THE CASE OF THE BENT BANKER

Copyright Dr Alan Solomon (1986-1995)

I first met Holmes when I was looking for lodgings at a reasonable
price, having just won my Doctorate, and before I'd set up my practice.
We took to each other immediately - he the tall, slender aesthete, and
me the short, tubby gourmand.

Not many people know that the original Holmes had a son.  In those days,
it was taboo to write about extra-marital affairs, but careful readers
of Dr Watson's chronicles will remember the glamorous Irena, and the way
that Holmes quite uncharacteristically let her escape.  The more astute
reader will have realised why - she was pregnant;  a little Holmes was
in the oven.  In the course of time, the son married and had children,
and named one of them after Grandfather.

Sherlock and I got along quite well.  We saw little of each other, as I
was out during the day attending my patients, and he seemed to prefer to
work at night.  He had a few foul habits;  his smelly pipe was not as
bad as his curious misapprehension that he could play the violin.  He
wouldn't let Mrs Hudson clean his room at all, muttering something about
the thicknesses of dust being important to his filing system.

We had a lot in common.  We both worked as consultants;  me in the
computer field, and he as a detective.  He once confided to me that he
hadn't really chosen that career, it had been more or less forced on him
by his name.  I could sympathise with that;  the name of Solomon has led
people to expect an unnatural amount of wisdom from me, and I didn't
choose computers - they chose me.

One day, a distinguished-looking gentleman was shown into our rooms.
"Good evening", said Holmes.  "I see that you are a banker, and have
been wrongly accused of fraud.  Would you like me to help you clear your
good name?" The banker looked startled.

Holmes loved doing that sort of thing.  He rarely explained how he
arrived at his conclusions, but I knew his methods, and could apply them
myself.  The gentleman had arrived in a Rolls Royce, and was wearing the
sort of velvet-trimmed coat that one associates with people in the
financial professions.  Stockbrokers come in Porsches and BMWs, so he
must be a merchant banker.  The only crime that bankers admit to is
fraud, and it must be a wrongful accusation, or he wouldn't be
consulting Holmes.

"I can see I have come to the right place," he said admiringly.  "Your
reputation is well deserved.  Let me explain the situation."

The bank had been losing money;  fairly large amounts were simply
disappearing from various people's accounts.  A vigorous internal
enquiry had led to the fact that Holmes' new client had been paying
large sums of money into one of his accounts, and he was unable to
explain this to the other bank directors.

"They intend calling in the police - Holmes, I swear by Mammon that I
did nothing that was wrong, only what any banker would have done.  You
must believe that I'm innocent, you have to help me."

Holmes looked thoughtful.  I knew from long acquaintance what was going
on in his mind.  He wasn't thinking about the crime, he was assessing
the net worth of his client, and working out how much the market would
bear.  It's something all consultants do - you have to estimate the
length of the wool before you begin to remove the fleece.

Holmes named a daily rate that made me swallow - I don't charge that
even for recovering data from a Novell file server.  But the banker
didn't seem to think it excessive, and agreed.

After he left, Holmes rubbed his hands together.  "Now, Watson, the
game's afoot".  He often said that, and I never did understand why - my
name never has been Watson.

For the next few days, I saw very little of him.  One of my patients had
become stricken with a severe file corruption, and needed a lot of
careful cutting and stitching.  But eventually, I'd got him back on the
mend (thank heavens for Debug) and was able to return to a normal
working day.

It was not until five days after our Banker had left that I saw Holmes
again.  He looked terrible.  His face was drawn and haggard, as of one
who hadn't slept for a week, and I thought I detected the effects of his
favourite drug.

"Doctor Solomon", he cried as soon as he saw me, "you *have* to help
me".  Now obviously I've heard this many times before, and I reflexively
swung into my bedside manner.  "There there, Holmes", I said.  "Sit
down, have a cup of tea, a muffin, and tell me all about it."

"It's terrible", he began.  "It's all computers!  Grandfather's notes
don't help in the slightest." Sherlock Senior had kept copious notes
during his life as a consulting detective, and over the seventy years
that he had practised, he'd built up case histories covering every
eventuality.  Sherlock Junior's main modus operandi was to find the case
that most closely fitted the situation under investigation, and work
from there.  "But there was no computer crime in Grandfather's day!"

So young Holmes had tried to strike out on his own.  He started off by
asking to see the ledgers, but of course they were all on the bank's
mainframe.  So poor Holmes had been thrown in at the deep end of
computing, trying to track transactions through a system that seemed to
have been written to baffle rather than enlighten.  "I'm sick of CICS,"
he said, "And COBOL is purely an employment protection scheme".  The
bank's programmers had taken great delight in throwing googlies at the
Great Detective, who had been intellectually defenceless for the first
time in his life.  He had then made the capital mistake of trying to
substitute hard work for intelligence and knowledge, and had worked
himself to exhaustion by working non-stop for 120 hours.  I prescribed a
sedative, hot with milk and lots of sugar, and when he'd finished his
cocoa, I tucked him up in bed.

He awoke late the next day, considerably refreshed, but depressed.
"This case is beyond me," he said miserably.  I think it was the idea of
such a luxuriant fleece slipping out of his grasp that really galled
him.  "I could perhaps help ..." I said.  "50-50" he said at once.
"70-30" I said, and we settled on 60-40.  "All we have to do", I said,
"is show that this crime could have been committed by an outsider.
Then, any competent lawyer would be able to cast reasonable doubt on the
rather circumstancial evidence that our client's bank account is more
than it should be.  In that situation, the bank would not even
consider a prosecution."

"Agreed," said Holmes, "but how can we show that this could have been an
outside job?  The bank says that its defences are impenetrable."

"By penetrating them ourselves, of course", I said.  Holmes looked
impressed.

End of part 1 - to be continued

Part 3

Holmes and I decided that we would do it - we'd break into the bank.
Really, the only way to convince someone in charge of security that
there are holes in his defences, is to find at least one of them.

First, Holmes spoke to the bank's chairman, who was apparently a member
of the same club as his brother.  He explained that we wished to test
the bank's defences, as the notorious criminal Moriarty was planning a
daring robbery, and we had reason to think that his bank was the target.
Naturally, he agreed at once, and wrote us a letter that would give us
immunity from prosection should we be discovered.

I've always wanted to crack a bank.  I've found that all security
systems have holes in them, usually left by the people who actually use
them.  If you make a password system sufficiently complex, people will
start writing them down, either on the terminal itself, or else in some
convenient place nearby.

People often use the same password for many different computers, and so
if you capture one, you capture them all.  But the simplest methods are
the best.  There is a well known list of ten passwords;  if passwords
are chosen by users, about 15% of them will choose something on this
list.  Passwords are rarely changed by users, and even if a password
change is forced, most users alternate between two, to avoid the
embarrassment of forgetting their password.  Many people use their own
name, or their wife or child's name.

For example, when Robert Schifreen hacked into Prince Phillips Prestel
mailbox, he did it by simply stumbling across a file containing a list
of passwords that some fool had left lying around on the computer;  the
subsequent prosecution was grossly unfair.  When I hacked Schiffreen's
Telecom Gold password, I did it simply for the fun of it, and avoided
using it because BT seem to have no sense of humour, and I didn't want
to suffer the same fate as he had (he minds his Ps and Qs a lot better
now).  I don't think that simply knowing someone else's password is a
criminal offence, and I have never told anyone else what it was, but I
used one of the simple techniques described above.

My favourite method is the "rabbit trap".  You write a program that
simulates the log-on sequence of the computer, and leave it running on a
terminal.  When someone logs in, you prompt for the password in the
usual way, write it to a file, then exit with some incomprehensible
error message, saying that the system is unavailable.

Not only rabbits can be caught that way, you can even catch quite
sophisticated people, if you can duplicate the log-on sequence fairly
accurately.  I nearly got caught that way myself once, but I noticed a
slight difference from the real log-on, broke out of the program, hacked
back to the file that it was writing and captured several other
passwords that way.  Then I changed the password of the person who had
left the trap for me, so that he wouldn't be able to get on the computer
without the humiliation of asking the system manager to change his
password.  Computers are good fun, even mainframes.

But I've never even tried to hack a bank computer, except for recovering
a few IBM AT hard disks for various banks.  The fuzz get pretty heavy if
you're caught with your terminal in a bank's computer, as they can't
understand the motive of cracking something simply because it is an
intellectual challenge.  This was a golden opportunity, and I set to
work at once.

The first thing I did was to phone around a few friends, asking for a
contact in the team of programmers who had set up the system.  This
could easily be a quite innocent request;  perhaps I admired the work
and wanted something similar written.  I hit a winner immediately - one
of the junior programmers on the team contacted me.  He was quite bitter
about the fact that as soon as the project had been completed, he'd been
made redundant (its a hard world) and happily explained to me how the
systems hung together.

Idle cash means wasted cash, he told me, so each day at five, the bank's
accounts are consolidated into a pool, which is then lent on the London
Inter Bank Overnight market at the prevailing rate of interest.  The
next day, the funds are redistributed amongst the various operations.
He also explained to me how the nominal ledgers fed into the general
ledger, how they accounted for VAT, the payroll system that they
operated (their own payroll was naturally done by another bank;  only
other organisations' payroll were done as a chargable service).  He
explained to me about the branch systems, the automatic controls on
lending and bond dealing, how the accounts were made up for the Old
Lady, the methods by which the tellers fed data into the system.  The
payroll interested me - perhaps I could find a way to funnel money out
of the bank via a dummy payroll?

Young George was very interesting, and he was clearly quite tickled to
be helping the Great Detective Sherlock Holmes in an investigation.
George didn't actually tell me anything naughty like a password, of
course, and I wouldn't have asked him for one.  There wouldn't have been
any point, anyway, as the first thing that the bank would do when he
left is cancel his user-id.  I did ask him what comms package he used to
log in, though, and he told me he used Procomm.  I expressed interest in
this, and asked him for a copy;  Procomm is Shareware, and really is the
best general-purpose comms package.

When the copy of Procomm arrived, I fired it up and had a look at the
phone directory.  Splendid - it was full of George's commonly-used
numbers, including the number he used to dial the bank.  I also had a
quick look at his keyboard macros, but there weren't any interesting
passwords there;  there's not a lot of point in logging on to an amateur
Bulletin Board under someone else's name, since most good amateur boards
will let use use it for free anyway.  But that phone number was
interesting.  Surely they couldn't be *that* stupid?

I took my little portable Mitsubishi tape recorder, and went to a public
call box to dial in;  no point in doing it from home.  Something at the
other end answered, and I recorded the tones it sent.  Back at home, I
listened to them carefully - they were not one of the standard modem
carrier tones.  So they weren't *that* stupid.

Holmes watched as I connected the output from the tape recorder to an
A-to-D interface card.  I then fed the tones into the port, digitised
them and captured it to disk.  I ran the digital data through a
frequency analyser program and sat and looked at the result.

I had a strong suspicion that I was looking at something illegal.  BT do
not let you make just any old noises over the telephone system.  If
you're a person, you can whistle, hum or even sing down the line, but if
you're a modem, there are only twelve different tones you can make.  You
aren't allowed, for example, to use the Bell tones (the US standard)
although no-one that I knew ever worried about this ban - how else can
you dial up a US service?

Perhaps the bank had special BT permission to use these different tones;
anyhow, it meant that I couldn't use my trusty Quattro to hack this
bank.  I needed a modem that would use these funny tones, so I needed to
find out what the bank was using.

I could have asked George, but he'd been very helpful, and I didn't want
to get him into trouble.  So I put out a fairly innocuous message over
Fidonet, and attached the file containing the digitised sounds.  "Does
anyone know", I asked innocently, "which modem uses these tones?".
Fidonet is an amateur packet switching system.  At 3 am each night Fidos
all over start autodialling each other to forward messages and files.
The system is configured as a star, with each central point acting as a
node in a bigger star, and each of those acting as a node in a
Superstar.

My message reached the 100 Fidos in my local network that very night,
and another 2000 Fidos all over the world the night after.  Each Fido
has perhaps a hundred regular callers, plus the Dutchies that the Fido
calls to pass messages back to people who'd rather be called by Fido
that the other way round.  So about half a million people had at least
seen my question, and it wasn't really too surprising that a couple of
them knew the answer.  I really don't know how people without access to
Fidonet manage to survive.

Apparently, the tones were characteristic of a modem called the Data
Guardian.  I contacted the manufacturer, and posed as a potential
customer.  I expressed my worry about people hacking into my Fido and
asked if they had anything that might help.  They should really have
smelt a rat, since Fidos are all public access - you actually encourage
strangers to call you.  But, of course, they weren't plugged into the
scene.  Totally blinded by greed at the thought of another punter lining
up for the slaughter, they wanted to send me an evaluation unit, but
with some difficulty I restrained them, and asked merely for a brochure,
plus the technical specs "so I can show it to my DP manager".  One of
these days my nose will be about twelve inches long;  those of you who
have seen me will probably have wondered how it got so large.  Mince
pies, mince pies.  But if I'd had an actual modem from them, they'd have
followed up with phone calls and sales calls, whereas this way, the
worst that would happen was that I'd get put on their mailing list.

I read their technical documentation.  They'd been fearfully clever.
The tones were very non-standard, and instead of use frequency or
phase-shift modulation, they used delta-modulation.  And they had a
password-and-callback scheme, so that you dialled the remote modem, and
after a few minutes, it would call back, but only to one of a list of
authorised numbers.

I explained all this to Holmes.  "Doctor Solomon, this is terrible - it
is a foolproof system", he exclaimed, "This means that we cannot enter
by this route." "On the contrary, my dear Holmes, it is excellent news.
This is precisely the sort of system that is vulnerable simply because
it is thought to be impenetrable.  The security staff believe the
propaganda of the supplier, and turn their attention to more interesting
areas, such as the blaggers.  They are most unlikely to monitor such a
secure system for signs of hacking." I had to admit, though, it didn't
look easy.  I had to use the tones that the Guardian used, then get it
to call me back, and only then could I start trying to guess an account
name and password.  You could see that they were quite right to believe
that hacking this system would be impossible.  I decided that I would
have to cheat, to do it a different way.

End of part 2 - to be continued

Part 3

We seemed to be trying to break into an impenetrable security system,
which used non-standard carrier tones, so an ordinary modem wouldn't
work, and which would only connect you by calling you back, and it would
only call you back if you were on the list of authorised numbers.

Have you noticed how often you find a clever way to solve a problem, but
when you explain it, you get told "Oh yes, of course you could do it
*that* way".  Mathematicians have a word for it;  we call it a "cook".
de Bonio calls it lateral thinking, and it you can master the trick of
it, you'll solve impossible problems without effort, at least sometimes.
The main thing you have to do is to think about what assumptions you are
making that aren't actually a condition of the problem, then relax those
assumptions.

I started off using the classic "War Games" approach.  First I had to
find out a few telephone numbers.  Fortunately, the Baker Street
Irregulars were still going strong, and this motley collection of
urchins was exactly what I needed.  I sent them down to the bank, armed
with clipboards and biros, to do some "market research".  At nine
o'clock and at five they were there, stopping anyone who looked like
they might work in the bank, and asking for their name, some questions
about ice cream, and their phone number "so that we can notify you if
you win the prize, a gallon of ice cream".

A couple of days of this netted me about a thousand numbers.  I set the
computer up to dial its way down the list but hang up after two rings.
But if the phone was answered within two rings, it would wait and note
whether it heard the characteristic Guardian tones at the other end.  I
repeated this each day; the thousand calls took about a minute each, a
total of 16 hours, so I simply left it to cycle while I thought up
another approach in case this one didn't work.

There was a good chance that it wouldn't work - I was trying to strike
the time between someone connecting his modem up but before he'd dialled
out.  And I didn't even have the assurance that any of my thousand
numbers was being used for this purpose.  While I left this running, I
tried another approach.

First, I wrote a program that would simulate the action of the Guardian
modem, using software and my digital-to-analogue converter instead of a
complex box of electronics.  This would obviously be far more expensive
than an actual modem, but I didn't want to attract suspicion by buying
their modem, and I was fairly sure that it wouldn't occur to them that a
modem can be emulated in software.

Next, I dialled up the bank.  The remote Guardian dutifully answered at
the other end, and this time, it recognised another Guardian's tones, so
sent a complex and encrypted sequence of signals that presumably
translated to "Who are you?".  Of course, I didn't have any answer to
this, so I sent a break signal, indicating that I'd hung up, even though
I hadn't.  To my surprise, the remote Guardian believed this, and didn't
drop the line.  Well.  I hung on to see what would happen next.  What
happened next was rather nice - someone called in.  It was very faint,
but I could just hear the remote Guardian send an interrogation, the
caller send a response, and then the remote dropped my line and the
caller.  Rats - I really wanted to hear what happened next, and I'm sure
you can guess why - a few more minutes, and the Guardian would have laid
bare its soul.

I sat and thought about all this for a while - it was very tantalising
to be so near.  I needed a stimulant.  "Holmes, could I have some of
your coke?" I asked.  As the impact of the coke fizzed through my body,
I began to experience another, familiar feeling.  "Holmes, could you pop
out and get me a couple of hamburgers, with fries and another bottle of
coke?" I find it hard to be inspired on an empty stomach - the rumbles
distract me.

As I munched my Big Mac, I thought about the situation.  The Guardian
thought that the line was cleared, so its logic said that it didn't need
to hang up;  a reasonable assumption on the part of the firmware
programmer.  But the second caller wasn't faking a hangup, so the
Guardian really was dropping the line, and the firmware program was
dropping all unused lines.  You could see the logic of it, and I began
to see the kind of neat, ordered mind that the firmware programmer had.
So what I needed to do was not to be on an unused line when the remote
Guardian dropped all the unused lines - obviously it wouldn't drop a
line that was in use.  But I couldn't be on an in-use line unless I was
logged in, and I couldn't log in unless I had already cracked this
system.  Ah, but.  I bet it would consider that a caller in the process
of logging in was in-use!

I set this up very carefully, because I didn't want some alert security
guard in the bank wondering why there were so many phone calls - I
wanted it to work first time.  I set up two computers, with two lines
and two fake Guardians.  I also hooked in my tape recorder.  With the
first machine, I dialled up the bank, and sent my break signal.  The
remote Guardian left me connected.  I waited, and waited, until someone
else logged in, then timing it carefully, I logged in with the second
computer.  When the Guardian dropped all the unused lines, it didn't
drop my second call-in, as it thought it was in-use, and I was able to
send a break signal to maintain contact.  The Guardian then dialled the
other caller, and I heard the other caller's Guardian answer, and heard
the exchange of encrypted passwords.  I now had quite a lot of
information, so I logged off to work out what to do next.

I played back my tape.  You could hear the various transactions, but
very faintly.  I tried to analyse the frequencies, but there was too
little signal and too much noise to be able to extract any information.
But one thing came through quite distinctly.  You could hear the little
clicks as the Guardian dialled out to the caller, and I counted them,
which gave me a phone number.  Knowing the phone number of an authorised
caller isn't enough, of course.  That doesn't get you into the bank.
But it was quite an achievement, and even Holmes was impressed.

Holmes asked me if I thought there was any chance of success.  I told
him that it was now excellent, but that a lot remained to be done.  He
pulled out his piped and sucked on it, one of his fouler habits.  "The
trial of our client starts next week - we have to be finished by then."
I explained that this job couldn't be hurried;  it had to move at its
own pace.  Have you ever noticed that when people think that something
is impossible, and you tell them that is is possible but difficult, they
then expect it to be done unreasonably quickly?  I think the cause is,
that if someone doesn't understand a skill, he usually thinks that there
is no skill involved.  Quite often, when I'm asked how I managed to do
something, I look modest and say that I was lucky.  Lucky?  Hah!

End of part 3 - to be concluded

Part 4

I was clearly making progress in breaking into this bank, since I could
fake the Guardian modem that they used, and now I knew the number of an
authorised caller.  But there was no way I could change my phone number
to that number, so when the bank Guardian called back, it wouldn't call
me, it would call someone else.

My next problem was to persuade the Guardian to call me instead of one
of its authorised numbers.  That looked like the strongest part of the
whole security system, and is probably the reason why so many people buy
these call-back modems.  Actually, it is the weakest link, as it
inspires such undue confidence.  I called up a friend of mine who calls
himself "Blue Adept".  He knows the telephone system the way I know the
PC, and it's always a good idea to seek advice from an expert.  I
explained what I needed, and he told me it was even simpler than I'd
thought, and explained how - he also gave me the frequencies I needed.

A lot of people, if they think about the telephone system, imagine huge
banks of relays and things that clunk round to select a line.  Bits of
it are still like that in this country, but in the US, and quite a lot
of the UK, they've brought the system into the 1970s.  It's all
electronic, now.  The switching is done by integrated circuits, the
wires are multiplexed or channelled into fibre optics and microwave
links - in fact, the world telephone network is really just a mammoth
computer.

Down the channels of this computer the data flows, but the same wires
than carry the data, also carry the control signals.  These tell the
telephone network what to do - how to route the calls, charging
information and so on.  It goes down the same wires as the voice call,
and is separated from it simply by the fact of being a special set of
tones.  That's why BT are very down on people using Bell tones or
anything else non-standard in the UK;  it could play havoc with the
telephone network control systems.  One simple example of the tones
controlling the call is when you use touch-tone dialling.

First, I needed to find out the passwords and protocols that would
reassure the Guardian - that was very easy to do.  I called the number
that I'd got by counting the clicks, which put me in touch with the
modem that had called the bank.  Then I sent the interrogation sequence
that I'd recorded from the bank's Guardian, and recorded the affirmative
answer from the machine I'd dialled.

I set up a third computer on a third telephone line - this was getting a
bit complicated.  I called into the bank's Guardian and did the break,
so the system left me on.  Then I waited for a caller.  After an hour or
so, I got one.  I called in with computer number two to stop the
Guardian dropping the line, and listened to the clicks as the Guardian
pulse-dialled to return the call.  Then, from my third computer, I sent
out the tones that told the telephone network to redirect the call to a
new number.

The phone rang.  My mock Guardian answered it, and in response to the
bank's guardian's interrogation, I played back the affirmative answer.
The Guardian connected me to the bank's mainframe, and I breathed a huge
sigh of relief.

Holmes was hopping about in a very agitated manner.  "Oo, Oo", he kept
saying.  I grinned at him.  "Now comes the tricky bit", I said.  His
face fell.  "You mean we're not through yet?", he asked.  "No", I said.
"I doubt if these guys are complete wallies.  The chances that I can
steal money via a remote terminal are approximately zilch." "Then why do
they put on so much security?", he said, "And what's the point of doing
what you've done?".  "One step at a time", I said.

Now I was into the computer, but I only had as much privilege as an
ordinary user, and I needed a lot more than that to actually do
anything.  Mainframes have elaborate security systems even once you are
on, and on most sites, ordinary users are allowed to log on, log off and
do very little in between.  I had a look at my privileges.  Sure enough,
I could read files that were in my account, create files, run a word
processor, and that was it.  What I really wanted was full Sysop
privileges, which would let me do anything, but only a Sysop can grant
Sysop priveleges.  Fortunately, there are several ways round this, but
please don't spread them around too much.  I asked the machine what it
was, and it said it was an IBM 4070, running CMS.  Splendid - a very
common machine, which should give no problems.  I logged off, and
explained the privilege problem to Holmes.  "It's a Catch 22", he said.
"Yes, but there's a cook", I replied.

I know the 4070 middling well, and CMS is a nice OS to use.  There's a
4070 I use regularly;  the nighttime operator gets bored, and I play
chess with him sometimes on his computer.  It means I have access to a
4070 without paying for it, and playing overnight chess isn't as time
consuming as it sounds, because I cheat and run a chess program.
Anyhow, I logged onto "my" 4070, and wrote a little program.  The
program had two purposes.  The main purpose was to use the system
service calls to change my privilege up to sysop level.  I explained
this to Holmes, and he asked me what the point was, since the operating
system wouldn't permit me to run it.  I added the second part of the
program, which was a straight Space Invaders game, using the code for
the one I have on my IBM PC.  I compiled it, checked that it worked,
then downloaded it to my PC.  Then I logged into the bank's computer
using the complicated system I'd set up, and uploaded the Trojan Horse
to the bank;  I called it INVADER.  I sent a message to the Sysop to say
that I'd found this file in my account, and thanking him for giving me a
copy of such a splendid game.  Then I logged off, and went to bed.

Next day, I logged in again, and had a look at my privileges.  Sure
enough, I had full Sysop privilege - I just love these impregnable
security systems;  nobody is ever suspicious, because there's no way any
outsider could get in, right?

So there I was, a Sysop, in full control of this 4070.  You would have
thought it was simple now to do anything I wanted, but I'm afraid not.
Banks really are paranoid about security.  Even computer personnel can
be dishonest, you know.  A Sysop sure enough has full control of the
computer, but the accounts data is kept on half-inch open reel nine
track magnetic tape, and is only loaded up when needed, or on request.
And these requests are logged by the tape operator, and he would
certainly notice something as unusual as a tape request in the middle of
the night.  So the computer was mine, but not the accounts data.

I asked Holmes to set up an account.  He loves dressing up - he calls it
disguises and I have to pretend not to recognise him.  He really went to
town on this one, though.  He does old ladies really well, and his
ruffian looks very convincing, but I'd never seen his Hare Krishna
before.

While he was out opening up an account, I went onto the computer again.
Computers are fairly complex things, so no-one is ever surprised when
they break down.  I wrote a disk device driver to replace the IBM one,
except that every time the disk was accessed, it reported a fault.  Then
I logged the fault, and logged the fact that the IBM engineer had been
called, and logged the FRN (fault report number).  By the time Holmes
returned clashing his cymbals, I was ready to go, and he had to do a
quick change into a white button-down shirt and a blue suit.  IBMers
always go everywhere in pairs, and it would have looked very odd if the
engineer had not been accompanied by the usual salesman.

We turned up at the bank, and told them we'd come to fix the fault,
quoting the FRN.  I flashed my IBM badge (do *you* know what an IBM
badge is supposed to look like?) and we were shown in to the machine
room.  I set to work, while Holmes kept up a constant stream of sales
patter.  Actually, he was rather good.  In fact, I started to worry that
he was too good - if they actually bought something from him, we could
have problems.  I took the cover off the disk drive and messed around
inside.  I mounted my diagnostics magnetic tape, while Holmes' chatter
distraced them from noticing the write enable ring on the tape.  I
copied the tape that they had mounted, disabled my disk device driver,
and told them that it was fixed, but that I'd be back the next day to
check it.

On the way out, we were checked by the security guards - I suppose they
wanted to be sure that we hadn't helped ourselves to a pile of money.
They'd counted one tape in, so they were quite happy to count one tape
out again.

When we got back to our digs, I mounted the tape on my tape drive, and
had a look at it.  Oh dear.  A tape holds an awful lot of data - I
seemed to have got a complete copy of their ledgers.  I worked all night
on that data, taking a bit from here and a bit from there, and putting
most of it into the Hare Krishna account.  I set up several new accounts
(but pre-dated them to various dates) in the names of the main board of
directors, and put the rest of the money into those.

The next day, we went down to the bank.  I gave my FRN again, flashed my
badge, and while messing around with the disk, copied my "diagnostics"
tape over the tape of accounts data.  I tidied up, dragged Holmes away
from the sale that he'd nearly made, and off we went.  On the way out,
we drew out quite a lot of money from our account, and went back to our
digs.

When we got back, we dumped the cash in the laundry basket, and grinned
at each other.  Holmes spoke first.  "I've had an idea ..."

It's nice here in Rio.  It's warm and sunny, and the natives are
friendly.  Ronnie Biggs gets a bit bumptious sometimes, and some of the
other expats are a bit unsavoury, but I suppose that's part of the price
you pay.  Our client got off, of course;  all the other directors
promised that if he kept his mouth shut, so would they, and no one would
be any the wiser.  And a prosecution was out of the question, otherwise
someone might have asked embarrassing questions about the sudden riches
of all the other directors.




Alan Solomon

The above is a work of fiction, of course.  There is not, and never has
been, a Great Detective by the name of Sherlock Holmes.