Ideas for Anti Robot Technology

[Boomerang]

Originally Posted by lwcamp

It's even easier than that. Make a list of targets. Assign each target a priority. Then do an indexed sort on the list (using heapsort, mergesort, quicksort, or whatever your favorite algorithm is) and take the top one. There will always be a top element in the list (as long as the size of the list is one or more), so there's no need to worry about random numbers.

Luke

At the top of every list I would expect to see do nothing. Better for the robot to risk being destroyed through inaction than to act and ruin important plans.

[Boomerang]

Originally Posted by sir_pudding

You can't expect to be able to write a series of procedural instructions that represent IQ 8 or whatever.

Not sure I understand your point. Isn't that exactly what zone minds do? Give their robots a set of instructions that they cannot deviate from.

[lwcamp]

Originally Posted by Boomerang

At the top of every list I would expect to see do nothing. Better for the robot to risk being destroyed through inaction than to act and ruin important plans.

That doesn't make sense. High priority targets should be attacked. There may be a conditional to compare the priority against a threshold, so that low priority targets are not attacked if the list contains only low priority targets. But your list will have a top element with highest priority (or at least, a priority equal to or higher than any element farther down the list), preventing any analysis paralysis.

Luke

[Johnny1A.2]

Originally Posted by Boomerang

Not sure I understand your point. Isn't that exactly what zone minds do? Give their robots a set of instructions that they cannot deviate from.

Not quite. The NUs work that way, yes.

The AUs, though, learn from experience and their 'minds' are much like human minds in many ways. They are locked into the goals the Z-mind assigns, but how they go about seeking those goals can vary widely, esp. with the smarter AUs.

Look at the example of the Tarantula in the canon RoS that becomes fascinated by human hunters. I doubt seriously the Z-mind would have incorporated such an instruction, it's a 'learned behavior'.

That's what makes AUs so useful, but it also makes them potentially troublesome. I suspect that they get a partial memory wipe or the equivalent on a regular basis.

[sir_pudding]

Originally Posted by Boomerang

Not sure I understand your point. Isn't that exactly what zone minds do? Give their robots a set of instructions that they cannot deviate from.

You aren't going to be able to write a series of procedural instructions that represent IQ 8 or whatever, and attempting to do so is futile. Even in-setting it is almost certainly more sophisticated than a long list of of If...Then statements.

So instead of trying to figure out what an NU would do based on procedures, just treat them like smart animals and if there is any doubt if their program covers something just roll versus IQ or an appropriate skill or default.

[Boomerang]

Originally Posted by sir_pudding

You aren't going to be able to write a series of procedural instructions that represent IQ 8 or whatever, and attempting to do so is futile. Even in-setting it is almost certainly more sophisticated than a long list of of If...Then statements.

So instead of trying to figure out what an NU would do based on procedures, just treat them like smart animals and if there is any doubt if their program covers something just roll versus IQ or an appropriate skill or default.

Okay, that makes more sense.

Probably what I will do in game terms is make IQ based rolls as you suggest and grant an advantage to the PCs in opposed IQ based rolls if they somehow get access to the robot's instruction set and have time to study it thoroughly.

[Boomerang]

Originally Posted by lwcamp

That doesn't make sense. High priority targets should be attacked. There may be a conditional to compare the priority against a threshold, so that low priority targets are not attacked if the list contains only low priority targets. But your list will have a top element with highest priority (or at least, a priority equal to or higher than any element farther down the list), preventing any analysis paralysis.

Luke

Maybe we are working on different assumptions. Zone minds would put a lot of emphasis on the rules for identifying and engaging targets especially high priority ones. It's those situations that are not clear cut where it is preferable for robot to do nothing rather than just start blasting things. Bear in mind that in tricky situations the far more intelligent Overseers can take over. The last thing the zone minds would want is the dumbest robots making the most important decisions.

[sir_pudding]

Most of the zoneminds (with the obvious exceptions) probably don't count which meatbag to kill first as an important decision.

[Boomerang]

Originally Posted by Johnny1A.2

Not quite. The NUs work that way, yes.

The AUs, though, learn from experience and their 'minds' are much like human minds in many ways. They are locked into the goals the Z-mind assigns, but how they go about seeking those goals can vary widely, esp. with the smarter AUs.

Look at the example of the Tarantula in the canon RoS that becomes fascinated by human hunters. I doubt seriously the Z-mind would have incorporated such an instruction, it's a 'learned behavior'.

That's what makes AUs so useful, but it also makes them potentially troublesome. I suspect that they get a partial memory wipe or the equivalent on a regular basis.

I like that approach. For AUs the factory settings will not allow much latitude but as they learn more they can drift away from the original programming, that is until they are reset at their next visit to the robot hotel. Of course sometimes an AU might not return to the hotel for some reason, allowing the possibility of robot PCs down the track.

[Boomerang]

Originally Posted by sir_pudding

Most of the zoneminds (with the obvious exceptions) probably don't count which meatbag to kill first as an important decision.

Fair point.