Decivre wrote:

Arenamontanus wrote:

Ok, now you have frustrated and slightly unhinged people writing your security code or checking your marketing materials. What could possibly go wrong?

I don't know if they would task indentures with such duties. Code-writing is something that requires real training, and even marketing materials might be trusted to those with real marketing knowledge. Indentures probably come from untrained stock, as the books mentioned that those talented people that escaped the fall without a body were first-priority for resleeving.

Untrained means that you can likely buy an AI to do the job instead. Remember that in EP very few jobs are just unskilled ditch digging: most require what we today would call university education. As I see it, the Fall flooded the market with desperate professional refugees. There was not enough shuttles to save all marketing execs, code review managers, dermatologists, rheologists and lobbyists - they were just as desperate as the nobodies, and at best just had a small advantage in getting in early in the uploading queue. Indenturing works just as well for professionals as for nobodies, except that 1) professionals tend to make more, so their contracts would tend to be shorter - a lot of them are now ex-indentures, and 2) creative work requires a lot of freedom in order to be creative. You can be indentured and yet have good working conditions - EP has a slight bias towards showing it as hell since it makes a more dramatic story and some of the authors might be a tad anticorporate. We never get to hear about the ex-indentures who are still working for their corporations.

Quote:

You'll be using indentures for low-training work. Operating doors, janitorial work, testing, mining, food staff, hauling, or basic assistant work. On a personal level, you probably have indentures doing housekeeping, basic home maintenance, guarding doors and providing sexual services (where it isn't illegal... sorry Jovians).

These jobs are all very close to what a cheaper AI can do. Why go about the bureaucratic mess of having an employee when you can have equipment? Some of them do require a transhuman, like testing, assistance and sex, but again you don't want the people doing them to be too discontent since it will be noticed by the customers.

Arenamontanus wrote:

Untrained means that you can likely buy an AI to do the job instead. Remember that in EP very few jobs are just unskilled ditch digging: most require what we today would call university education. As I see it, the Fall flooded the market with desperate professional refugees. There was not enough shuttles to save all marketing execs, code review managers, dermatologists, rheologists and lobbyists - they were just as desperate as the nobodies, and at best just had a small advantage in getting in early in the uploading queue. Indenturing works just as well for professionals as for nobodies, except that 1) professionals tend to make more, so their contracts would tend to be shorter - a lot of them are now ex-indentures, and 2) creative work requires a lot of freedom in order to be creative. You can be indentured and yet have good working conditions - EP has a slight bias towards showing it as hell since it makes a more dramatic story and some of the authors might be a tad anticorporate. We never get to hear about the ex-indentures who are still working for their corporations.

I actually see the relationship between AI and indentures to be one of strategy. There are certain scenarios where indentures are more beneficial to have than AI: [list=1][*]Costs. An AI has a cost up-front (whether it be programmers to create it, or the price of a license to acquire it), whereas an indenture contract has it's cost curtailed to the expiration period (if they make it without voiding their terms). [*]Public relations. The system (inner system especially) has an issue with AI by large, and there is a benefit to replacing those AI with indentures... it puts a human face on the labor force. Furthermore, it can be spun to make the contract-holder look charitable. While us (the audience) sees the repressive use of people in a manner that is only three-shakes from slavery, the public masses within the setting hear about how these "benevolent" hypercorps are giving the destitute a chance to earn a body. [*]Actual benevolence. There is the very real possibility that there are hypercorps that do want to do right by their indentures. Such seems to be the case for modern Somatek, which has over the years integrated many former uplift indentures into the fold as full-employees... and as such has drastically shifted their views to a far more tolerant and understanding one about the minds of uplifts. They may very well treat their uplift indentures with a lot of relative respect. I don't see it being a standard for all hypercorps. Cognite certainly sounds like a company that treats it's indentures little better than modern labs treat guinea pigs.[/list] As for writers' bias, I don't think the issue is so much anti-corporate as they are basing the concept of indentured servitude on the historical precedent. Indentured servants in the 17th and 18th centuries weren't treated very well at all, and were for all intents and purposes slaves. In fact, it was considered gauche for nobility to treat indentured servants well... they knew what they were getting into when they signed the paper, so being nice to them is a show of weakness. That concept might still be present amongst the hyperelite of 10 AF.

Arenamontanus wrote:

These jobs are all very close to what a cheaper AI can do. Why go about the bureaucratic mess of having an employee when you can have equipment? Some of them do require a transhuman, like testing, assistance and sex, but again you don't want the people doing them to be too discontent since it will be noticed by the customers.

But AI aren't necessarily cheaper. You still have to pay up front for a license or a programmer. An indenture can be acquired for no down payment, and you might be able to negotiate them down to a sleeve that is so cheap that you can nanofacture it yourself out of scrap a couple hours prior to the contract expiring. And if your servant is discontent? Well, customer satisfaction is a requirement in their contract. Too much of that and their time has been wasted, and they certainly don't want that to happen. Fear is likely a powerful motivator for compliance among the indentured masses.

Decivre wrote:

A videogame and a simulspace are likely going to still be two very different things in Eclipse Phase's time. A simulspace is a simulated reality, coalescing all the simulated physics and object representation under a massive rendering system. A vidgame, however, can severely reduce the mechanics of reality to whatever's necessary in order to produce entertainment. [...] Such shortcuts wouldn't be so easy when there are multiple actors within the system. A simulspace might have a dozen to hundred of people all operating within it, and need to render everything that any one of them might sense or interact with. That has the potential to eat an insane amount of resources.

A simulspace doesn't have to simulate physics anymore than a videogame does. If you want your game or simulspace to have detailed physics, you can. If you're running a virtual office, you're not going to care one iota about the atoms in the walls - it is just a wall that blocks vision and movement and doesn't need any properties beyond that. If you don't want to bother with walls or even embodiement, you don't need that either. Simulspace does in no way imply a detailed simulation of physics.

Decivre wrote:

Smokeskin wrote:

The argument doesn't hinge on the x60 speed. At x20, x10, x5, the case still holds. If state of the art is x60, then what about a mere 10% of that? We can agree that must be trivial, can't we? That's x6 speed. Trying to land a job where you can take your time off in real space would be like asking for a job where you only work 1 day a week.

I don't think that's necessarily trivial. The fastest computer today slightly tops 10 petaflops, which means that 10% of that capability is about 1 petaflop... well beyond anything most people have in their homes, and probably more powerful than the majority of business systems. In fact, the server I used to work on operated at 300 teraflops... and the GPU of a high end gaming rig today operates at a little over 500 gigaflops. That's [sup]1[/sup]/[sub]20,000[/sub] the power of the most powerful computer. So if the most powerful computer today could run a 60x simulation, a high-end desktop could run a [sup]1[/sup]/[sub]333[/sub] speed simulspace. Microfractions of the speed of the slowest simulspaces within the setting.

So by your way of looking at it, how much more powerful is a server than a high-end desktop? 3 times? So you're saying that EP servers can only run simulspace at around 1% speed? Sounds like your example is utter BS.

Decivre wrote:

Smokeskin wrote:

So you have infomorph employees working full time, and people sleeved in physical morphs working just one day a week. Who do you think will dominate in the workforce? Do you think CEOs (many of whom we know are perfectly willing to sacrifice health and personal life to satisfy their ambition and drive) will work and recreate in simulspace and get 6 times as much work done? What sort of CEO do you think the board of directors will hire, the one that works 6 times as much or the one that works at realtime speed? When the CEO selects the people to work with, do you think he wants people that also work on x6 speed like him? Do you think he'll hire people that take their time off in simulspace at x6 speeds, or do you think he'll hire people that only log into simulspace to work every 6 days of his subjective time? You can continue the line of reasoning on your own. When you then factor in that not only is there a huge competitive advantage in never living in realtime but that the cost of living in simulspace as an infomorph is much cheaper than having a physical body, that just makes it even clearer that infomorphs will dominate.

But again, this is all assuming that your company has access to the simulspace capacities necessary to gain that edge. I don't imagine it's very common, and in fact I work under the assumption that simulspace averages on the slow side. A common simulspace probably operates at around 80% realtime (I work under the assumption that most simulspaces aren't put together by pros and hypercorps), maybe a little faster. 1x is likely the benchmark standard for a simulspace system. To me, 60x simulspaces are the equivalent of Ultraviolet nodes in Shadowrun... mythical in scope, with few people ever catching glimpse of one.

Ok, just to be clear, since you've come over to my point of view there's no reason to argue against me anymore. This whole argument started with me saying that I don't allow x60 simulspaces because of the massive impact it would have on society, which you then argued against, and now you're working on the assumption that simulspaces don't run faster than 1x either, while this is where you started (bolding is mine):

Decivre wrote:

http://www.eclipsephase.com/no-limit-limited-technology-synthmorph-infom... Not all simulspaces will be run at 60x acceleration. Computers are still confined by processor limits, so larger scale simulations will be a greater load, reducing the speed as a result. [b]A virtual office might run at 60x acceleration[/b] but a virtual representation of the big bang (relevant to physicists) might run far slower.

Since it seems we now agree that there aren't x60 virtual offices (or x6 even), everything is fine. I can easily accept that there are some rare supercomputers that can run minds at x60 speed, just like there are seed AIs and nanoreplicating swarms.

I think the computational demands for a simspace are completely determined by the number of users, since what you need to simulate is essentially only what they experience and things that influence it. A human has a sensory bandwidth of about 10 gigabits per second, and I once estimated the computational requirements to fill this with believable vision, sound, tactile and chemical signals to around 100 TFLOPS. So each user would need around this, plus some slight overhead for unseen activities. The important part is that you don't simulate stuff that doesn't contribute to the experience. Greg Egan put it well in the opening of "Permutation City", mentioning that the protagonist knew that when he turned away from a painting it was "actually" replaced with an averaged rectangle for the ambient light calculations. Later on he went to have a shower just to annoy the owner of the simulation, since the hydrodynamics calculation was more expensive. Incidentally, I found this excerpt on BoingBoing a minute ago: http://boingboing.net/2012/06/14/excerpt-from-rapture-of-th.html

Quote:

Of course, the sim is far too realistic. A virtual champagne bath should somehow manage to keep the champagne drinking-temp cold while still feeling warm to the touch. And it shouldn’t be sticky and hot and flat; it should feel like champagne does when it hits your tongue—icy and bubbly and fizzy. And when Huw’s nonbladder feels uncomfortably full and relaxed in the hot liquid and she lets a surreptitious stream loose, it should be magicked away, not instantly blended in with the vintage Veuve to make an instant tubworth of piss-mimosa. This is what comes of having too much compute-time at one’s disposal, Huw seethes. In constraint, there is discipline, the need to choose how much reality you’re going to import and model. Sitting on an Io’s worth of computronium has freed the Galactic Authority—and isn’t that an unimaginative corker of a name? — from having to choose. And with her own self simulated as hot and wide as she can be bothered with, she can feel every unpleasant sensation, each individual sticky bubble, each droplet clinging to her body as she hops out of the tub and into the six-jet steam-shower for a top-to-bottom rinse, and then grabs a towel —every fiber slightly stiff and plasticky, as if fresh out of the wrapper and never properly laundered to relax the fibers—and dries off. She discovers that she is hyperaware, hyperalert, feeling every grain of not-dust in the not-air individually as it collides with her not-skin.

Note the importance of deliberate *non-realism*. A lot of this non-realism also would saves computational power. But it also takes design skill, which might be expensive. Putting plausible dirt in the corners of the sim is harder than it looks, and important for the sense of reality.

Decivre wrote:

Even so, you'll still need a computer with enough power to calculate what needs to be rendered and when. That isn't going to be a small task. Plus, there are the potential for elements that need to be simulated for later periods of time. A simulspace with flowers might need to keep track of the progress of those flowers, so that when a person returns to that spot, they will have grown. The same is true for other scenarios. A storm-ravaged virtual forest will still need to simulate a number of things while people aren't within, so that people will later see the effects of the storm.

That depends on how much extrapolation you and your simulspace visitors are willing to accept. You can save a lot of computer cycles by just developing an algorithm that creates the effects of storm damage, rather than actually modelling the whole thing, at the expense of possibly-recognizable patterns killing your sense of authenticity.

Pyrite wrote:

That depends on how much extrapolation you and your simulspace visitors are willing to accept. You can save a lot of computer cycles by just developing an algorithm that creates the effects of storm damage, rather than actually modelling the whole thing, at the expense of possibly-recognizable patterns killing your sense of authenticity.

Exactly. I'm saying that you'll have a wide swath of simulations. Some very realistic and potentially simulating every single elements, while others are broadly unrealistic and may only simulate things within the local environment of every user. It'll all depend on the needs of the user, as well as the purpose of the simulspace.

Arenamontanus wrote:

Does a simulated tree falling make a sound if there is nobody in that part of the simspace? I think most computationally constrained simspaces will simply ignore that part of the physics: there is no point in wasting computations that have too small effects. Really cheap simspaces just flip the tree from "normal tree" to "overgrown log". More advanced ones invoke a physics engine and run a simple mechanical model, noting if any secondary effects happen (like blocking a stream) and producing a sound/vibration simulation if potential listeners are close. If there are just a few bots, they can be put in scared mode without any sound - the boids fly away. Only very "rich" simulations play out the full physics.

I can see that. The more inferior simulspaces likely look more and more like a bad videogame as they have to cut down and take shortcuts to maintain speed. It would be interesting to find out how much of the system is taxed simply determining what falls within the sensory limits of every user, so the system knows what it doesn't need to render.

Arenamontanus wrote:

Note that humans are amazingly easily fooled in VR. You don't need much hints to convince people that the physics and optics are correct despite fairly serious errors. I seem to recall that distances can be off by 10% and people completely miss it. Connoisseurs will of course look for imperfect Torrance-Sparrow reflections and timbre mismatch in the ambient sound, but they are like us who watch movies in order to call out errors in the special effects.

Perhaps, but that doesn't necessarily mean that transhumans are as easily fooled. As cognition improves with modification, it may become harder and harder to creating a convincing simulated reality. You mentioned that the human mind probably accepts about 10 gigs of data per second, but how much harder would it be to create a believable reality for someone who's enhanced senses have double that intake? How do you fool someone with distances when they have a nerve cluster or cyber implant specifically designed for gauging distances to the millimeter? There might even be a subtle competition between genehackers and cyberware engineers who are tasked with giving (trans)humanity better senses with which to discern reality with, and simulspace designers that are tasked with fooling those senses.