and other maniacal
eccentrics, such as this writer, demanding access as a constitutional
God-given right.

There are at least 50 people on the line to get coffee and cakes, 10 times
the normal line and the jockeying for outlets is getting worse and worse --
how did we reach this stage where we are both so dependent and so
vulnerable? --- and what does this mean when we are in an age when we are
concerned about `cyberwarfare', which we are told may take out our
electrical grid?

Obviously, better computer security cannot help deal with havoc caused by
hurricanes nor with electrical outages because of downed trees and wires
but, when we see through this disaster, just how much more dependent we are
now on electric power than ever before, we can only imagine what it would
be like if someone were able to knock out the entire grid.

The present outage is limited to a small, albeit highly populated, section
of the east coast of the United States—and, driving 5 miles over here to
our local `refugee center',  I saw the large majority of businesses closed,
traffic lights out of operation, and gas stations unable to pump gas. In
New York City, the entire area south of 34th Street is without electricity
with thousands of businesses and hundreds of thousands of individuals
without power. One can only imagine what would be the impact of a
nationwide electrical shutdown—and, of course, the grid is controlled by
computer systems.

No matter what our technologists do, hackers, crackers, virus writers, etc.
all seem to be able to get around the safeguards which they install. For
years, the Computer Emergency Response Team (CERT— has been
warning users about security problems in Microsoft products, particularly
Internet Explorer and Outlook.  One is sure that Microsoft has been
addressing these problems as it finds out about them.  Yet on Oct 2012 25,
it issued a new report, Vulnerability Note VU#948750—Microsoft Outlook
Web, explaining a system hole under which an attacker could `execute
arbitrary scripting code'.

Microsoft is certainly not the only culprit in the security area. We have
all heard of infiltration of bank, credit card, on-line services (Yahoo,
etc.), and even Federal Government systems—infiltration that leads to
identity theft, financial loss, password compromises, and vandalism—and
what we have heard is only the tip of the iceberg. 2600: The Hacker
Quarterly magazine regularly publishes vulnerabilities of systems which,
hopefully, are soon repaired by at-risk firms (A weekly radio show, *Off The
Hook*, hosted by the editor of 2600, Emmanuel Goldstein, is heard on WBAI,
99.5 FM and is streamed at

It is obvious that what our virus programs, security systems, and systems
administrators have been doing isn't working—at least not 100% of the
time, and that is what is really required to protect our cyber

So, what to do? Dr. Peter G. Neumann, who has been monitoring computer
security for SRI International for forty years ,,, and has edited the Risks
Digest since 1985, analyzing the constantly changing technology world --
from the mainframe to the iPad—and the security challenges that the
constant innovation brings (for a full profile on Dr. Neumann, see the
recent *New York Times* article) is ready for a different approach.

  [Modesty suggests I truncate the rest of this.  John, Many thanks for
  the plug!  I strongly recommend his writings.  PGN]

Creative Disruption is a continuing series examining the impact of
constantly accelerating technology on the world around us. These changes
normally happen under our personal radar until we find that the world as we
knew it is no more.

Sandy: NYU hospital power outage... may have been from safety sensors

Danny Burstein <>
Sat, 10 Nov 2012 00:14:27 -0500 (EST)
(Re: RISKS-27.07) When the Con Ed substation serving a large part of
southern Manhattan was flooded out during Hurricane Sandy's storm surge, the
hospitals mostly uneventfully went to emergency backup power.

The glaring exceptions were NYU Hospital (and Bellevue).

It's now starting to look like the problem at NYU was exacerbated by some
safety switches.

(The usual cautions, of course, about early reports apply)

[ny times]

At this point, Dr. Grossman said, he could only theorize as to why the
generators had shut down. All but one generator is on a high floor, but the
fuel tanks are in the basement. The flood, he said, was registered by the
liquid sensors on the tanks, which then did what they were supposed to do in
the event, for instance, of an oil leak. They shut down the fuel to the

Re: Verizon FIOS phone service

"Bill Hopkins" <>
Tue, 6 Nov 2012 20:08:47 -0500
Solomon (RISKS-27.06) mentions Verizon wired service outages.  We lost power
locally for about 12 hours during Sandy's visit to the area.  The FIOS box
has a battery backup to deliver telephone service "for up to eight hours" in
a power outage.  Internet access died after a couple of minutes (the router
is on an UPS) and I assume the TV signals did also.

Phone service died in less than 8 hours, but when I plugged the FIOS box into
another UPS, both the phone line and Internet access came back.  Things were
stable until the power came back.

Whether this would be true with a more general power failure (we could see the
lights on further down the hill) will be the subject of a future "natural experiment."

Re: When your fuel pumps are below sea level... (Burstein, R-27.07)

Simson Garfinkel <>
Mon, 5 Nov 2012 21:41:30 -0500
Danny Burstein (and others) made passing reference to the 1965 blackout:

I looked into this back in 1996 for an article I was writing at the time.
Below is summarized from the notes I made back then...

According to the New York Times 1965 Index, p. 323, the November 9th
blackout of 1965 resulted in 800,000 people being stranded on the NYC
subway; many were evacuated, but 10,000 were stranded past midnight.
Governor Rockefeller ordered up 10,000 National Guardsmen to report to
armories to help residents and police. Military vehicles carried elderly and
the sick to hospitals.  All radio stations halted, but many resumed
broadcasting within 15 minutes.

NY Telephone Col, operating at full capacity on emergency diesel generators,
reported a record number of phone calls.

The Buffalo area darkened for only 40 minutes, but in New York City the
blackout lasted 13 1/2 hours.

The 1965 blackout followed a major 1961 blackout of Manhattan, which took
place on June 13, a 96 degree day. During the 1961 blackout Manhattan was
dark for between 2.5 and 4.5 hours. In 1963 Consolidated Edison assured the
government that a recurrence of the 1961 blackout would be unlikely.

Niagara Mohawk engineers said that the immediate cause of the 1965 breakdown
was a "quarrel" between giant generators in which some got out of phase with
others. The generators cut out one-by-one. It was difficult to restart the
generators without power.

"Once the northward power flow had been cut off through the Ontario hydro
plant, the current reversed direction, overloading lines in much of upstate
New York and triggering automatic cutoff devices there. Then, New England
and New York City power systems automatically tried to fill the power
vacuum, which imposed intolerable burden on their generating facilities and
these plans in turn cut themselves out. That is how the failure spread."

W. Sullivan discussed the blackout in light of speculation that civilization
is doomed by its increasingly complex technology. He cited opposing views
that such emergencies brought new proof of human ingenuity and adaptability.

Computers were seen as tools for preventing future blackouts.

*The New York Times* ran an editorial, Aladdin's Lamp Blacks Out. The
editorial said, in part, "Short of a nuclear bomb, the most crippling
affliction that can befall a modern metropolis is a total power failure.
The blackout that crippled New York and most of the Northeast last evening
was a dismaying reminder of the vulnerability of any community to a severing
of its electric lifeline."

On 7 Dec 1965, the Federal Power Commission issued its report on the
november 9th power failure. It said that the power failure would not have
occurred if the power systems involved had been following more careful
operating policies. It said that the immediate cause of the blackout was an
automatic shutdown of the power distribution line between the US and
Canada. The line had circuit breakers which were set to make the line cease
operating if the power load exceeded 375 million watts. That set point was
chosen in 1963 and had not been reviewed. "In the time since the setting was
determined, the average power load on the line controlled by the relay
increased to 356 million watts, and thus ordinary upward fluctuations in
power tripped the relay and started the whole blackout."

There were many concerns at the time that the increased interconnection of
power systems was responsible for the blackout. The Commission stated flatly
that more, rather than fewer, interconnections between power systems in
different areas were neededto provide reliable electrical service.

The report found that emergency vehicles and been rendered unusable during
the blackout because NYC gasoline pumps could not be run manually.  The
Commission charged the petroleum industry with finding some means of
operating gasoline pumps at service stations when electric power fails. The
Commission further said that elevators in the city should be equipped with
manual devices to move them to a landing.

The report noted the possibility of a failure of the proportions involved of
9 Nov 2012 had never been considered and said that studies were urgently
required based on the more stringent assumptions.

The day the report was released, 6 Dec 1965, southeast Texas was blacked
out for 25 minutes.

Please report problems with the web pages to the maintainer