Welcome! These forums will be deactivated by the end of this year. The conversation continues in a new morph over on Discord! Please join us there for a more active conversation and the occasional opportunity to ask developers questions directly! Go to the PS+ Discord Server.

My crew and I are still struggling with the backup idea ...

16 posts / 0 new
Last post
Skelshy Skelshy's picture
My crew and I are still struggling with the backup idea ...
they are on a mission and they have 3:40 minutes left until orbital bombardment hits. That got them thinking what they would really loose. We all were wondering how frequent backups would be given you had backup insurance, and if there was any specific cost to each backup? Can the individual trigger a backup? Can they backup themselves to safety seconds before destruction? It seems with unlimited storage and bandwidth, the answer might be you can do them at any time, as long as you're meshed? Earlier I thought death is just property damage and one of the few ways to get to you really is to drive you insance. If some joker resleeves you in a low pain tolerance morph for torture (or in a fancy VR like in Altered Carbon) what do you really have to lose? Just find a pre-insane backup and you're good to go. They'd have to hack your backup or infect you with something nasty but not deadly and slow acting so you'll have to trade off life experience and or a fresh restore. What are sensible limits? Have to do backup in person in a lab? Thanks S.
Arenamontanus Arenamontanus's picture
Skelshy wrote:It seems with
Skelshy wrote:
It seems with unlimited storage and bandwidth, the answer might be you can do them at any time, as long as you're meshed?
Sounds reasonable, but in practice this might be a bad idea. Ongoing backups presumably includes sending the changes from previous states: this requires a continuous stream, and breaks will mess up versions unless you have a smart protocol. It will also means a constant stream of data that might be detectable. More importantly, if something gradually goes wrong with your brain like an exsurgent infection, it can be harder to figure out how far to roll back. I would hence go for intermittent backups that map the entire state. The bandwidth requirements are iffy. EP tries to ignore storage demands and bandwidth (since they always turn out to be embarassingly low in games after a few years), but I think there are limits. A brain most likely requires on the order of tens or hundreds of terabytes (I have done various calculations and estimates for this); compression might reduce this by an order of magnitude. The channel capacity of the mesh is the real bottle neck. Current wireless systems have a limit of around 850 MB/s, making a full backup take about 3.5 hours. The ultimate theoretical capacity of laser communications http://www.rle.mit.edu/quantummuri/publications/additions4_05/shapiro_5.pdf is on the order of terabits per second. So then a full backup takes about 10 minutes. Assuming the channel is not crowded by other people trying to leave. Most of the time I hence assume backups are made to cortical stacks rather than over the mesh, and when people sleep or rest, the stack backup data is moved properly to the storage facility.
Extropian
Arenamontanus Arenamontanus's picture
Hmm, building on the previous
Hmm, building on the previous post: Let's assume high-bandwidth mesh links are via lasers to tiny base stations built into most surroundings. Your equipment has a power of one milliwatt, which is -30 dBW and hence your optimal bandwidth is about 1 Tb/s. So a full ego can be transmitted in minutes (especially if egos only encompass the difference from a standard brain and these are decently small; I have my doubts about that, but there are issues of compressibility of neural connectivity too). Note that figure 4 in the above paper assumes a 1 km link (in vacuum): you have a decent range, although air, dust and so on decreases it a bit. Now, what happens when a team tries to escape the Imminent Doom? (which kindly enough doesn't interfere with communications... I am much more fond of Imminent Dooms that also jam communications ;-) ) In this case there will be several milliwatt sources shining away, adding noise light, but this is likely (if I understood the paper right - I am *not* a signals guy!) not a problem with the right kind of detectors. The real problem is available spectrum: each signal has a bandwth which is limited by its coherence bandwidth, which in turn is dependent on delay spread. If we assume people in a 1 km area the delay spread will be 300,000 Hz, and you can transmit a huge number of egos (since the frequencies are on the order of 10^14 Hz). But if the people are crammed into a room, with distances measured in meters, then the delay spread will be 300 million Hz. Still more than enough for a few people, but now the spectrum can only handle a few hundred thousand - and this was *theory*, using perfect sensors and equipment, no people getting in the way of the lasers, no back-scattering and so on. It is at this point the evil GM gleam comes into my eye and I start drawing slowly advancing progress bars in the air...
Extropian
NewtonPulsifer NewtonPulsifer's picture
This thread was a duplicate;
This thread was a duplicate; most of the discussion that already happened occurred here: http://eclipsephase.com/my-crew-and-i-are-still-struggling-backup-idea As to bandwidth, I'm of the opinion we've haven't nearly reached the limits for bandwidth. For lasers we have 84.5 terabits per laser, 12 on 1 fiber strand: http://www.theregister.co.uk/2012/09/23/ntt_petabit_fibre/ For wireless we have spectral efficiency of 95.7 bit s^−1 Hz^−1 by exploiting the orbital angular momentum of photons: http://www.nature.com/nphoton/journal/v6/n7/full/nphoton.2012.138.html EDIT: Your mit.edu link is getting "Page not Found" so I couldn't read it. EDIT 2: Here's some bits per second per hertz to compare to the 95.7 achieved OAM link has a spectral efficiency of 95.7 bits per hertz LTE maxes out at 16.32 bits/Hz 802.11n is 2.4 bits/Hz Digital TV (DVB-T) is just 0.55 bits/Hz So we're still in the very early stages of exploiting wireless bandwidth. There's plenty more to come. http://www.extremetech.com/extreme/131640-infinite-capacity-wireless-vor...
"I fear all we have done is to awaken a sleeping giant and fill him with a terrible resolve."- Isoroku Yamamoto
Arenamontanus Arenamontanus's picture
NewtonPulsifer wrote:Your mit
NewtonPulsifer wrote:
Your mit.edu link is getting "Page not Found" so I couldn't read it.
Hmm, the page seems to have disappeared recently. The paper is behind a paywall at http://www.opticsinfobase.org/jocn/abstract.cfm?uri=jon-4-8-501 Nice to know the MIT estimates were rather conservative.
Extropian
OneTrikPony OneTrikPony's picture
NewtonPulsifer wrote:OAM link
NewtonPulsifer wrote:
OAM link has a spectral efficiency of 95.7 bits per hertz
I forgot about that other thread. Thanks for bringing up OAM again. Orbital Angular Momentum only works if you assume everyone using it is wearing a rather large spiral reflector hat to transmit. LOL Regardless; you still have the issue of bandwidth between your transceiver and the spiral reflector antenna that gives your signal its angular momentum. Sure, you might speculate that, in the future, someone has figured out how to do OAM without the big spiral reflector but that's a hard bit of physics to handwave.

Mea Culpa: My mode of speech can make others feel uninvited to argue or participate. This is the EXACT opposite of what I intend when I post.

NewtonPulsifer NewtonPulsifer's picture
OneTrikPony wrote
OneTrikPony wrote:
NewtonPulsifer wrote:
OAM link has a spectral efficiency of 95.7 bits per hertz
I forgot about that other thread. Thanks for bringing up OAM again. Orbital Angular Momentum only works if you assume everyone using it is wearing a rather large spiral reflector hat to transmit. LOL Regardless; you still have the issue of bandwidth between your transceiver and the spiral reflector antenna that gives your signal its angular momentum. Sure, you might speculate that, in the future, someone has figured out how to do OAM without the big spiral reflector but that's a hard bit of physics to handwave.
Here they've done it holographically in silicon: http://www.ncbi.nlm.nih.gov/pubmed/23232408 Plus the Alan E. Willner experiment above did *not* use a spiral reflector either. Is there a thread where someone gave you the idea that taking the garden shears to a satellite dish antenna is the only way OAM will work? It's not - its just the cheapest/easiest with off the shelf components if you wished to do it right now. One thing about EP, is they have mind-bogglingly powerful *optical* computers. It follows that the exact same optical tricks they use inside their optical computers will work in long distance communications. Simply put in EP bandwidth will scale with computing power.
"I fear all we have done is to awaken a sleeping giant and fill him with a terrible resolve."- Isoroku Yamamoto
OneTrikPony OneTrikPony's picture
I havn't paid for any of the
I havn't paid for any of the white papers but all of the publicly available info, from a few years ago and just recently leads me to believe that this is a non starter. If I'm familiar with the willner experiment you refer to, it had a range of how many meters? One. Because OAM multiplexing is effected by atmospheric turbulence and requires a clear LOS. The reflector isn't necessary to transmit an OAM signal it's just the only way anyone has imagined to transmit that signal over a few hundred meters in atmosphere.
Quote:
It follows that the exact same optical tricks they use inside their optical computers will work in long distance communications.
Not really. That's a huge stretch. How do you scale your SLM to transmit in atmosphere over say; 1000 meters? Regardless; you still have the issue of Directionality. The EP mesh works with no large investment on infrastructure. This is not the tech you're looking for. :)

Mea Culpa: My mode of speech can make others feel uninvited to argue or participate. This is the EXACT opposite of what I intend when I post.

Tango Tango's picture
Without touching the science
Without touching the science part of this, i feel uploading a backup should take an hour or two at minimum. Swapping the cordical stack for a new one is of course the fastest way, but that means physical delivery...and risks associated with that (which of course is a positive thing as far as the gaming experience is considered). EP:s intro story features some sort of emergency neutrino farcaster implanted in a morph that can send the ego out in an instant, but i feel gadgets like those should be the exception to the rule. I feel that players should instead get creative; Start recording LD audio/video after every backup, and implant emotional responses as metadata in the video's timeline. Should the player's find themselves in a pinch you described, just upload that file in the cloud and retrieve it later for analysis.
- "Mom's chicken soup, maybe?"
NewtonPulsifer NewtonPulsifer's picture
OneTrikPony wrote:I havn't
OneTrikPony wrote:
I havn't paid for any of the white papers but all of the publicly available info, from a few years ago and just recently leads me to believe that this is a non starter. If I'm familiar with the willner experiment you refer to, it had a range of how many meters? One. Because OAM multiplexing is effected by atmospheric turbulence and requires a clear LOS. The reflector isn't necessary to transmit an OAM signal it's just the only way anyone has imagined to transmit that signal over a few hundred meters in atmosphere.
Atmospheric turbulence is a solved problem at least to 3km or so. Also, why is LOS necessary? A scattered OAM photon will work just fine. Yes, it will give up some of its momentum to whatever it scattered through or off of, but if I transmit at +88 OAM and it arrives at -82 OAM it can still be received and carry useful data.
OneTrikPony wrote:
NewtonPulsifer wrote:
It follows that the exact same optical tricks they use inside their optical computers will work in long distance communications.
Not really. That's a huge stretch. How do you scale your SLM to transmit in atmosphere over say; 1000 meters?
SLM? Is that an initialization for Spatial Light Modulator? I'd say nobody would use an SLM in EP. You want to use constructive/destructive interference and optical resonance chambers. SLM is lo-tech. EDIT: Wrt to SLM - on reflection I suppose we are basically talking about the same thing. Please forgive the pedantry. When I said the above I had things like photon re-use in mind (if you can make a reversible computing optical computer, you can make a reversible optical communicator. And by optical I mean EM waves/photons of any type). This way you have huge reductions in energy used to transmit.
OneTrikPony wrote:
Regardless; you still have the issue of Directionality. The EP mesh works with no large investment on infrastructure. This is not the tech you're looking for. :)
Why is directionality an issue? You'd want to use a lot (and by a lot I mean potentially millions) of scanned arrays anyways - this is the logical way forward with MIMO style technology (regardless of how much Orbital Angular Momentum adds to things). Speaking of MIMO it looks like NTT DoCoMo already got 5 gigabits per second all the way back in 2007 (that's 50bits per hertz) by using 12 MIMO antennas for LTE-A (that's 4G LTE-Advanced). EDIT2: I'm not really seeing how the mesh isn't using a large investment of infrastructure? It has a huge infrastructure - just much of it is privately owned by individuals. Are there going to be publicly used fiber runs? Yes. Just like public sidewalks, parks, and roads. If the economics hurts your suspension of disbelief, you can have people having to pay 100 credits a year in taxes for the backhaul infrastructure in your EP Universe.
"I fear all we have done is to awaken a sleeping giant and fill him with a terrible resolve."- Isoroku Yamamoto
OneTrikPony OneTrikPony's picture
Good Link! Thanks. It's
Good Link! Thanks. It's pretty rare these days that I don't google myself into a paywall. I do not have a suspension of disbelief issue. I don't really see how the mesh *isn't* using a large investment of infrastructure either. If I had a suspension of disbelief problem it would be with the mesh as presented in the core. In EP there is no privately owned mesh infrastructure, no American Tower Corp. no AT&T. Right? The mesh is publicly owned in the sense that every one who uses the mesh actually caries a piece of it in their skull. That's all the mesh is; an ad-hock network built of participating nodes. Those participating nodes are inserts and public spimes. Don't blame me, I didn't write it. :) It's a eutopian concept. I like it because, in EP, I wouldn't have to pay AT&T $100 a month to tell me how much of American Towers capacity I *Can't* use. On the otherhand your entire communications infrastructure has to fit inside the skulls of the people who use it. That's a situation that makes me Extremely skeptical that use of the term "unlimited bandwidth" actually means unlimited. I have no problem with the way the mesh works and is used as described in the setting. I hope it's actually possible some day. My suspension of disbelief issues and skepticism are with the ideas that the mesh is capable of carrying real-time (or even frequent) wireless backup traffic. Perhaps the paper you just led me to will help with that skepticism.

Mea Culpa: My mode of speech can make others feel uninvited to argue or participate. This is the EXACT opposite of what I intend when I post.

NewtonPulsifer NewtonPulsifer's picture
Here's another one that isn't
Here's another one that isn't behind a paywall: "Influence of atmospheric turbulence on optical communications using orbital angular momentum for encoding" 4-Jun-2012 http://arxiv.org/pdf/1204.5781.pdf If you want to just extrapolate observed trends there's "Nielsen's Law of Internet Bandwidth" which would have bandwidth increase by about 6x10^8 times in the same time as computing power went up 1x10^10 times If your baseline for a handheld computer was 12 gigaflops and 8 megabits (usable real-world) per second for today's computing power, then you'd have 120 exaflops and 480 terabits per second. If an ego file was 20 petabits in size, you should be able to copy it in 42 seconds. Bump that another 10 years or so of growth and you have 12 zettaflops and 27.4 petabits. You can see the huge difference even 10 years makes. For your EP Universe you could take it the other direction. 10 years earlier. Somehow they managed to squeeze an ego into 1.2 Exaflops through clever engineering, but wireless takes like 2400 seconds to copy an ego. A fiber link can increase it by 10x (consumer grade) or 100x (server grade) faster, however, for a 240 or 24 second transfer. That might allow a GM to create some dramatic tension of having a firefight while the ego file copies.
"I fear all we have done is to awaken a sleeping giant and fill him with a terrible resolve."- Isoroku Yamamoto
Telos Telos's picture
Backups and Save Points
The players in my group have a simple mechanic. When you perform a backup take a snapshot of your character sheet. This can be done w cell phone cams or even a GM digital cam for archiving your character's backups. This is effectively keeping track of your latest brainprint. Most backup insurance policies perform the backup every six months. You can pay more for nonstandard backups but typically even then this process is only done every month. For those who are less tech savvy or camera deprived you can also keep a Rez Log which tracks expenditures and changes to the character in question. The only other method(despite what real-world science indicates) is to pay for the highly cherished Neutrino Farcaster. This will allow you to escape that impending doom if you're within the Sol solar system. Any other methods to "backup" a character on the fly i.e. the mesh is just a pc workaround to avoid rez loss and shouldn't be considered as it takes away from the spirit of Eclipse Phase. Yes you will die and sometimes even under the best of circumstances you lose time and character advancement(rez expenditures). Hopefully this helps.
Welcome to the network.
NewtonPulsifer NewtonPulsifer's picture
I have to 100% disagree.
I have to 100% disagree. Making backups/forking ponderous and difficult is a 180 degree turn from the spirit of Eclipse Phase. [b][i]Evacuating a Cyberbrain[/b][/i] [i]Characters inhabiting a synthmorph cyberbrain may voluntarily choose to evacuate by copying themselves as an infomorph onto another device. This takes 1 full Action Turn. See Infomorph Resleeving, below.[/i] [b][i]Spare[/b][/i] [i]Spare morphs are small, cheap, lightweight, synthetic shells designed to be used as a replacement should someone’s original morph be killed or destroyed. [b]A cortical stack (retrieved from the character’s previous morph) can be easily plugged into the spare morph, effectively resleeving them[/b] (they must make Integration, Alienation, and Continuity Tests as normal). Once plugged in, it takes only 3 Action Turns for the cortical stack’s data to be read and checked for integrity and the ego to run inside the spare’s cyberbrain.[/i] Now saying the spirit of Eclipse Phase [i]as written[/i] with regards to the ease of backups/forking makes it a less satisfying game depending on a GM/group's style of gameplay is another thing. To repeat,[i]as written[/i] the spirit of Eclipse Phase is convenient backups/forking/merging - not difficult, ponderous, or painful. If you argued [i]as written[/i] that convenient backups makes for a worse gaming experience, I'd tend to agree with you - but I would not agree that it fits the spirit of Eclipse Phase.
"I fear all we have done is to awaken a sleeping giant and fill him with a terrible resolve."- Isoroku Yamamoto
Scottbert Scottbert's picture
NewtonPulsifier has it.
NewtonPulsifier has it. Transferring a stored ego from a file, or from a cyberbrain, is fast and easy. HOWEVER, you're limited by radio range and wireless link quality. Not quite to the level some folks posit above -- but I could see even a smart protocol needing several actions over a poor link. I think radio ranges are given in the gear section. ALSO, you're limited by how accessible the backup is. Someone in a synthmorph can send their ego right away -- but people in biomorphs and pods usually _don't_ have access to their cortical stack (this is a security measure, so that if their personal mesh is compromised it can't be used to wipe, corrupt, or copy their stack). So, if you're in a biomorph or a pod you don't have a backup to send. (Note that the device in the story that radios stack contents back has two modes: Normally, it regularly transmits a copy of your stack /when in radio contact/ with a nearby storage facility. Once, it can use a tiny bit of antimatter to send a neutrino burst anywhere in the solar system, but the process of generating the energy needed fries the morph. Costly, but it's there for exactly the type of situation your characters are in.) Making a backup takes ten minutes for a biomorph, and of course you can save a copy every time you resleeve anyway. I'd say it depends on the situation -- some people might be paranoid enough to make daily backups, others might do it less often.
NewtonPulsifer NewtonPulsifer's picture
This has been covered in a
This has been covered in a couple of other threads, but its one thing to prevent mesh access to your cortical stack, and another thing entirely to prevent all convenient access. There's a compromise. Like a locked access jack wired to the cortical stack. In addition, there's no reason not to encrypt the data on the cortical stack. http://eclipsephase.com/are-cortical-stacks-encrypted http://eclipsephase.com/cortical-stacks-are-terrible-backups Part of the result of the discussion in the first thread was the realization that there is unbreakable file level encryption in Eclipse Phase (unbreakable unless the crypto keys are obtained). Otherwise crypto-cred couldn't work. Which begs the question of why one can't have their cortical stack encrypt the data and output the file(s) as write-only.
"I fear all we have done is to awaken a sleeping giant and fill him with a terrible resolve."- Isoroku Yamamoto