The RISKS Digest
Volume 16 Issue 92

Thursday, 16th March 1995

Forum on Risks to the Public in Computers and Related Systems

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

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

Contents

Health card rips off ATM for $100,000
Roy Beimuts
A340 shenanigans
Les Hatton
Mistake of platform-specific instructions
Stanton McCandlish
The Manchurian Printer
Simson L. Garfinkel
Re: Scientology Blackmail Risk
Lance A. Brown
Jon Green
Re: Internet-Finland Privacy
Michael Jennings
Jumping to conclusions? (Lifeguard)
Peter da Silva
Re: Microsoft and Lotus spreadsheet errors
Bear Giles
Society and the Future of Computing
Phil Agre
Info on RISKS (comp.risks)

Health card rips off ATM for $100,000

"AGR CANADA, Research Station, Melfort, Sask." <SYSTEM@skrsme.agr.ca>
15 Mar 1995 12:18:46 -0500 (EST)
>From Saskatoon StarPhoenix, March 14:

"BC man sentenced for ATM scam"

     PORT HARDY, B.C. (CP)  —  A Vancouver Island man who used his health
card to steal $100,000 from a bank machine has been given a year in jail.
Richard Lee Mose, 22, of Port Alice, B.C., was found guilty of theft in
Campbell River court.  A Bank of Nova Scotia official said a processing
problem allowed Mose to use a non-bank card on an automatic teller.  With
his health-care card, Mose completed several transactions over several
hours and got $109,000 dollars.  But the bank machine recorded the
information stored on the health card, and police traced it back to Mose.

Roy Beimuts, Melfort, Saskatchewan     beimutsr@em.agr.ca


A340 shenanigans

Les Hatton <les_hatton@prqa.co.uk>
15 Mar 1995 15:56:26 +0000
The BBC news at 08.30 reported a slight problem which occurred on the
morning of 15 Mar 1995 with the ultra high-tech, packed full of software
and therefore utterly wonderful Airbus A340.

Apparently on the final part of its approach to Gatwick, both the pilots
screens went blank, to be replaced by a polite little message saying "Please
wait ...".  Somewhat unnerved, the pilots requested that the plane turn
left, but it turned right instead.  They then tried to get it to adopt a 3
degree approach to the runway, but it chose a 9 degree plummet instead.  At
this point, from the report, they appeared to gain manual control and landed
safely.  It is not clear who will pick up the dry-cleaning bill.

Vis a vis this sort of thing, I was at a talk recently, given by the CAA (UK
Civil Aviation Authority), at which it was stated that in the past
generation of civil aircraft, most of the software problems were reported in
the Flight Management System.  Not surprisingly, this was the most complex
part of the aircraft software system.  Not any more it isn't.  During the
talk, it was also admitted that the newer generation of aircraft such as the
A340, other software systems including active systems were "at least as
complicated".  So what next ?
 I suppose it follows on nicely from the story in the October 1994 Risks
whereby a Japanese Air Force T-4 jet trainer ejected one of its pilots.
Perhaps it didn't like him.  :-)

Les Hatton, Ph.D. C.Eng, Director of Research & Engineering, Programming
Research Ltd, England   les_hatton@prqa.co.uk    +44 (0) 1 932 888 080


Mistake of platform-specific instructions

Stanton McCandlish <mech@eff.org>
Wed, 15 Mar 1995 17:29:33 -0500 (EST)
One thing that's bugged the hell out of me for several years is the
failure of many people giving instructions on finding net resources
to do so in a generic manner.  Here's a great example:

       1.  Get into the WWW.
       2.  Enter the letter G so you can enter the address for the
           test site.
       3.  At the prompt, enter the following:

               http://sunsite.unc.edu/jembin/mb.pl

This was from an announcement off of the Net Happenings list. Anyone
following these instructions will get nowhere unless they are using lynx
(unless there's another browser that uses the same keystrokes to specify
a URL, which I doubt.)  The first item implies that the instructions are
generic, for whatever means one uses to browse WWW sites (and,
incidentally, indicate a fundamental misunderstanding of the technology,
since the WWW is not a "thing" that one "gets into", but is a method, a
process, a standard.

All the author had to do was put:

       URL:http://sunsite.unc.edu/jembin/mb.pl

At any rate, the application-dependent specificity of the instructions,
though not a problem for anyone that's been using Internet tools for
very long, is potentially very confusing and frustrating to less
experienced users.  And these users exist in ever greater numbers, as
systems like AOL and Delphi become increasingly Internetly.  I don't
mean to single out AOL for any more maligning, but as a forum sysop
there besides my other online services duties, I can say from firsthand
experience that the technical skill level of AOL users is, on average,
staggerly lower that that of the typical shell account user, or almost
anyone else online for that matter.

Typical questions I receive from AOL users are: "What happened?"  "How do
I read this file?"  "If I want to download an ASCII file do I have to use
the ASCII protocol, or can I use ZModem?"  "What do I join?"  "Help!"
"What is a 'GIF graphics viewer'?"  Those aren't snippets. Those are
entire message bodies.  Such people, confronted with the site instructions
I quoted from Net Happenings, are never going to reach the site in question,
will probably pound people like me with more questions like the above
example in an effort to figure out what to do, and may eventually give
up in disgust (a few might actually RTFM and get a clue, but not
everyone is a born geek.)

So, there are two RISKS to keep in mind here:

1) If you use platform-specific instructions, you are doing other readers a
disservice, and should at very least note that they are platform-specific
(or application-specific, or whatever, as apropos).  Better yet, just use
generic instructions that anyone can use.

2) If you need help with something, but don't *include enough information*
for them to actually help you, you're unlikely to get a useful response.
Include info such as: what kind of system you have, what it's relevant
configuration is, what went wrong in detail, what all the items (files,
whatever) are specifically that you are referring to, etc.

          Stanton McCandlish
        mech@eff.org

Electronic Frontier Foundation


The Manchurian Printer

Simson L. Garfinkel <simsong@pleasant.cambridge.ma.us>
Wed, 15 Mar 1995 21:28:37 -0500
The Manchurian Printer, (C) 1995, Simson L. Garfinkel
[The Boston Sunday Globe, March 5, 1995, Focus Section, Page 83]

Simson L. Garfinkel

Early this month, Hewlett-Packard announced a recall of 10,000 HP OfficeJet
printer fax copiers. The printer's power supplies may have a manufacturing
defect that could pose an electrical shock hazard. HP says that it
discovered the problem with its printers during routine testing; HP was
lucky: printers can be very dangerous devices. A typical laser printer, for
example, can draw hundreds of watts of power, generate internal temperatures
high enough to burn a wayward human hand, and even, under the right
circumstances, start a fire.

Most manufacturers, of course, try to design their printers to minimize such
risks. Increasingly, however, there is a chance that companies might
intentionally design life-threatening flaws into their products so that the
flaws can be exploited at a later time. These fatal flaws might be
intentionally built into equipment manufactured overseas, as a kind of
"insurance policy" in the event of a war between that foreign country and
the United States. The flaws might form the basis for a new kind of
corporate warfare. Or the flaws might be hidden by disgruntled employees
contemplating extortion or revenge.

Indeed, U.S. military planners are increasingly worried about this sort of
possibility, they place under a heading "Information Warfare." Nevertheless,
although the threat of Information Warfare is very real, an even bigger
danger is that the Department of Defense will use this threat to convince
the new Congress to repeal the Computer Security Act of 1987. This would
effectively allow the National Security Agency to declare martial law in
cyberspace, and could place the civilian computer industry into a tailspin.

To understand what the military is afraid of, imagine the Manchurian
Printer: a low-cost, high-quality laser printer, manufactured overseas, with
built-in secret self-destruct sequence. For years these printers could lay
dormant.  But send them a special coded message---perhaps a long sequence of
words that would never normally be printed together---and the printer would
lock its motors, overheat, and quickly burst into flames. Such an attack
might be the first salvo in an out-and-out war between the two countries.
Alternatively, an enemy company might simply use printers to start selective
fires, damage economic competitors, take out key personnel, and cause
mischief.

Unlike the movie the Manchurian Candidate, the technology behind the
Manchurian Printer isn't science fiction. Last October, Adobe accidentally
shipped a "time bomb" in Photoshop version 3.0 for the Macintosh. A time
bomb is a little piece of code buried inside a computer program that makes
the software stop running after a particular date. Adobe put two time bombs
into its Photoshop 3.0 program while the application was under development.
The purpose behind the time bombs was to force anybody who got an advance,
pre-release copy of the program to upgrade to the final shipping version.
But when it came time to ship the final version of Photoshop 3.0, Adobe's
engineers made a mistake: they only took out one of the bombs.

An engineer inside Adobe learned about the problem soon after the product
was shipped, and the company quickly issued a recall and a press release.
Adobe called the time bomb a "security code time constraint" and said that
"although this is an inconvenience to users, the security constraint neither
damages the program or hard drive, nor does it destroy any files."

It only takes a touch of creativity and a bit of paranoia to think up some
truly malicious variants on this theme. Imagine that a company wants to make
a hit with its new wordprocessor: instead of selling the program, the
company gives away free evaluation copies that are good for one month.
What's unknown to the users of this program is that while they are typing in
their letters, the program is simultaneously sniffing out and booby-trapping
every copy of Microsoft Word and WordPerfect that it finds on your system.
At the end of the month, all of your wordprocessors stop working: Instead of
letting you edit your memos, they print out ransom notes.

Any device that is equipped with a microprocessor can be equipped with such
a booby-trap. Radios, cellular telephones, and computers that are connected
to networks are particularly vulnerable, since an attacker can send them
messages without the knowledge or consent of their owners. Some booby- traps
aren't even intentional. What makes them particularly insidious is that it
is almost impossible to look at a device and figure out if one is present or
not. And there is no practical way to test for them, either. Even if you
could try a million different combinations a second, it would take more than
200 years to find a sequence that was just 8 characters long.

* * *

Information Warfare isn't limited just to things that break or go boom. The
Department of Defense is also worried about security holes that allow
attackers to break into commercial computers sitting on the Internet or take
over the telephone system.

"This nation is under IW attack today by a spectrum of adversaries ranging
from the teenage hacker to sophisticated, wide-ranging illegal entries into
telecommunications networks and computer systems," says a report of the
Defense Science Board Summer Study Task Force on Information Architecture
for the Battlefield, and issued last October by the Office of the Secretary
of Defense.

"Information Warfare could pervade throughout the spectrum of conflict to
create unprecedented effects. Further, with the dependence of modern
commerce and the military on computer controlled telecommunication
networks, data bases, enabling software and computers, the U.S. must
protect these assets relating to their vulnerabilities," the report warns.

Information warfare changes the rules of war fighting, the report warns. A
single soldier can wreak havoc on an enemy by reprogramming the opposing
side's computers. Modern networks can spread computer viruses faster than
missiles carrying biological warfare agents, and conceivably do more damage.
Worst of all, the tools of the information warrior are readily available to
civilians, terrorists and uniformed soldiers alike, and we are all potential
targets.

Not surprisingly, the unclassified version of the Pentagon's report barely
mentions the offensive possibilities of Information Warfare---capabilities
that the Pentagon currently has under development. Nevertheless, these
capabilities are alluded to in several of the diagrams, which show a keen
interest by the military in OOTW---Operations Other Than War.

"They have things like information influence, perception management, and
PSYOPS---psychological operations," says Wayne Madsen, a lead scientist at
the Computer Sciences Corporation in northern Virginia, who has studied
the summer study report.  "Basically, I think that what they are talking about
is having the capability to censor and put out propaganda on the networks.
That includes global news networks like CNN and BBC, your information
services, like CompServe and Prodigy," and communications satellite
networks. "When they talk about 'technology blockade,' they want to be able
to block data going into or out of a certain region of the world that they may
be attacking."

The report also hints at the possibility of lethal information warfare.
"That is screwing up navigation systems so airplanes crash and ships runs
aground.  Pretty dangerous stuff. We could have a lot of Iranian Airbuses
crashing if they start screwing that up," Madsen says. Indeed, says Madsen,
the army's Signal Warfare center in Warrenton, Virginia, has already invited
companies to develop computer viruses for battlefield operations.

Our best defense against Information Warfare is designing computers and
communications systems that are fundamentally more secure. Currently, the
federal organization with the most experience in the field of computer
security is the National Security Agency, the world's foremost spy
organization. But right now, NSA's actions are restricted by the 1987
Computer Security Act, which forbids the agency from playing a role in the
design of civilian computer systems. As a result, one of the implicit
conclusions of the Pentagon's report is to repeal the 1987 law, and untie
the NSA's hands. Indeed, the Pentagon is now embarking on a high-level
campaign to convince lawmakers that such a repeal would be in the nation's
best interests.

This argument confuses security with secrecy. It also ignores the real
reasons why the Computer Security Act was passed in the first place.

In the years before the 1987 law was passed, the NSA was on a campaign to
expand its power throughout American society by using its expertise in the
field of computer security as a lever. NSA tried to create a new category of
restricted technical information called "national security related
information." They asked Meade Data Corporation and other literature search
systems for lists of their users with foreign-sounding names. And, says
David Banisar, a policy analyst with the Washington-based Electronic Privacy
Information Center, "they investigated the computers that were used for the
tallying of the 1984 presidential election. Just the fact that the military is
looking in on how an election is being done is a very chilling thought. After
all, that is the hallmark of a banana republic."

The Computer Security Act was designed to nip this in the bud. It said that
standards for computer systems should be set in the open by the National
Institute of Standards and Technology.

Unfortunately, the Clinton Administration has found a way to get around the
Computer Security Act. It's placed an "NSA Liaison Officer" four doors down
from the NIST director's office. The two most important civilian computer
standards to be designed in recent years---the nation's new Escrowed
Encryption Standard (the "Clipper" chip) and the Digital Signature Standard
were both designed in secret by the NSA. The NSA has also been an unseen
hand behind the efforts on the part of the Clinton Administration to make
the nation's telephone system "wiretap friendly."

Many computer scientists have said that the NSA is designing weak standards
that it can circumvent, so that the nation's information warfare defenses do
not get in the way of the NSA's offensive capability.  Unfortunately,
there's no way to tell for sure. That's the real problem with designing
security standards in secret: there is simply no public accountability.

In this age of exploding laser printers, computer viruses, and information
warfare, we will increasingly rely on strong computer security to protect our
way of life. And just as importantly, these standards must be accountable to
the public. We simply can't take our digital locks and keys from a Pentagon
agency that's saying "trust me."

But the biggest danger of all would be for Congress to simply trust the
administration's information warriors and grant their wishes without any
public debate. That's what happened last October, when Congress passed the
FBI's "Communications Assistance for Law Enforcement Act" on an unrecorded
voice vote. The law turned the nation's telephone system into a surveillance
network for law enforcement agencies, at a cost to the U.S. taxpayer of $500
million.


=========WHAT FOLLOWS ARE CAPTIONS FOR THE ART===========
Photo: Box of Microsoft Word 6.0

Even though it's illegal, a lot of people like to "try out" software by
making a copy from a friend before they plunk down hundreds of dollars for
their own legal copy. Computer companies say that this is a form of software
piracy: many who try never buy. More than 2 billion dollars of software is
pirated annually, according to the Business Software Alliance.

One way that companies like Microsoft and Novel could fight back is by
booby-trapping their software. Sure, customers wouldn't like it if that stolen
copy of Microsoft Word suddenly decided to erase every letter or memo that
they've written in the past month, but what legal recourse would they have?
=====================
Photo: Cellular Telephone

Is your cellular phone turned on? Then your phone is broadcasting your
position every time it sends out its electronic "heartbeat."  Some law
enforcement agencies now have equipment that lets them home in on any
cellular telephone they wish (similar technology was used recently to catch
infamous computer criminal Kevin Mitnick). Perhaps that's the reason that
the Israeli government recently ordered its soldiers along the boarder to stop
using their cellular telephones to order late night pizzas: the telephone's
radio signal could be a become a homing beacon for terrorist's missiles.
===================
Photo: Floppy Disk

Beware of disks bearing gifts. In 1989, nearly 7000 subscribers of the British
magazine PC Business World and 3500 people from the World Health
Organization's database received a disk in the mail labeled "AIDS
Information Introductory Diskette Version 2.0". People who inserted the
disks into their computers and ran the programs soon found out otherwise:
the disks actually contained a so-called trojan horse that disabled the
victims' computers and demanded a ransom.
==================
Photo: A computer with a screen from America Online, and a modem

Several years ago, users of Prodigy were shocked to find that copies of
documents on their computers had been copied into special "buffers" used by
Prodigy's DOS software. Prodigy insisted that the copied data was the result
of a software bug, and it wasn't spying on its customers. But fundamentally,
if you use a modem to access America Online, Prodigy or Compuserve, there
is no way to be sure that your computer isn't spying on you while you surf
the information highway.
==================
HP's recall affects only OfficeJet  printers with serial numbers that begin
US4B1-US4B9, US4C1-US4C9, US4BA-US4BU, or US4CA-US4CK. Worried
about your OfficeJet? Call HP at (800) 233-8999.
===============
Simson L. Garfinkel writes about computers and technology from his home in
Cambridge, Massachusetts.


Re: Scientology Blackmail Risk (Vilkaitis, RISKS-16.91)

"Lance A. Brown" <lab@biostat.mc.duke.edu>
Wed, 15 Mar 1995 10:58:37 -0500
Vilkaitis is not correct.  Postings on alt.security.pgp stated that Finnish
authorities secured a warrant to seize the equipment the Finnish Anonymous
Server runs on.  The owner of the Server negotiated a deal with the
authorities where he released the identity of _one_ user of the Server and
the authorities didn't seize the equipment.

My understanding of the behind-the-scenes goings on is that the Church of
Scientology is bringing copyright charges against one of its former
ministers who is now a vocal critic of the CoS on the Internet.  The
sequence of events, as I understand it, is that someone used the Finnish
Anonymous Server to post allegedly copyright material on USENET.  The CoS
asked the FBI to talk to Interpol who talked to the Finnish Police about
getting the ID information of the anonymous poster.  Once this ID
information was released by the owner of the Server it was immediately
handed over to CoS people.

   [Also noted by Kevin.P.Maguire@jpl.nasa.gov (Kevin Maguire) and
   "Matti E. Aarnio [OH1MQK]" <mea@mea.cc.utu.fi>.  PGN]


Re: Scientology Blackmail Risk (Vilkaitis, RISKS-16.91)

Jon Green <jonsg@diss.hyphen.com>
Wed, 15 Mar 1995 09:40:35 +0000 (GMT)
  [... more as noted by Lance Brown deleted ...  PGN]

Nonetheless, this does represent a worrying precedent.  There are persistent
rumours that the entire user base of at least one anonymizing service has
been compromised by covert action by a security agency, and that's just the
start.  As has been pointed out elsewhere, any agency monitoring
international communications (NSA in the US and GCHQ in the UK, to name two)
should have little trouble matching anon ID with real ID if the message is
in plaintext and the server in another country.  Matching messages where the
first leg is PGP-encoded (and the server decodes before retransmission) would
be more difficult, but by no means impossible.

The only sensible conclusion is that anon remailers provide anonymity from
your peers, not from the law.  If you use them illegally, you may well be
identified.  Them's the breaks.

jonsg@hyphen.com  jon@sundome.demon.co.uk


Re: Internet-Finland Privacy (RISKS-16.91)

Michael Jennings <M.J.Jennings@amtp.cam.ac.uk>
15 Mar 1995 17:16:33 GMT
>Case #1 ...  A Swedish journalist-researcher "reveals" that an Anonymous
>Case #2 ...  Finnish Police receive a request from U.S. law enforcement

    There have been suggestions on the net (in alt.privacy.anon-server,
I think) that these two events may well have been related: specifically that
the Church of Scientology might have been indirectly responsible for the
'This anon server is used by pedophiles: shock, horror' stories in the first
place, as an attempt to discredit the anon server in order to make the
police more likely to raid it for them/ get it shut down. This is only
speculation, but it is consistent with their style. It is their standard
policy to attempt to discredit their opponents through character
assassination at the same time as they attack them through legal means.

    For instance, one of the recent posters of copyright material to
alt.religion.scientology was described in passing in a Scientology press
release as someone who conducted execution-style killings of his pets in
front of his children. Paulette Cooper, who wrote a book entitled _The
Scandal of Scientology_, found a circular (supposedly written by `a
concerned neighbor') circulating around her apartment block suggesting her
`removal from our residence, and if possible, have her put under appropriate
psychiatric care.' Several critics of scientology have been accused (and
sometimes tried) for crimes that they did not commit, after having
(apparently) been framed by scientologists.

    Many people have used the anon server to post critical articles
about Scientology.  I suspect the church would like it discredited. (Massive
amounts of information about this and the church in general can be found at
http://falcon.cc.ukans.edu/~sloth/sci/sci_index.html#diary)

Michael Jennings, Department of Applied Mathematics and Theoretical Physics
The University of Cambridge.        mjj12@damtp.cambridge.ac.uk


Jumping to conclusions? (Lifeguard)

Peter da Silva <peter@nmti.com>
Wed, 15 Mar 1995 09:44:33 -0600
> "Anybody who shoots at you from any direction would be immediately located
> and subject to return fire," says Thomas Karr, head of the Lab team.

I don't read this as "the weapon would automatically return fire" but "the
police officer would be able to return fire".


Re: Microsoft and Lotus spreadsheet errors (Bellovin, RISKS-16.90)

Bear Giles <bear@tigger.cs.colorado.edu>
Wed, 15 Mar 1995 16:56:25 -0700
>They ended up doing the calculation, and storing a compressed table giving
>the difference between the calculated values and the legal ones.  Never mind
>reality — custom ruled.

I can beat that.  I recently ported a mess of old FORTRAN meteorological
code to C++, and I extensively expanded the validation suite in the process.

During the process I learned that a number of my newly ported functions were
returning values a hair off (typically < 0.5%, when the number of
significant digits should have resulted in errors two orders of magnitude
smaller).  I traced the problem to the fact that I was using the best
currently known physical constants, while the standard reference tables were
using values from the 1960s.  Reasonable, since it was published in 1965,
but I had to address the differences in the results.

After consulting with the working meteorologists, I eventually put the
1960-era physical constants back into the software.  As a practical matter,
the resolution of the available data is still coarse enough that the small
differences won't make any difference, and it is easier for them to compare
the final results than for FORTRAN programmers to understand scientific C++
classes.

But it still grates.  At least I was able to eliminate a number of duplicate
functions.  Previously, someone had made a token effort to include a
validation suite (with typically <10 test cases per function), but they made
no effort to identify slow and/or inaccurate functions.  A tradeoff of speed
for accuracy is often justified, but who would ever want to use a function
which is significantly slower and less accurate than another one?

Bear Giles  bear@fsl.noaa.gov


Society and the Future of Computing

Phil Agre <pagre@weber.ucsd.edu>
Wed, 15 Mar 1995 11:52:07 -0800
The conference on Society and the Future of Computing (SFC'95) will be
held from June 11th to 14th in Durango, Colorado.  This conference is
an initiative of the US Public Policy Committee of the Association for
Computing Machinery (USACM).  Its focus is on opportunities for socially
beneficial applications of computing technology: visions of what's
possible ten years from now and agendas for computer science research
that can make those visions come true.

Conference speakers include:

   Gary Chapman, University of Texas, Austin
   John Cherniavsky, National Science Foundation
   Peter J. Denning, George Mason University
   Linda Garcia, Office of Technology Assessment
   S. Joy Mountford, Interval Research Corporation
   Don Norman, Apple Computer, Inc.
   Roy Pea, Northwestern University
   Paul Evan Peters, Coalition for Networked Information
   Virginia E. Rezmierski, University of Michigan
   Leslie Sandberg, Institute for Telemedicine
   Paul Young, National Science Foundation

Full information about the conference is available from the conference
web pages at:

  http://www.lanl.gov/LANLNews/Conferences/.sfc95/sfcHome.html

or send an e-mail message that looks like this:

  To: rre-request@weber.ucsd.edu
  Subject: archive send sfc-95

Poster sessions will be an important part of the conference.  For
full information on poster submissions, see the conference web pages
or contact Doug Schuler <douglas@scn.org>.  The deadline is April 1st.

Student scholarships are available as well; information is available
from sfc95-students@lanl.gov.

The basic idea is to gather 250 people in a nice place to work hard and
have fun learning how to take social issues into account when setting
agendas for computer science research.  We hope you can join us.

Phil Agre, UCSD

Please report problems with the web pages to the maintainer

x
Top