Fuzzy logic.

[The Rock]

I was just thinking about how fuzzy logic could fit into Star Trek, from the very beginning with The Original Series all the way up to the newest shows like Discovery, Strange New Worlds, Lower Decks, and Picard. It got me wondering how this kind of thinking could change the way the tech and characters work across the whole franchise.

If you haven’t heard of fuzzy logic before, it’s basically a way of thinking where things aren’t just yes or no but can be kind of, maybe, or probably. It’s all about dealing with shades of gray instead of things being completely true or completely false. That feels a lot more like how decisions actually happen in real life, especially in tricky situations.

Imagine if the computer on the original Enterprise could say, “This nebula is probably safe,” instead of just safe or unsafe. Or if Spock could say that a plan is mostly logical instead of acting like logic is always perfect or absolute. Then jump forward to The Next Generation or Deep Space Nine, where Data or Odo could use fuzzy logic to weigh incomplete information and make decisions based on probabilities and gut feelings instead of just yes or no answers. It would make them seem more human and relatable when they face tough choices.

Think about Voyager or Enterprise. When the crew is out in unexplored space dealing with unknown species or risky technology, fuzzy logic could help them make better guesses instead of needing perfect data that they rarely have. It would fit perfectly with the uncertainty and improvisation those shows highlight.

Now fast forward to the newer shows like Discovery, Strange New Worlds, Lower Decks, and Picard. Fuzzy logic could make the ship’s computers and AI like Zora or the La Sirena’s systems feel a lot more human by giving them the ability to deal with uncertainty in a more natural way. It could explain how Stamets or Seven of Nine make tough decisions based on incomplete data or unpredictable alien tech. Lower Decks especially could have some fun with that by showing how fuzzy logic makes their transporter or ship systems sometimes only mostly work, which could add to the humor and tension.

What do you all think? Would fuzzy logic make Star Trek’s technology and AI feel more real and grounded across the series? Or would it take away some of that clear-cut Starfleet logic and idealism we love? Would it add more excitement and unpredictability or just make things more complicated?

 

[knightgrace]

Okay, I have studied fuzzy logic since the early 1990s. And the most shocking thing of all, is that I am still not a politician.
I know extremely shocking.
What fuzzy logic does is to multiply the truth value by some number from zero to one.
By doing this, it affectively, decreases the number of 'If-then-else' statements to describe a given production system. Or increases the power of a single 'If-then-else' statement by, typically a factor of ten.
For a production system(original called "expert system "), one requires a minimum of about 800 rules, this means that a fuzzy logic system is the equivalent to an 8000 rule based system, typically. Do keep in mind that the ten fold improvement is typical, not fixed.

An example: say you want a computer controlled mile long train. In a 'crisp' logic system you have to have a rule for each unit of distance to come to a full stop. In a a fuzzy logic system the rule would be, " begin at one mile from the station light braking, at one half of a mile began moderate braking, at 100 feet begin heavy braking. "

The fuzzy logic computer, will scale to a continuous increase in braking, in stead of stair step.

Please read 'Fuzzy Future' by Bart Kesko( not too sure about the author's name ( it has been over thirty years after all. )

The problem with production systems is that they are brittle. Any, even in the slightest, and you get garbage.

So what replaced them? Or rather what was added to the logic? Answer: statistical logic. What did this do? Look up Dragon Dictate..
As of 1999, it permitted speaker independent users, with better than 99% reliability. But it wasn't perfect. Keep in mind that the original voice understanding programming was very speaker dependent, and very limited in vocabulary. But statistical logic had its limitations. Then came deep learning neural networks in 2014. This basically perfected voice understanding. It is in your phone now, and has been, for quite awhile. But actually identity based processing?

Just look at ChatGPT3. After fifteen or so minutes, it wants to kill you....for real.
Why?
The programmers wanted absolutely no biases at all, so it doesn't know right from wrong. This is deep.
Very deep.
Think about it.
The Ten Commandments ARE a very great biased system. For good. But good is a problem, for some.

 

[The Rock]

Okay, I have studied fuzzy logic since the early 1990s. And the most shocking thing of all, is that I am still not a politician.
I know extremely shocking.
What fuzzy logic does is to multiply the truth value by some number from zero to one.
By doing this, it affectively, decreases the number of 'If-then-else' statements to describe a given production system. Or increases the power of a single 'If-then-else' statement by, typically a factor of ten.
For a production system(original called "expert system "), one requires a minimum of about 800 rules, this means that a fuzzy logic system is the equivalent to an 8000 rule based system, typically. Do keep in mind that the ten fold improvement is typical, not fixed.

An example: say you want a computer controlled mile long train. In a 'crisp' logic system you have to have a rule for each unit of distance to come to a full stop. In a a fuzzy logic system the rule would be, " begin at one mile from the station light braking, at one half of a mile began moderate braking, at 100 feet begin heavy braking. "

The fuzzy logic computer, will scale to a continuous increase in braking, in stead of stair step.

Please read 'Fuzzy Future' by Bart Kesko( not too sure about the author's name ( it has been over thirty years after all. )

The problem with production systems is that they are brittle. Any, even in the slightest, and you get garbage.

So what replaced them? Or rather what was added to the logic? Answer: statistical logic. What did this do? Look up Dragon Dictate..
As of 1999, it permitted speaker independent users, with better than 99% reliability. But it wasn't perfect. Keep in mind that the original voice understanding programming was very speaker dependent, and very limited in vocabulary. But statistical logic had its limitations. Then came deep learning neural networks in 2014. This basically perfected voice understanding. It is in your phone now, and has been, for quite awhile. But actually identity based processing?

Just look at ChatGPT3. After fifteen or so minutes, it wants to kill you....for real.
Why?
The programmers wanted absolutely no biases at all, so it doesn't know right from wrong. This is deep.
Very deep.
Think about it.
The Ten Commandments ARE a very great biased system. For good. But good is a problem, for some.

Man that was a really solid explanation. Thanks for breaking it down like that. It made fuzzy logic click for me in a way I hadn’t thought about before. The train example especially helped. Fuzzy logic really does feel like using a dimmer switch instead of just flipping things on or off which makes a lot more sense when you think about how things work in real life. Stuff isn’t usually all or nothing. There’s always some middle ground.

And those old systems you mentioned sound super fragile. It’s crazy to think about how they could totally break if just one little thing went wrong. The way statistical models came in and helped smooth things out and then deep learning took it even further is just wild. Voice recognition is such a good example going from barely working for most people to now being built into basically every phone and smart device. That’s some next-level progress right there.

Your point about bias is also really interesting. On paper getting rid of all bias sounds like the perfect goal. But sometimes it actually makes things feel less real or even kind of off. Like trying to be too neutral or perfect can take away what makes decisions human the gut feelings the values the messy stuff that makes us who we are. That definitely gave me a lot to think about.

Thanks for sharing all this. It’s cool to hear from someone with real experience in fuzzy logic and it’s made me see how all these systems have evolved over time. Definitely looking forward to learning more and thinking about how this all applies to stuff like AI and even Star Trek tech.

 

[knightgrace]

I am going to get very geeky here - fair warning issued.


Let us use the Decipher Star Trek the Role Playing Game, instead of FASA...( which will be brought in as I feel the need.)

Okay, Decipher, had a very interesting set of rules, which unlike FASA, as far as I can tell, doesn't include size...

Why is size so important for sensors?

Because there is a problem with binary logic, or crisp logic. It is quite simply yes/no, on/off, true/false... using Decipher Game ideas as a starting point, a human being emits energy while being alive. Heat and electromagnetic energy. Humans, in good health emit heat energy primarily at 37°c...with temperature variations according to what part of the body is emitting the heat energy, the effectivity of the sensors for detecting through the ambient levels of heat...that is if the ambient temperature is at 37°c, detection is going to be a bit more difficult. At an ambient temperature of 0°c, a human being is going to stick out quite clearly...

But, here is the problem: a 64 bit word length central processing unit, is only taking in a very small amount of data, per unit of time. Meaning that the more data that the hand device (Hand Scanner (Star Trek:Enterprise), Tricorder (Star Trek: TOS; TNG; DS9; VOY, SNW, DISCO, and so on)) the longer the processing time is.
But, what exactly ate you looking at? Exactly how do you make or more exactly does the device do it? The answer is; templates. But templates are required for each range setting, point blank, short, medium, and long. But crisp logic only handles yes/no...such that there ate only two templates per range address. Large and small, as determined by the range to the center of mass. Any attempt to have more than two increases the processing time. But what is a human? Templates are supplied, but...

Humans can be Large, small, fat, thin, different colors, depending upon the ambient lighting, male/female. Too exacting specifications, and you are going to miss the entire range of humanity. Fuzzy logic to a certain extent can do a better job than crisp logic. But as has been found out, statistical logic, does better. But you require larger and larger sample numbers to handle the requirements. Deep Learning networks, will respond better, but then you have another problem. Power.
Power? Exactly just how much power does it take to run the CPU of a Hand Scanner/Tricorder? The next question is just how much power does it take to run the active sensors? Remember that active sensors transmit a pulse of energy, and the farther out that you have determined that individual pulse, has to go, the more power it takes.

Now about size. Using a Field mouse versus a human male at two meters versus 10.5 centimeters...remember that a field mouse is a mammal, and therefore emits heat...in other words it gives its presence away. Much to a rattlesnake's liking... Now a third thing. An amoeba...much, much smaller than one centimeter.
I introduce you as well to another mammal, the Polarbear. Larger and heavier than you. But in a polar environment, extremely difficult to detect via heat energy. Why? It's fur, is nearly a perfect insulator. What this means is that you must not rely solely upon heat. For the amoeba, it is so small that detection is almost impossible at ranges beyond one meter. You may detect a multitude of them in a fresh water pond, however, but not by heat. But by spectral analysis. In particular, not simple spectral analysis, nor multispectra analysis ( though if enough are present in a water sample...), nor by hyperspectral analysis ( though the probability is much higher 243 bands of light going from far infrared to ultraviolet versus seven bands of light for multispectral). But transspectral analysis, which I define as being continuous spectrum from far infrared to far ultraviolet. [Side note: if these ghost hunters used even limited spectral analysis...]

 

[Timofnine]

I was just thinking about how fuzzy logic could fit into Star Trek [snip]

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[FormerLurker]

Fuzzy logic is what Spock has after a large glass of Romulan Ale.