Remove ads

A Sword Without a Hilt: A Song of Ice and Fire/D&D 3.5 Crossover

Reader mode
Snowfire said:
Work out how to create a stable electrical power source from Dragon's Fury (@Azel railgun project) power units
Click to expand...
Click to shrink...
I'm never getting rid if that idea again, will I?

My next engineering goal would be trains and steam power generation btw. It's something I want to put the SD university on.


As for downloading the internet, it bears keeping in mind that most of it is utterly useless to Viserys. Why would he want stuff like Twitter or Facebook?

Downloading a few good libraries and online published research papers would be more then enough.

Cue Viserys learning about the scientific publishing mafia. Cue undiluted disgust and hatred.
 
egoo said:
...What's the problem with it, again?
:???:

I genuinely liked the theme, the math, and the general design of the thing :/
Click to expand...
Click to shrink...
It also needs a good deal too much science to come up with. Like... magnet field theory, induction, electricity and so on.
Since then, I've tried very hard to keep to sensible progressions of existing technology and capabilities while the railgun skips a lot of necessary precursor knowledge.

So far our tech advances are:
- built a magic powered oven that doesn't need an elemental bound to it and produces much more heat at the cost of being much more expensive and needing a higher-level caster
- from cheap thermal power, we improved on ancient steam toys to build something that can kill people really damn hard
- ignored the common wisdom of building flying ships by shackling elementals into it like beasts of burden, instead opting to adjust a high-level spell to simulate a planar effect, thus giving us superior engines at the expense of, yet again, being hideously more expensive and needing significantly better casters to build them

And... that's about it. Just a bit of lateral thinking and breaking with traditional ways of doing things.

From here on, I would like to go further in the steam direction, as we are now entering the point where we can just assign muggles to expand on the tech, instead using our casters to do more fundamental research.

I'm especially eager to get Serpentstone from the ruins in Sothoryos, so that we can look into logic engines.
 
Azel said:
It also needs a good deal too much science to come up with. Like... magnet field theory, induction, electricity and so on.
Since then, I've tried very hard to keep to sensible progressions of existing technology and capabilities while the railgun skips a lot of necessary precursor knowledge.

So far our tech advances are:
- built a magic powered oven that doesn't need an elemental bound to it and produces much more heat at the cost of being much more expensive and needing a higher-level caster
- from cheap thermal power, we improved on ancient steam toys to build something that can kill people really damn hard
- ignored the common wisdom of building flying ships by shackling elementals into it like beasts of burden, instead opting to adjust a high-level spell to simulate a planar effect, thus giving us superior engines at the expense of, yet again, being hideously more expensive and needing significantly better casters to build them

And... that's about it. Just a bit of lateral thinking and breaking with traditional ways of doing things.

From here on, I would like to go further in the steam direction, as we are now entering the point where we can just assign muggles to expand on the tech, instead using our casters to do more fundamental research.

I'm especially eager to get Serpentstone from the ruins in Sothoryos, so that we can look into logic engines.
Click to expand...
Click to shrink...
Well, a railgun basically is "make something go very fast and hit". Magitech solution? Look into telekinesis and, in general, force spells. Or delayed blast fireball & pressure chamber filled with water.
For the internet thingie - fae might pay good for facebook?
 
MTB said:
Well, a railgun basically is "make something go very fast and hit". Magitech solution? Look into telekinesis and, in general, force spells. Or delayed blast fireball & pressure chamber filled with water.
For the internet thingie - fae might pay good for facebook?
Click to expand...
Click to shrink...
I've looked into alternative solutions, including a gravity gun using portable holes, but they are too expensive to be practical.

In regards to more firepower, my current idea is to use the plinth from Valyria and VS to figure out a way to build a genuine tele-force-gun like Tesla envisioned. Basically building an item that can fire a cranked up Magic Missile.
 
Vote closed. Awards and joust next then the proposal.
Adhoc vote count started by DragonParadox on Mar 15, 2019 at 5:55 AM, finished with 209 posts and 16 votes.
Click to expand...
 
Azel said:
I'm never getting rid if that idea again, will I?

My next engineering goal would be trains and steam power generation btw. It's something I want to put the SD university on.

As for downloading the internet, it bears keeping in mind that most of it is utterly useless to Viserys. Why would he want stuff like Twitter or Facebook?

Downloading a few good libraries and online published research papers would be more then enough.

Cue Viserys learning about the scientific publishing mafia. Cue undiluted disgust and hatred.
Click to expand...
Click to shrink...
I think trains would be quite feasible. We've already developed a lot of the magitechnology and concepts that would be necessary to make it work.

And the Imperial road network, from the very beginning, has been built with the space for rails to be added.
 
Goldfish said:
I think trains would be quite feasible. We've already developed a lot of the magitechnology and concepts that would be necessary to make it work.

And the Imperial road network, from the very beginning, has been built with the space for rails to be added.
Click to expand...
Click to shrink...
Actually, I've designed them with full rails already, to be used by horse drawn carriages. The only thing we need is adding stations and switches at the major intersections to direct a train.

The only stumbling block is that they will be prodigiously expensive, since they need a onboard power-plant for 2,000 IM, plus controls, which I would price at 2,500 IM.

So we would need... 100 tons of steel, 2,000 IM and 500 IM of enchanting per locomotive.
Well. And another good batch of steel to make the actual cars of course.
 
Azel said:
Actually, I've designed them with full rails already, to be used by horse drawn carriages. The only thing we need is adding stations and switches at the major intersections to direct a train.

The only stumbling block is that they will be prodigiously expensive, since they need a onboard power-plant for 2,000 IM, plus controls, which I would price at 2,500 IM.

So we would need... 100 tons of steel, 2,000 IM and 500 IM of enchanting per locomotive.
Well. And another good batch of steel to make the actual cars of course.
Click to expand...
Click to shrink...
The locomotives themselves should require the great majority of the enchanting. The cars would need some minor work, but that's hardly worth mentioning.

We really could do steam powered trains with a minimum of fuss. We don't need to worry about fuel, adding water to the boiler, inferior materials or manufacturing techniques leading to boiler explosions, etc.
 
bigbow said:
Ten years is kind of wildly optimistic for hard AI/singularity though (or pessimistic, depending on how you see the whole thing). 2044 seems like the most reputable estimate in my mind, though many bump it up by 20 or 30 years (no way).
Click to expand...
Click to shrink...

Recent articles on the state of neural networks disagree. For example, the one that was given a passage that had some French in it, and promptly taught itself French. Not particularly good French, but it wasn't told to learn French in the first place. It chose to do so by itself.

We're much, much closer than people think.
 
Goldfish said:
The locomotives themselves should require the great majority of the enchanting. The cars would need some minor work, but that's hardly worth mentioning.

We really could do steam powered trains with a minimum of fuss. We don't need to worry about fuel, adding water to the boiler, inferior materials or manufacturing techniques leading to boiler explosions, etc.
Click to expand...
Click to shrink...
The cars shouldn't need enchanting at all. The only thing really useful there is making sure everything is Hardened to minimize wear and tear.

On that note, the distance SD <-> Tyrosh when going by the roads is roughly 330 km. So if we can get a train going that manages 100 km/h, we could have a express train going the route every 4 hours.
 
Snowfire said:
Recent articles on the state of neural networks disagree. For example, the one that was given a passage that had some French in it, and promptly taught itself French. Not particularly good French, but it wasn't told to learn French in the first place. It chose to do so by itself.

We're much, much closer than people think.
Click to expand...
Click to shrink...
As someone working in the field, I can assure you that there is a lot of hype involved in these bold claims and often only a dubious amount of substance.

In fact, there is a lot of doubt on the validity of many such research results as it turns out that Deep Learning systems are often exceedingly brittle when presented with unexpected input and sometimes don't even deliver correct results, just results that look correct to the observing human.

Especially neural network approaches show themselves good at pattern matching, but abysmal at something close to actual thought.

The corollary is that machines are very much like humans in one particular regard: It's much easier for them to appear smart then to be smart.
 
Azel said:
The cars shouldn't need enchanting at all. The only thing really useful there is making sure everything is Hardened to minimize wear and tear.

On that note, the distance SD <-> Tyrosh when going by the roads is roughly 330 km. So if we can get a train going that manages 100 km/h, we could have a express train going the route every 4 hours.
Click to expand...
Click to shrink...
I was thinking of minor stuff to make improve the speed while reducing wear and tear on the cars and the tracks; Grease-based lubrication, Unseen Servant-based brakemen, Continual Flame-based lighting, etc.
 
Azel said:
The cars shouldn't need enchanting at all. The only thing really useful there is making sure everything is Hardened to minimize wear and tear.

On that note, the distance SD <-> Tyrosh when going by the roads is roughly 330 km. So if we can get a train going that manages 100 km/h, we could have a express train going the route every 4 hours.
Click to expand...
Click to shrink...
Would permanancied Animate Objects help? Because that spell is perpetuum mobile in a can. Or a cheaper version for the same effect? With a command word for start/stop the movement?
 
Snowfire said:
Recent articles on the state of neural networks disagree. For example, the one that was given a passage that had some French in it, and promptly taught itself French. Not particularly good French, but it wasn't told to learn French in the first place. It chose to do so by itself.

We're much, much closer than people think.
Click to expand...
Click to shrink...
Azel said:
As someone working in the field, I can assure you that there is a lot of hype involved in these bold claims and often only a dubious amount of substance.

In fact, there is a lot of doubt on the validity of many such research results as it turns out that Deep Learning systems are often exceedingly brittle when presented with unexpected input and sometimes don't even deliver correct results, just results that look correct to the observing human.

Especially neural network approaches show themselves good at pattern matching, but abysmal at something close to actual thought.

The corollary is that machines are very much like humans in one particular regard: It's much easier for them to appear smart then to be smart.
Click to expand...
Click to shrink...
I read an article last week about a new text generating program. It's obviously not an AI yet, but it's really skilled at producing entire blocks of reasonable text of all sorts with only a few keywords. Some of the examples of its work were pretty funny, but kinda eerie as well.

It'll be interesting to see where that particular avenue of AI research goes.
 
Azel said:
As someone working in the field, I can assure you that there is a lot of hype involved in these bold claims and often only a dubious amount of substance.

In fact, there is a lot of doubt on the validity of many such research results as it turns out that Deep Learning systems are often exceedingly brittle when presented with unexpected input and sometimes don't even deliver correct results, just results that look correct to the observing human.

Especially neural network approaches show themselves good at pattern matching, but abysmal at something close to actual thought.

The corollary is that machines are very much like humans in one particular regard: It's much easier for them to appear smart then to be smart.
Click to expand...
Click to shrink...

I am aware that the current systems have flaws. But even so, the capacity for them to learn and intuit is a lot higher then people outside of the field often think. And, to be fair, I don't think either of us knows what's actually going on in Google's AI development section.

The general argument that I've seen placed against your point on Deep Learning, however, is that at the end of the day a lot of what human thought is is just, well, pattern matching. We're following that progression, and we've gotten a lot further, faster, than I think some of those involved in the dev process expected.

At the end of the day, I guess we'll see. But I have been convinced by others in the field, that 'within a decade' is plausible. Maybe not likely, but it's plausible. And that's equal parts awesome and terrifying.
 
Snowfire said:
I am aware that the current systems have flaws. But even so, the capacity for them to learn and intuit is a lot higher then people outside of the field often think. And, to be fair, I don't think either of us knows what's actually going on in Google's AI development section.

The general argument that I've seen placed against your point on Deep Learning, however, is that at the end of the day a lot of what human thought is is just, well, pattern matching. We're following that progression, and we've gotten a lot further, faster, than I think some of those involved in the dev process expected.

At the end of the day, I guess we'll see. But I have been convinced by others in the field, that 'within a decade' is plausible. Maybe not likely, but it's plausible. And that's equal parts awesome and terrifying.
Click to expand...
Click to shrink...
If our first AI goes batshit insane, let's hope for a Deus rather than a Skynet...
 
Goldfish said:
I was thinking of minor stuff to make improve the speed while reducing wear and tear on the cars and the tracks; Grease-based lubrication, Unseen Servant-based brakemen, Continual Flame-based lighting, etc.
Click to expand...
Click to shrink...
Grease based lubrication would be too expensive. I'm using it for the Moonchasers turrets and it's really, really costly. There was a stone in the Underdark that works as if it was constantly under the effect of Grease and if we dig that up, I would definitely use it for these purposes, but right now it would be much more economic to use good old animal fat.

Unseen Servant brakes would be kinda iffy due to low energy output. I would rather use good old hydraulic brakes.

Continual Flame is a given. We are lighting our whole city with it, so no reason not to do the same with train cars.
 
Snowfire said:
I am aware that the current systems have flaws. But even so, the capacity for them to learn and intuit is a lot higher then people outside of the field often think. And, to be fair, I don't think either of us knows what's actually going on in Google's AI development section.

The general argument that I've seen placed against your point on Deep Learning, however, is that at the end of the day a lot of what human thought is is just, well, pattern matching. We're following that progression, and we've gotten a lot further, faster, than I think some of those involved in the dev process expected.

At the end of the day, I guess we'll see. But I have been convinced by others in the field, that 'within a decade' is plausible. Maybe not likely, but it's plausible. And that's equal parts awesome and terrifying.
Click to expand...
Click to shrink...
I'm... less then convinced. The field has definitely made some great strides in recent years and we are definitely making progress towards the goal, but within a decade is much too rosy a prediction in my opinion. With emerging evidence that our current models are fundamentally flawed, we need to rethink a lot of things. It's true that a lot of human cognition is basic pattern matching and we are getting there to teach that part to machines, but what is still missing is the glue that keeps it together. The pieces that turn disparate patterns into abstract models. We thought we could do that with Deep Learning and neural networks, but it's here that we are realizing that it's still just pattern matching we've made and that it becomes very brittle when scaled up like that.

The other thing is that the tech sector is struggling in regards to innovation since the past decade. It has become a lot of smoke and mirrors, with much effort being poured into making dumb things look very smart and fancy for the marketing and sales divisions to turn it into money. One spectacular example is the whole Blockchain hype of last year. They hyped up a technology with very specific applications and capabilities as something that would change the world forever and solve all our problems. The actual results have been... less then impressive. A lot of money and especially a lot of good minds were wasted on chasing a buzzword that very few people actually understood, but everyone clamored to have.

At it's core, I doubt that our current socio-economic environment is conductive to promote the true breakthroughs we would need.
 
Azel said:
Grease based lubrication would be too expensive. I'm using it for the Moonchasers turrets and it's really, really costly. There was a stone in the Underdark that works as if it was constantly under the effect of Grease and if we dig that up, I would definitely use it for these purposes, but right now it would be much more economic to use good old animal fat.

Unseen Servant brakes would be kinda iffy due to low energy output. I would rather use good old hydraulic brakes.

Continual Flame is a given. We are lighting our whole city with it, so no reason not to do the same with train cars.
Click to expand...
Click to shrink...
Grease that can be used on command at 1st level only costs 180 IM. I don't mean to use it to lubricate the internal components of the car, but to be able to apply it to the track itself. Not really necessary, though, if we do this properly.

The Unseen Servant brakemen wouldn't need to exert much force, merely trigger a braking mechanism. Reliable hydraulics are vastly more complicated than we need for this project. Just build each individual car with a mechanical braking mechanism.
 
Time to quibble !

Snowfire said:
Not really likely, to be honest. We're, at present, probably...oof. Ten-twenty years out? And probably on the low end of that. I have friends in Google and, whilst they aren't allowed to talk about the specifics, what they're able to discuss is pretty incredible. And, I'll admit, a little frightening too. For Plaentos to hit Singularity would require massive expansion, development, and most specifically the training of specialists in fields that simply do not exist there right now.

Even with magic, there's a limit to how quickly you can do that. Fabricate and the Creation spells are great for industrial revolution gear, but they can't really keep up with the fine detail necessary for silicon chips. Or most other modern pieces of technology. And you can't just transplant that sort of technological capacity. You need the underlying infrastructure, which just isn't there, and can't exactly be 'magicked' into place, either.
Click to expand...
Click to shrink...
I disagree for three reasons.

1. Planetos doesn't need to be fully industrialised to hit singularity. Singularity is "creation of an artificial intelligence capable of rapidly self-improving and/or self-replicating" and generally presumed to be able to produce material things or at least conduct research and engineering tasks.
Not "post-scarcity society" or anything. Using magic we can get infinite raw materials without needing to mine/collect/handmake them and we can use Polymorph Any Object cheese to make precise machinery, so we can simply have an expensive research center in SD if we want to achieve Singularity.

2. There is huge potential for magic to shorten this process immensely. Not only is it possible to create infinite Awakened Constructs with infinite mental stats (assuming cheese isn't a problem - this is theory crafting) AND infinite skill ranks, but magic allow for great strides in the neurology, psychology, etc necessary to attempt to create sentience. Divination also allows us to skip A LOT of testing, research and false leads. I think that in a motivated and rich D&D setting with access to IRL tech and research, it is achievable within a single lifetime.

3. "Time is an issue" is a ridiculous objection for a game that has literal time travel and temporal acceleration within the rules :D

I am absolutely not going to push for Singularity in ASWAH, but I do think that it is entirely possible within a reasonable timeframe.
 
Goldfish said:
Grease that can be used on command at 1st level only costs 180 IM. I don't mean to use it to lubricate the internal components of the car, but to be able to apply it to the track itself. Not really necessary, though, if we do it right.

The Unseen Servant brakemen wouldn't need to exert much force, merely trigger a braking mechanism. Reliable hydraulics are vastly more complicated than we need for this project. Just build each individual car with a mechanical braking mechanism.
Click to expand...
Click to shrink...
Do not grease the track or wheels. That's a horrific accident in the making. What you want to grease is the axles, not the parts that keep the train where it's supposed to be.

As for the brakes, I did just want to trigger them with hydraulics, but using dead-weight designs to supply the actual force. Unseen Servant could work fine there too. How much would it cost per car?
 
Azel said:
Do not grease the track or wheels. That's a horrific accident in the making. What you want to grease is the axles, not the parts that keep the train where it's supposed to be.

As for the brakes, I did just want to trigger them with hydraulics, but using dead-weight designs to supply the actual force. Unseen Servant could work fine there too. How much would it cost per car?
Click to expand...
Click to shrink...
200 IM per car for a constant effect Unseen Servant brakeman. They could also be tasked with helping to clean the cars between trips.
 
Goldfish said:
200 IM per car for a constant effect Unseen Servant brakeman. They could also be tasked with helping to clean the cars between trips.
Click to expand...
Click to shrink...
So:

Locomotive:
- 2,000 IM (ritual) for the boiler
- 500 IM (enchanting) for the controls
- 900 IM (enchanting) for a decanter to supply water
- 100 tons of steel

Cars:
- 30 tons of steel
- 200 IM (enchanting) for an Unseen Servant

Then I would suggest the "average" train to be the locomotive, 8 cargo wagon transporting 100 tons of goods each and 2 passenger wagons for roughly 50 persons each.
That would come out to 400 tons of steel, 3,400 IM enchanting and 2,000 IM ritual cost for the boiler.
 
Reader mode
You must login or register to reply here.
Remove ads
Back
Top