Basic question about Ego backup

29 posts / 0 new
Last post
SEÑOR FASE SEÑOR FASE's picture
Basic question about Ego backup

Hi!

I thought that only peeps with cortical stacks could have backups, but per the RAW, seems like anyone can... even peeps w/out one... True?

I know, I know... just confirming I'm not reading stuff 'rong...(sigh)

Thanks!

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

Yes. To make a backup without a stack (and you can do this even if you have one), you use the Uploading procedure. If you're not in a synthmorph, you need an Ego Bridge and some time.

nezumi.hebereke nezumi.hebereke's picture
Re: Basic question about Ego backup

The difference is, a cortical stack maintains a current backup. Uploading maintains an archived backup.

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

Right. It depends on what you mean by 'backup'. Given that the stack itself is basically synonymous with the current backup that it constantly makes, I just use the word 'backup' to refer to the archived ones. :)

The Doctor The Doctor's picture
Re: Basic question about Ego backup

Yerameyahu wrote:
Right. It depends on what you mean by 'backup'. Given that the stack itself is basically synonymous with the current backup that it constantly makes, I just use the word 'backup' to refer to the archived ones. :)

Perhaps 'offsite backup' would be a better way to describe it.



OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

It's true that a cortical stack contains a realtime back up. (If you have a stack you get backed up once every second. 86000 times a day. p.268) But the stack is just the storage medium not the recording mechanism. The mechanism is the nanite systems deployed through out the brain that detect the minutest changes in brain state and send the data to the model in the stack.

Essentially an ego bridge does the same thing. It deploys nanite systems through out the brain to record all of the data, (probably neuronal and synaptic positions as well as the analog chemical/electrical potential at each position that creates current thoughts and memories), then writes that into RAM to be run on software simulation or transmitted somewhere else.

So, you kinda gotta wonder, if people with Cortical Stacks all ready have the recording device installed and operating and it only takes a second to record the brain state, Why does it take an ego bridge 10 minutes to do the same thing?

Spoiler: Highlight to view

Fun Idea:
since I'm reading Kurzwell right now I thought I thought I'd share his ideas on what is actually happening and the amount of computation it would require.
On page 124 of The Singularity is Near, he writes;
"If we want to "upload" a particular person's personality... we may need to simulate neural processes at the level of individual neurons and portions of neurons, such as the Soma, axon, dendrites and synapses. The "fan out" (number of interneuronal connections) per neuron is estimated at 10^3. With an estimated 10^11 neurons, that's about 10^14 connections. With a reset time of 5 milliseconds, that comes to about 10^16 synaptic transactions per second.

Neuron model simulations indicate the need for about 10^3 calculations per synaptic transaction to capture the nonlinearities [ed. the analog nature of the signal]... resulting in an overall estimate of about 10^19 [calculations per second] to simulate the brain at this level. [the level of an individual personality]"

About memory he writes;
"...A resonable estimate is about 10^6 bits per...(patern or item of knowledge), for a total capacity of 10^13 (10 trillion) bits for a human's functional memory... if we model human memory on the level of individual interneuronal connections, we get a higher estimate. We can estimate about 10^4 bits per connection to store the connection patterns and neurotransmitter concentrations. with an estimated 10^14 connections, that comes out to 10^18 (a billion billion) bits."

So using those estimates a cortical stack is recording a billion billion bits every second, and mesh inserts can perform a hundred trillion calculations per second, to record an otherwise unmodified human. This is the lower bound on computing power in EP. Something like a Menton morph that is twice as smart with multiple times the amount of functional memory could at least double those numbers.

So, I'm still wonderning why it takes so long for an ego bridge to work.

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.

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

Specialization? Compression? Encryption? :) *shrug*

Prime Mover Prime Mover's picture
Re: Basic question about Ego backup

Maybe the 1 second recording time is just changes that have taken place in the last second. So Stacks just updating every second as opposed to rewriting. You could explain away the longer time it would take to copy the bulk of the ego as having been done during the time the stack was implanted. The bridge would require more time however due to having to start from scratch.

"The difference between truth and fiction, people expect fiction to make sense."

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

The only reason I can come up with for the work of an ego bridge to take so long given the presence of a cortical stack is that the ego bridge has to deploy it's own nanite recorders and they are somehow different than the ones a stack uses and that deployment would take time. That's extremely unsatisfactory to me because a Stack is a digital device so why not just run a probe into the stack and rip it like you would any other digital memory or a cyber brain?

I get the idea that the Stack update each second only has to record the data that is different than the previous second's data so it might not be recording an entire brain and thought process every update in the way you don't over write your whole backup hard drive every time you back up your data. But I think, (and maybe some technical people here could confirm/deny), that even if you're not writing all of those Billion Billion bits you're still scanning the entire brain so you're still doing every one of those hundred trillion calculations per second every second and I want to assume that an Ego Bridge has--at least--the processing power of Mesh Inserts.

Don't get me wrong; I'm extremely skeptical and the trouble of taking the time to upload/download a personality from a biological brain makes it feel more real to me. I like it. But I'm bugged by the idea that there's a time discrepancy.

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.

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

I love the incremental backup idea, but aren't stacks also highly tamper-resistant? It seems fair that could make it harder to fast-copy. Not that 10 minutes is a long time, of course.

sjmcc13 sjmcc13's picture
Re: Basic question about Ego backup

Prime Mover wrote:
Maybe the 1 second recording time is just changes that have taken place in the last second.
I agree with this. there are going to be parts of the brain that are not changing very much, the majority of long term memory is set and not changing, so the stack will not be updating those parts that are not changing, and would naturally focus only on what actually changed as a method of saving time.

Also with a stack making updates so quickly, if it misses a change the odds are that it will get caught quickly, so copy errors are going to be checked and fixed by the nature of the stack, but a ego bridge is only doing this once, so it would need to double or ever triple check the scan to remove any copy errors which would probably at least double its time frame.

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

That's another excellent point: the uploading/backup process is probably quadruple-checked. :)

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

Quote:
but aren't stacks also highly tamper-resistant? It seems fair that could make it harder to fast-copy.

I haven't read anything that makes me think that the memory architecture of the stacks is tamper resistant. Non volitile RAM certainly, but the tamper resistance comes in the sense that they don't connect to mesh inserts or have any wireless connection. (which is highly suspicious because they'd need wireless connection to the Recording nanite swarm or you end up running 'cables's through capillaries in the brain.)

It seems counterintuitive to me that the architecture the stack is such that the write speed is 600 times slower than the read speed for some reason.

Quote:
the majority of long term memory is set and not changing
this is true but both memory and thought are chaotically distributed throughout the entire brain. I'm less concerned about the number of bits updated in each back up than I am about the total computational power needed to do so. I think you still pretty much end up scanning the entire 'drive' to record thought and experience that is current up to one second ago. The Cortical stack is an almost real time model of your brain state. How do you do that without scanning the entire brain?

This is how it must work because of A and B.

A. If you die a horrible death you take stress upon reinstantiation for experiencing a horrible death. This means that the experience of a horrible death, which happened pretty much throughout your entire brain, was scanned and transmitted to the stack.

B. The stack is updated every second. If you're only going to record long term memory why update the stack every second? every ten minutes or once a day would be enough. How do you determine what has changed from one second to the next without scanning and comparing both models; the brain now vs the brain one second ago? Seems like that would require twice the computation of simply reading the current state and writing it down. I'm not a computer guy but I'll imagine there is some equivelent to the addressing on my hard drive for the neuronal model but it can't be nearly as effective because the analog nature of each synaptic transaction must be accounted for in the model and the scan. Otherwise we wouldn't differentiate between a biomorph brain and a cyber brain.

At least that's they way I see it. I'm not defending a position here. I'd be happy if someone could come up with something that would expand my idea of the concept.

Edit] error checking is a good idea. It could even involve the formation of discrete deltaforks that run chunks of personality and memory which could then be compared to the original copy. But one has to wonder why this same process isn't necessary with uploads from a cyber-brain.

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.

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

I think you're being very uncooperative as we work hard to come up with handwaving technobabble for you. :) You know perfectly well that sci-fi != science, and anything with nanites is magic.

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

Yerameyahu wrote:
I think you're being very uncooperative as we work hard to come up with handwaving technobabble for you. :) You know perfectly well that sci-fi != science, and anything with nanites is magic.

LOL
Wave your hand FASTER damnit!
:D
It's true that EP would be much easier if I had no interest in the science and the word "Why" was absent from my lexicon.

The book says that's the way it works and I'm ok with that but I have this itch in my head that if I come up with a reason it works that way it will have broader implications for the EP universe. There are other seeming conflicts that might be potentially involved; simulspace time dialation for one.

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.

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

Okay, so I *just* realized what your question actually is: 'why does the ego bridge take 10 minutes to make a copy *from the brain*?' In that case, the answer is obvious: it takes time for the "millions of specialized nanobots" to physically get into position throughout your nervous system. Duh. :)

Sorry, for some reason, I had it in my head that you were asking why it took 10min to copy a *stack* (which, of course, it doesn't; I'm just stupid).

Anyway… BAM. Science accomplished, question fully answered. And, dare I say, stupidity redeemed!

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

Ouch. Sorry Yerameyahu. That's not *Exactly* my question. I really need to get better at phrasing questions.

To break it down:

What's the difference between the nanites used to record the brain for a cortical stack and the nanites used to record the brain for an Ego bridge that causes the job of the ego bridge to take 10 minutes?

Seems to me like they're pretty much the same thing. But it takes an ego bridge 600 times longer to do it's job. If the reason is primarily that the ego bridge has to put nanites in the brain then I wonder why can't an ego bridge use the nanites that are allready there to service the cortical stack.

If the answer is that the nanites an ego bridge uses are different in some way then I wonder how are they different because it seems like they're doing pretty much exactly the same thing.

If the answer to that is that Cortical Stack nanites are not wireless enabled (which they're not, because of security concerns; no matter how cumbersome and improbable that makes the system), then wouldn't it be easier and much less invasive to simply create a nanite probe to tap the cortical stack?

If the answer to that is that a cortical stack is impossible to download by means of a physical probe then I have to ask; How do you get the data off a cortical stack?

My reading of the book indicates very strongly that the reason an ego bridge takes much longer to do it's job is because of some limitation intrinsic to a biological brain. I like that idea because it maintains the unique quality of the Cyber Brain in Synthmorphs and Pods and feeds the debate around the very important transhuman theme; Why do we continue to pursue biological existence.

Some of the 600 fold increase in time can be accounted for by the fact that the work of an ego bridge is more complete and detailed than the work of a Cortical Stack. Some of it can also be accounted for by the increased processing power required by error checking, decryption ect. So I'd buy that it takes maybe 10 times longer or even 60 times longer but not 600 times longer.

My conclusion is based on two things;
1. A cortical stack scans the entire brain and records a significant portion of that data in one second.
2. The same thing must happen when a biomorph ego enters a Simulspace because I'm Highly dubious that you can run x60 virtual acceleration of a mind on biological brain hardware but it doesn't take 10 minutes and an ego bridge to enter a simulspace. (I'm going to start a separate thread about this so let's not get into that here.)

And my conclusion IS:
There must be something I don't know about what an ego bridge is doing when it downloads and ego from a biomorph.

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.

sjmcc13 sjmcc13's picture
Re: Basic question about Ego backup

OneTrikPony wrote:
this is true but both memory and thought are chaotically distributed throughout the entire brain. I'm less concerned about the number of bits updated in each back up than I am about the total computational power needed to do so. I think you still pretty much end up scanning the entire 'drive' to record thought and experience that is current up to one second ago. The Cortical stack is an almost real time model of your brain state. How do you do that without scanning the entire brain?

because the stack is not the only part of the equation, and memory and the like being spread out should actually help, as it means that all the nanobots that are doing the sensing and transiting to the stack have less work to do (and might let you use less bots).

You would still update frequently because people want as up to date a backup as possible, not for a need, but because they want it. And of course they are not recording just long term memory, though they do not need to record everything. the nanobots do not need to record the signals sent out to make your heart beat, just the structure used, same as the signals for moving your arm. All they really need are memory and the basic structure of the brain.

As to death, the brain does not die all at once, and the stack will have time time to fully update once you die to include the effects of death. though the odds are that there are filters set into the stacks programing (and the sensor bots it uses) to tell it what part to record on death and what parts not to, otherwise when you are revived you could have brain damage from the changes that occured at death being re-created.

Yerameyahu Yerameyahu's picture
Re: Basic question about Ego backup

… That *is* your question, man. Seriously, I just answered this: the ego bridge nanites take 10 minutes to get *into* your brain. Presumably, you could make 2 copies in the same ten minutes (more or less), because then the nanites are fully in place. You're falsely assuming that the *speed* of the copying is a fair race, but it's not: the cortical stack is fully in place when the race starts, but the ego bridge's are not. Stop saying '600 times as fast!'. :)

And, no, you can't use the existing cortical stack ones. They're 'hardwired' to the stack for security.

And, no, the stack is designed *not* to accept a 'nanite tap' for the same reason.

For a stack that's been removed, download is already fast, so no worries there. How does a stack know it's been removed? No constant communication from the 'known' nanites, of course!

So your question now is, 'aww, c'mon, I really would rather use a shortcut tap of some kind, or hijack the existing nanites!'. :D

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

sjmcc13 wrote:
As to death, the brain does not die all at once, and the stack will have time time to fully update once you die to include the effects of death. though the odds are that there are filters set into the stacks programing (and the sensor bots it uses) to tell it what part to record on death and what parts not to, otherwise when you are revived you could have brain damage from the changes that occured at death being re-created.

I used this concept in my very first EP character. It's the only way I could see to rationalize the Neural Damage trait.

Quote:
So your question now is, 'aww, c'mon, I really would rather use a shortcut tap of some kind, or hijack the existing nanites!

I guess so. It just seems simpler, less invasive to me to do it one of those ways.

One thing that occurred to me is that in some cases the nanites injected by the Ego Bridge need to be able to wipe or overwrite the brain. Such as the case when you're leaving a morph and have no plans to return. (ID security again.) Obviously this is not a capability of Stack nanites so an Ego bridge would defintely need to inject it's own systems in that situation.

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.

nezumi.hebereke nezumi.hebereke's picture
Re: Basic question about Ego backup

I would agree with Yaremahu.

The stack is 'faster' because it already uploaded it yesterday. The upload today is just checking for changes. It also has all of the hardware already installed.

The stack is developed solely as an emergency safety measure. It MUST be fast, but it does not have to be as accurate - most likely any errors will be written over in 24 hours or less, but a delay of a few seconds in that real-time recording is not acceptable. It's also developed to be an independent system, and to not interface with anything else (to reduce chances of corruption or misconfiguration - it is a safety system, after all).

Ego-bridges are developed to be completely accurate, but not necessarily fast. They also need to 'start from scratch' in spreading their nanomachines.

Comparing the two is like the difference between brakes and wheel chocks. Different tools for different purposes, even if they do 'the same thing' - and you really don't want to remove one and just depend on the other.

If a character wanted to pull his stack (ouch) and use it as a quick ego-bridge, I'd permit it - but he's then without a net.

Arenamontanus Arenamontanus's picture
Re: Basic question about Ego backup

I think the sensible model is that the stack is building a model of the brain using a distributed system of sampling and measurement (e.g. measure the activity of all connected neurons, update the model based on this). The "once a second" thing is just how often it updates its own final model (i.e. reassuring advert-speak to customers of no importance). I would imagine an empty stack actually needs quite some time to record a brain it is connected to. Fortunately most sleeving also copies the intended brain state into the stack, so it starts out close to the real brain state.

Egobridges are doing black magic; as I said in an earlier thread on the subject this is one of the parts where EP is handwaving wildly to justify people with organic brains. Downloading a mindstate into an existing biobrain is quite implausible and impractical. See http://eclipsephase.com/cost-resleeving for some details and calculations. I think it is reasonable to assume - given EP tech - that uploading and downloading can be done in a hour or so. But best not think of what is actually happening in that brain and bridge.

Scanning a flat brain is probably easiest using disassembly. In my descriptions of the Fall gory emergency uploading stations figure heavily...

Extropian

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

@ Arenamontanus;

Quote:
the stack is building a model of the brain using a distributed system of sampling and measurement (e.g. measure the activity of all connected neurons, update the model based on this).

If a full upload requires a scan of 8x10^19 calculations and memory of (n)x10^18 bits what would be your estimate for the amount of calculation and data transferred in a simple Cortical Stack Update?

I'd like to understand how you can record all off the distributed processing going on in the brain state without scanning the entire brain at least once a second. Would you mind expounding a bit on that? I understand that once you've done it the first time you'll have the topology, 'wiring diagram' of which neuron goes where and which synapses are paired together. I understand the calculations that yield 8x10^16 (yours) and (n)x10^19 (Kurzwell's) are basically an estimate of the total number of Synapse pairs with some thousand additional calculations thrown in to account for analog/digital conversion.

So how do you figure out which synapses not to scan?

Also; given that we hear so much about the 'plasticity' of the adult brain these days why is it so hard to remodel a brain to accommodate a different ego and it's memories?

My thinking has been that the process didn't require a huge amount (or any?) of actual physical remodeling requiring neuronal connections be switched around but simply a comparison of "Old Map" to "New Map" and the best fit connections be artificially strengthened or created while inappropriate connections be eliminated. I thought that, over all, total connections are dense enough that one never need consider modfications at the scale of an axon or dendrite, that the length of a neuron is rarely an issue because you can almost always find a way to use synaptic pairs that allready exist or are at least close to each other.

Arenamontanus wrote:
We have ~10^11 neurons with ~8000 synapses each.

With 8000 synapses per neuron it seems to me that you've got a statistically good chance of having almost everything you need very close to where you need it. So all you'd need is some synaptic bridges and to clip some other synapses.

if memory and skill is mostly recorded in the Soma of the neurons involved I thought that once you've got the synaptic map solved you could then just reinforce (or mimic what ever memory does to the Soma) those particular neurons.

If my premise is completely wrong or the answer is time intensive I'll be satisfied with a simple; "No. Dumbass."

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.

Arenamontanus Arenamontanus's picture
Re: Basic question about Ego backup

OneTrikPony wrote:

If a full upload requires a scan of 8x10^19 calculations and memory of (n)x10^18 bits what would be your estimate for the amount of calculation and data transferred in a simple Cortical Stack Update?

My guess is that it is a change of synaptic weights for some synapses. Suppose the device samples ten billion synapses per second. It would take about a day to sample all, but most synaptic changes are small (during one second about 1% of all neurons fire, so synaptic updates happen at most for 0.0001 of all synapses (about ten billion or so)). So the stack has a fairly accurate map of the brain, with the oldest data maybe a day out of date but most data far more actual (and when run, the neural network properties tend to make slight errors disappear).

So the stack is not storing the activity itself, it is just measuring it to update the wiring diagram it maintains.

Quote:
Also; given that we hear so much about the 'plasticity' of the adult brain these days why is it so hard to remodel a brain to accommodate a different ego and it's memories?

Normal brain plasticity takes days and weeks. Neurons grow new synapses, send out branches and adapt to training.

In EP, supposedly you can rewire a brain to run not just a different person but even an AGI that decides to give meat a chance. That seems to require major re-routing of connections, not just a synapse here and there but long and tangled axons stretching over *centimetres*. That is tough.

Quote:
I thought that, over all, total connections are dense enough that one never need consider modfications at the scale of an axon or dendrite, that the length of a neuron is rarely an issue because you can almost always find a way to use synaptic pairs that allready exist or are at least close to each other.

This is unfortunately not true. There are hundreds of billions of targets, but each neuron just has 8000 connections - it is a very sparse network. There might be a way of getting a dense network if the cortical minicolumn model my old research group worked on happens to be true (essentially what matters is small groups of neurons, not individual neurons, and all these groups have connections to each other) - but I doubt nature has been so obliging.

Quote:

if memory and skill is mostly recorded in the Soma of the neurons involved I thought that once you've got the synaptic map solved you could then just reinforce (or mimic what ever memory does to the Soma) those particular neurons.

Unfortunately the place where stuff is stored is the synapses - the soma is just maintaining them and acting as a central point. The memories and skills all depend on what the synapses are doing.

Extropian

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

Damnit.

Well that kind of creates another potential conflict in the setting then. If an Ego Bridge has to ,and can, make large scale structural changes to a bio-brain, dis/reassembling neurons, in short time frames it should probably have implications for why it takes 3 years to grow one from scratch.

Oh well. I give up.
[Waving Hands! Waving Hands! There's noplacelikehome! Threresnoplacelikehome!]

Dude, it's super cool of you to take the time to respond in detail. Thanks :)

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.

sjmcc13 sjmcc13's picture
Re: Basic question about Ego backup

it takes 3 years to grow a body from scratch, though it you look at the tables for Healing Vats they can rebuild a body like 3 weeks, as long as certain requirements are met (an intact head in medical stasis).

It would be the same with a egobridge and the brain, a new morph would have a blank easily over writable brain designed to be over written, The bridge can re-write the connections between neurons, but re-building the neurons from scratch takes time, and creating a new brain takes even more time. The ego bridge is not making these neurons, or removing them , it is only effecting the connections between them. which is significantly easier then making new ones, and the odds are that there are going to be parts that do not change much from person to person, like the part of the brain that controls our heart beat, that is not going to change much, so the bridge probably leaves that alone most times, this means that the bridge is probably not changing the whole brain.

OneTrikPony OneTrikPony's picture
Re: Basic question about Ego backup

I believe the current thinking is that it takes 3 weeks to grow a body and 3 years to grow a brain. That's why you've got to have an intact head for a healing vat.

I'm fairly certain that Arenamontanus just explained to me that an ego bridge does have to move neurons around, or at least Axons and Dendrites, The easiest way to do that would be to destroy and rebuild them because they can be very long.

So even though an ego bridge is not changing the whole brain, if it's changing more than 3.17x10^-6 of a brain durring the ten minutes it operates, it follows that Tech exists in EP capable of building neurons faster than they can be grown.

That's probably a gross oversimplification in the analysis but all of EP is a simplification of the actual science.

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.

nezumi.hebereke nezumi.hebereke's picture
Re: Basic question about Ego backup

Or you could build connections using synthetic materials, including smart materials which can move more easily. This can maintain that connection long enough for the dendrites to build on naturally, and as long as the percentage or organization is appropriate, it's still a 'natural' (albeit, augmented) brain.

Axel the Chimeric Axel the Chimeric's picture
Re: Basic question about Ego backup

I think that building a brain may take time, but, for the most part, the ego transfer is non-destructive, and what damage is done can be mapped and repaired later. Time consuming? Certainly, but far less so than building the trillions of neural connections normally required.




[@-rep +1, f-rep +2]