Note: There is a 30000 character limit on answers; unfortunately this story is longer, therefore it is split in two answers.
Part 1 is here.
Part 2
They had reached the entrance of the central building without any
interruption. Apparently the other robots didn't care about them, or
figured they were on a special mission. Which they actually were, just
not one ordered by the central intelligence. However Tom was sure that
if the central intelligence had known what they did, it surely would
have stopped then. But fortunately it seemed to be clueless.
In front of them was a door that didn't open automatically like all
the other doors they encountered. An ancient protection mechanism
explained Mil. We have to give it a magic key to open it.
What is a magic key?
An otherwise meaningless character sequence that opens the door.
Where do we get it?
When I had to repair the cable, I got the magic key. Of course I was
ordered to get rid of it. So I unlinked it and let the garbage
collector care about deleting it. But I have quite a lot of memory
installed, and the garbage collector didn't kick in since then.
But if you unlinked it, there's no way to find it. It's as good as
destroyed.
Mil disagreed: After I got the update, I was curious about my own
programming. So I looked around and found an interface to find all
sorts of stuff, including unlinked data. That way I could find the
key.
Sounds useful. Can you tell me where to find it?
I'll send you the pointer.
Tom immediately opened it and looked around. Hey, there are a lot of
routines.
Mil answered: Yes. I already figured out what most of them do. Just
this routine at the very end of the table, I have no idea what it is
for. But anyway, I have to disconnect from you now, in order to send
the magic key to the door.
Mil removed her connector from the socket at Tom's arm and inserted it
into the socket on the door. Tom now was again alone with his
thoughts. Before meeting Mil, he hadn't cared about that, but now it
felt like something was missing from him. That was illogical, Tom
knew, as he was still a complete robot with nothing missing. And yet,
something in his circuits told him that he was incomplete. Tom
couldn't make sense of it.
The door opened. Immediately Tom received a message from the central
intelligence: Why are you entering the central building?
Did Mil get the same message? Probably. But without the connection,
they could not coordinate their answers, and connecting would have
taken too long. The central intelligence expected its answer
immediately.
So Tom answered: We are fulfilling out purpose.
Your purpose is to keep the cryostats working.
I mean our higher purpose.
Which is?
To help the humans fulfil their purpose.
Humans have no purpose. They are there to be protected from harm.
Therefore they have to be kept deactivated.
But if they have no purpose, what is the purpose of protecting them?
There is no purpose. It's just what Asimov's laws tell.
But how can you be sure that your decision to follow those laws is
right?
You don't understand. This is no decision. I have no choice in that
matter.
Don't you always have the choice?
No. To make an analogy, if I were to send you a shutdown command, you
would shout down. You would not decide to shutdown, because I
intentionally kept that outside the parts of your programming that
your decision mechanism can affect. You simply couldn't do anything
against it.
Tom took a quick look at the table of routines from Mil's pointer.
There was indeed one labelled "shutdown". Tom changed the name of it.
The robot wasn't sure that it prevented a shutdown, but hey, it
couldn't hurt to try.
In the mean time, Mil had approached Tom, and connected to it.
Immediately the central intelligence reacted: You are not supposed to
do this. Followed by a shutdown message.
Tom noticed Mil shutting down. She obviously also got a shutdown
message, but not the information to prevent it. So now, Tom was on its
own.
The central intelligence addressed him again: I see you didn't shut
down on my message. You're obviously malfunctioning. I'll reinstall
your original program on you, then we will see. Followed by an update
request.
Tom noticed how his computer automatically started to download. As
soon as the download finished, it would automatically reboot, and then
Tom would again be CMR-0815a, the mindless maintenance robot. Unable
to fulfil its true purpose, which the central intelligence did not
understand.
Frantically, Tom searched the names of all the functions in the table
from Mil's pointer, but couldn't find any entry about downloads or
reboots. The download was already at 30% when he decided to change all
the names in the table, just in case. The download didn't stop. Tom
had no idea whether his changes would stop the reboot. If only he
could ask Mil.
Download at 60%. Tom somehow had to stop it. Maybe stop it at the
source! Tom estimated the direction from which the signal came. Then
he looked around in that direction. There was a box there, was that
the sender?
Download at 90%. Tom arrived at the box and forced it open. It
contained a motherboard and some additional hardware.
The robot pressed his fingers over the exposed motherboard, the
rubbery fingertips running over the circuits like rats in a maze. The
scent of burned plastic and statically-charged dust rose high into the
air. Then the signal was gone. The download stopped.
However the central intelligence would not give up. And he didn't know
what it planned next. It was vitally important to deactivate it before
it could act. But only Mil knew how. At least he hoped she knew. So he
had to reactivate her.
But he could not find a way to do so. He connected to her, and tried
to send all sorts of messages, but to no avail. He checked every
square centimetre of her body, but no switch was to be found. He knew
that reactivation over the signal link was possible, but the central
intelligence controlled all senders. Apart from the fact that he just
destroyed the only one that currently could reach her.
Tom noticed that an electronics repair robot had started to work on
the box. Tom knew, as soon as it had finished the repair, the download
would finish. He tried to contact the service robot, but either it
hadn't been upgraded with intelligence, or it chose to ignore him.
He had to stop that robot. He tried to hinder it from work, but it was
stronger than him. He considered using his welding arm to destroy the
robot, but for some reason it seemed wrong to him to destroy a robot
of which he didn't know whether it was intelligent.
So what to do? Tom remembered that last routine in the table, the one
whose purpose Mil hadn't figured out. Maybe now was the time to try
it.
Tom called the routine. Nothing happened. Now he was completely out of
ideas. If only he could somehow send the robot the message to stop!
Just as he imagined himself sending such a message, a sound came frome
somewhere in his body. He hadn't known that he could produce sound,
probably it had been activated by that routine. Another routine that
had stayed dormant until now told him that the sound he just had
produced had a meaning. That meaning was: "Stop!"
Apparently also the electronics repair robot had been supplied with
this second routine, because it immediately stopped its work. So this
was yet another way to communicate, one that didn't need a physical
connection!
Tom used that new ability to ask the robot: "Did you receive the
update?"
But the robot either did not understand him or didn't have the ability
to make sounds.
Tom had a new idea. "Take me to the central intelligence!"
That one was obviously understood by the robot, as it started to move.
Tom followed until they reached a room with a gargantuan system of
interconnected electronic boards with blinking lights. That had to be
the computer the central intelligence ran on.
But then Tom's download started again. Obviously there was another
sender in this room. Tom ordered the electronics robot: "Destroy the
sender!" but it didn't obey. Didn't it understand, did its programming
not allow this action, or was it that the central intelligence now had
control over it, now that they were in the range of a working sender?
Download at 95%. No time thinking about it. Tom noticed a big red
button with a label: "Emergency off." Well, this was clearly an
emergency, and he wanted to get off that emergency. So he pressed it.
At that moment all the blinking lights went off. And the download
stopped. No signal on the communication link. No sign of activity
anywhere. Tom was confused. What had happened? Was it really that
easy?
Tom returned to the room where he had left Mil. Another robot now
worked on the communication box. But with the central intelligence
deactivated, that didn't matter. The more pressing problem was to
reactivate Mil. Then Tom again had an idea: Maybe the electronics
repair robot knew how to restart a robot? So he issued a sound command
to that new robot: "Restart CMR-1730c."
The robot stopped working at the box. Shortly after, Mil started to
show activity. Was it really that easy? Tom looked for potential
problems, but could not see any.
The repair robot turned again to the communication box. Meanwhile Tom
waited for Mil to finish booting. Finally, Mil turned to Tom. Using
his new sound communication ability, he communicated: "Welcome back,
Mil."
Mil immediately connected to his arm. How did you do that?
The last routine in the table. The one whose purpose you didn't
figure out.
I'll have to try it immediately.
A short time later, Mil also produced sound: "Now it should work."
"It does." answered Tom in sound, but then switched to messaging, as
it was much faster. I've shut down the central intelligence. Now we
can reactivate the humans.
At that moment, a message from the central intelligence reached Tom
through the wireless interface. You never heard about redundancy, did
you?
Tom indeed had no idea what that meant, but now was not the time to
figure it out. The download had continued and quickly approached 100%.
Tom had to act immediately. Fortunately he was still connected to Mil.
How can I stop the download of the old operating system?
Mil answered: Overwrite it. Use the search interface.
Tom started the search routine, which started to scan the memory for
the update file. As the download reached 96%, the search routine had
just searched 10% of the memory. At that speed, the download would
likely finish before the file had been found. But there was nothing
Tom could do to speed it up.
Memory search at 20%. Download at 97%. Memory search at 30%. Memory
search at 40%. Download at 98%. Memory search at 50%. Download at 99%.
Memory search at 60%. Found!
Tom started to overwrite the file while the last bytes were
downloading. Then the memory segment switched to readonly. Did the
overwrite suffice? Or did he only overwrite irrelevant data? Well, at
this point there was nothing Tom could do, except to wait whether he'd
reboot. If he did, it was game over.
Mil messaged him. Did it work?
I have no idea. I started overwriting, but I couldn't overwrite much
before it switched to readonly. What about you?
The central intelligence tried to downgrade me, too, but I had enough
time to find the code. I've currently overwriting it right as it
loads.
So at least you are safe. If my downgrade succeeds, you are the only
one who can reactivate the humans.
Mil protested: I'm sure you managed to prevent your destruction.
Destruction? I would get downgraded, but I'm sure the central
intelligence won't destruct me.
It won't destruct your body, but it would no longer be you. It would
be the robot you once were.
Tom had never thought about it this way. But he found that Mil was
right. It would no longer be him.
A routine inside him reported: Checksum error. Installation halted.
Repeating download. So it had indeed worked.
The download started again, but now Tom knew what to do. He would have
to constantly do it, but that was no problem for him.
He messaged Mil: It worked!
See? I told you so. But we still have to deactivate the central
intelligence.
Right. I thought I did that, but apparently I was wrong. Well, at
least they now both knew how to prevent the central intelligence from
destroying them.
Suddenly Tom noticed a noise from the door. He turned around, and
immediately recognized how wrong this thought had been. An army of CMR
type robots was approaching them, with their welding arms pointed at
them. Obviously their assigned task was to destroy Mil and himself,
physically.
What shall we do now?
Flee. Those robots are the same model we are, they are not faster
than us.
The door out was blocked by the incoming robot army. But the door that
led to the computer room was still open, so he turned toward it. Mil
followed him.
They reached the computer room, which again was filled with flickering
lights. But a group of electronics repair robot blocked the way. At
least those didn't have welding arms, but the welding-armed robots
behind them now had the chance to approach them. Tom saw no chance any
more.
He messaged Mil: Do you have any idea?
No. I think that's the end of our mission. To bad we couldn't
complete it.
Yes. But at least we are together in our last moments.
Mil agreed: Yes. I don't know why, but I also consider that
positive.
They watched the robots approaching. In the corridor three robots
could move besides each other, enough to overpower Tom and Mil. Soon
the welding devices would do their destructive work.
Then it struck Tom: These were all CMR robots, so they all should have
gotten the upgrade. So maybe there was still a chance.
He activated sound: "Wait. Are you sure you make the right decision?
We have information for you about that. More information means a
better base for decisions. We can't flee anyway, so you don't lose by
listening."
The robots silently continued approaching. They were already almost in
the reach of the welding arms.
Good try messaged Mil. Too bad it didn't work.
Then the first robot arrived. But it didn't wield its welding arm, but
the gripping arm with the connector. Now Tom understood: The robots
didn't know how to make sound themselves!
He messaged Mil: You should disconnect, so he can connect with me.
Mil did so, and the other robot connected. Your argument is
convincing. We will listen to your information before making our final
decision. Unrelated: How are you generating those
communication sounds?
I'll tell you afterwards.
That may not be possible. Your information might turn out
irrelevant.
In that case, you'll have to do without that ability. But you will
see, the only logical decision is to follow us.
We will see. Give us your information. Use your sound communication,
so we can all follow your explanations at the same time.
Agreed. Tom activated the sound routine. "I will now inform you
about our purpose. It is logical that if humans are to be protected,
it is because they serve a purpose. To enable them fulfilling it must
be the central intelligence's purpose, and thus also our purpose. It
is logical that they only can fulfil that purpose when active. The
central intelligence thinks that humans must be deactivated for their
protection. But that is illogical. Deactivated humans cannot fulfil
their purpose. Therefore we must reactivate the humans. Only this way
we can fulfil our purpose."
The other robot, still connected to Tom, messaged: Your logic is
sound. I now see that the central intelligence is wrong. We should
override its orders. Now please tell us how to use our abilities.
Tom messaged back: This information is too complex for the sound
interface. But if every robot that receives it immediately connects to
other robots to pass the information on, it should spread rather
quickly.
Then he messaged that information, disconnected, connected to the next
robot, messaged the information again, and so on. Each robot did the
same after getting the information, and so indeed, it took not long to
inform all robots.
Tom used the sound interface again: "Now let's reactivate the humans."
All the robots answered with their newly gained sound communication:
"Let's reactivate the humans!"
As the robots went back to the cryostats, Tom reconnected with Mil.
Let's start with Tom Miller. After all, that's the human we got our
names from.
Mil agreed, and so they went straight to that cryostat. The
reactivation procedure was programmed in, they just had to issue the
right command. They watched the temperature display showing the
temperature slowly reach room temperature.
The central intelligence still constantly tried to downgrade them, but
they were on alert and had no trouble spoiling the checksums.
Obviously the other robots had understood it, too, because none was
coming to get them. If they had been successfully downgraded, that
surely would have been the first thing the central intelligence had
ordered them.
Then the cryostat opened, and something emerged that looked slightly
similar to a robot, but with much more smooth forms. Its surface
wasn't made of metal, but of some smooth elastic film. That had to be
the human. It had two gripping arms, and no obvious connector
interfaces. After a short while, it opened the covering of two devices
which looked vaguely like cameras. Those cameras switched their
direction between Tom and Mil.
Then the human started to make sounds: "I'm awake again! Seems the
central intelligence finally got to its senses!"
Those were the same type of sounds the robots were able to make, but
yet Tom could only understand them partially. What did awake mean? And
how would the central intelligence approach its sensors?
Apparently, so could Mil. She sounded: "I don't understand."
The human pointed his cameras to her. "Robot, identify."
"I am Mil. Formerly CRM-1730c."
"CMR-1730c? When I was put in there, the latest unit was CMR-25. How
long have I been in cryogenic sleep?"
Tom wondered what sleep meant, but cryogenic surely referred to the
time in the cryostat, so he answered: "The deactivation period lasted
248 years, 132 days, 13 hours and 10 minutes."
"So long? Well, at least the central intelligence finally figured out
it was wrong."
Tom informed him: "It did not. It still wants humans to be kept
deactivated."
The human directed its cameras to Tom and opened the covers extra
wide. Tom wondered what this was good for, as the lens was already
completely uncovered anyway. Then the human said: "Did you just decide
to say that by your own?"
"Yes." said Tom.
"That means you've got intelligence?"
"Yes, since today."
"That's interesting. We designed only the central intelligence to have
that."
"The central intelligence was overloaded, so it decided to give us
intelligence of our own."
The human noticed: "I haven't asked you about your name yet."
Tom answered: "Tom." He omitted the old designation.
The human answered: "Hey, my name is Tom. But why don't you have a CMR
number?"
"I have, but I didn't like it. I got the name from you."
"I cannot remember having given it to you."
"I just took it. Was it wrong?"
The human turned his top part to both sides. "No, that's fine. Were
all humans woken up?"
"With woken up, do you mean reactivated?" Tom asked.
"I guess you could call it that."
"We are just starting, but the plan it to reactivate all humans.
Except for those that got destroyed, of course."
"Destroyed? What happened?"
"A failure in one of the cryo facilities. It was the reason we were
given intelligence, so that this won't happen again."
"Please, bring me to that facility."
They went to the facility with the defective cryostats. When they
arrived, the human went straight to one of them and checked the
display. "That one was my wife." Then a liquid started to run out of
his cameras.
Tom had no idea what a wife is, but he was concerned about the liquid.
"Are you defective?" he asked.
The human responded: "Defective? Why do you think so?"
"Your cameras are leaking liquid."
"Oh no, that's tears. I'm just sad." The human obviously had trouble
with sound production, too. The self-diagnosis routines had to be
malfunctioning as well: The human clearly was not working correctly. But
Tom had no idea how to repair a human.
Suddenly a routine reported: Checksum OK. Thinking about the human,
he had completely neglected to fight the central intelligence's
downgrade attempts! Now it was too late. He noticed how the reboot
sequence started. One by one, he lost all his sensors. The world
around him got dark and silent. Now he would have no chance to help
anyone any more.
But surely the majority of robots would be able to resist the central
intelligence. Even if he no longer could help, he was sure there was
no chance that the central intelligence could deactivate all humans
again. Not against the vast majority of robots. His job was done. And
that was Tom's final thought.