Fiber Optic Latency on a FE model

I have yet to see somebody refute this:

Quote from: Robosteve on November 06, 2008, 10:50:29 AM

The receivers use complex statistical algorithms to predict what the incoming signal will be before it arrives.

I have yet to see you show any evidence of this.

Quote from: Robosteve on November 06, 2008, 09:51:36 PM

I have yet to see somebody refute this:

Quote from: Robosteve on November 06, 2008, 10:50:29 AM

The receivers use complex statistical algorithms to predict what the incoming signal will be before it arrives.

I have yet to see you show any evidence of this.

The whole point of a conspiracy is that there is no evidence.

Quote from: Robosteve on November 06, 2008, 09:51:36 PM

I have yet to see somebody refute this:

Quote from: Robosteve on November 06, 2008, 10:50:29 AM

The receivers use complex statistical algorithms to predict what the incoming signal will be before it arrives.

I have yet to see you show any evidence of this.

It's gotta be a troll post, I mean, can someone actually believe there is an alg to figure out the next sequence of bytes in an encryption key?  In a datastream involving human user input?  Not to mention it would introduce a ton of corrupt data as it could never be 100% effective... over even 10% for that matter¹. 

It's a nice idea though:  I'd love to crack it, then see what it would predict for downloading tomorrow's lotto numbers. 


1. unless it was predicting data from a Tom post:  "I don't see any"... [PREDICT TEXT:=proof.]   

It's gotta be a troll post, I mean, can someone actually believe there is an alg to figure out the next sequence of bytes in an encryption key?  In a datastream involving human user input?  Not to mention it would introduce a ton of corrupt data as it could never be 100% effective... over even 10% for that matter¹.

Even a completely random algorithm would be correct 50% of the time when predicting binary code. They've just programmed it to make predictions based on what data has been received before, such that as more and more data comes in, it learns from its mistakes and its efficiency improves. I would say that it is currently greater than 99.9% efficient for us not to notice any adverse effects.

Quote from: paddy on November 06, 2008, 11:07:39 PM

It's gotta be a troll post, I mean, can someone actually believe there is an alg to figure out the next sequence of bytes in an encryption key?  In a datastream involving human user input?  Not to mention it would introduce a ton of corrupt data as it could never be 100% effective... over even 10% for that matter¹.

Even a completely random algorithm would be correct 50% of the time when predicting binary code. They've just programmed it to make predictions based on what data has been received before, such that as more and more data comes in, it learns from its mistakes and its efficiency improves. I would say that it is currently greater than 99.9% efficient for us not to notice any adverse effects.

You gotta be kidding me.

By that logic, you could create a blank binary string large enough to fit the next winning lotto numbers, flip a coin for each binary digit, and have a 50% chance of getting it right.   

It's 50% to predict a single bit, not binary code.  It's the same with a coin flip:  1/2 heads.  Getting heads twice:  1/4, 3=1/8...1/16 1/32 1/64 1/128 1/256 1/1024... sensing a pattern here?

Not to mention, the whole point of sending data is to send some information that is not known.  I message someone in Europe to 'pick a number between 1 and 100' and you really think the odds can be any better than 1 in 100?  I ask them to send a compressed video of their birthday...you think that can be fudged?? 

How can you possibly think this??

Funny how the cables aren't even close to straight lines in this map. Almost as if they were laid down using faulty navigational data...

Just saying, is all...

Goldstein, you really need to learn to chill out. You'll give yourself an ulcer.

Quote from: ragnarr on November 06, 2008, 10:00:04 PM

Quote from: Robosteve on November 06, 2008, 09:51:36 PM

I have yet to see somebody refute this:

Quote from: Robosteve on November 06, 2008, 10:50:29 AM

The receivers use complex statistical algorithms to predict what the incoming signal will be before it arrives.

I have yet to see you show any evidence of this.

The whole point of a conspiracy is that there is no evidence.

First off, the receivers do not know what the signal is nor does it care.  These are dumb layer 1 devices that send both voice streams and data across the same line.  It does not know the difference between them.  It is another device further up in the stream that reads the stream and splits up the IP packets from the voice. 

Second, I am not going to waste my time trying to disprove everything that you make up on the basis of "they could be doing it".  Why don't you start by explaining how they might go about doing this, what kind of experts would you need, how many, etc.  Then it may be worth addressing in detail.

Quote from: ragnarr on November 06, 2008, 08:54:37 PM

Funny how the cables aren't even close to straight lines in this map. Almost as if they were laid down using faulty navigational data...

Just saying, is all...

Or it could be that map is not official, or the fact that because the earth is round and that is a flat map, distortions would have to be made.  The map below is accurate with respect to the cable runs.

I have yet to see somebody refute this:

Quote from: Robosteve on November 06, 2008, 10:50:29 AM

The receivers use complex statistical algorithms to predict what the incoming signal will be before it arrives.

This should refute your claim:

Then again, we may need the jumbo roll.

Funny how the cables aren't even close to straight lines in this map. Almost as if they were laid down using faulty navigational data...

Just saying, is all...

The cables don't look like straight lines as the map is a flat projection of a spherical surface. The curves are cause by the distortions of the projection method.

I have yet to see somebody refute this:

Quote from: Robosteve on November 06, 2008, 10:50:29 AM

The receivers use complex statistical algorithms to predict what the incoming signal will be before it arrives.

Because it's silly, that's why. You can't predict the next bit of a random binary string, and if we could develop algorithms to predict what the next bit was going to be, we'll drop the bits it could predict to get a better throughput.

By that logic, you could create a blank binary string large enough to fit the next winning lotto numbers, flip a coin for each binary digit, and have a 50% chance of getting it right.   

No, because you have to get them all right together for that to work.

It's 50% to predict a single bit, not binary code.  It's the same with a coin flip:  1/2 heads.  Getting heads twice:  1/4, 3=1/8...1/16 1/32 1/64 1/128 1/256 1/1024... sensing a pattern here?

And they would only need to predict one bit at a time. Hence, 50%.

Not to mention, the whole point of sending data is to send some information that is not known.  I message someone in Europe to 'pick a number between 1 and 100' and you really think the odds can be any better than 1 in 100?  I ask them to send a compressed video of their birthday...you think that can be fudged?? 

Yes. Videos get streamed all the time; it's possible to statistically predict what one frame will be based on the previous one. If the person picks a random number between 1 and 100, the receiver will look at all the cases where people have ever had to pick a random number, and guess which number is going to be picked.

Because it's a crock of sh1t and people can tell your just trolling. If we could use statistical information to predict an incoming signal (in its entirity) before it arrives, you've have discovered som kind of magical telepathic means of communication. There'd be no wires, satellites, pseudolites, stratelites, blimps. We'd just have computers "predicting" correctly, what messages were due to come in.

If you can do this you'll never work another day in your life. Unfortunately you can't.

There are many means of reconstructing signals, but this comes from redundancy in the transmission.

Shoo shoo little troll.

If there was no information being transmitted, how would the receiver know what to base its predictions of what's coming next on?

First off, the receivers do not know what the signal is nor does it care.  These are dumb layer 1 devices that send both voice streams and data across the same line.  It does not know the difference between them.  It is another device further up in the stream that reads the stream and splits up the IP packets from the voice. 

Who told you that? The Conspiracy?

Because it's silly, that's why. You can't predict the next bit of a random binary string

So now all information being transmitted everywhere in the world is random? Forget organised text, music and video, we'll just send a random stream of bits across and hope it makes sense!

and if we could develop algorithms to predict what the next bit was going to be, we'll drop the bits it could predict to get a better throughput.

A better throughput? You mean it would seem like the signals were travelling faster than they really are, almost as though the cables were laid over a longer distance than RET would have us believe?

Yes. Videos get streamed all the time; it's possible to statistically predict what one frame will be based on the previous one.

Come now Steve, you know darned well that any frame to frame predictions occur when the video is compressed and encoded, not when it's transmitted.  A repeater only sees data as 0s and 1s.

And they would only need to predict one bit at a time. Hence, 50%.

Wrong, cable testers use protocols, which the other end has to recognize.

Yes. Videos get streamed all the time; it's possible to statistically predict what one frame will be based on the previous one. If the person picks a random number between 1 and 100, the receiver will look at all the cases where people have ever had to pick a random number, and guess which number is going to be picked.

Learn what a frame is, I think you mean datagram.

Streaming video and VOIP uses UDP in the transport layer, which is not a reliable protocol like TCP.  With UDP, there is no checking at the session layer, it is not needed, if a single datagram in a stream is lost, or corrupted, it doesn't matter to the end user because it looks like a momentary blip in the video or a crack in the sound (for VOIP).  The exception for this is streaming video from webpages, which runs over TCP.  However in that case the video is buffered.

If there was no information being transmitted, how would the receiver know what to base its predictions of what's coming next on?

It doesn't, its a layer 1 device.

Quote

First off, the receivers do not know what the signal is nor does it care.  These are dumb layer 1 devices that send both voice streams and data across the same line.  It does not know the difference between them.  It is another device further up in the stream that reads the stream and splits up the IP packets from the voice.

Who told you that? The Conspiracy? 

I am a network engineer.

So now all information being transmitted everywhere in the world is random? Forget organised text, music and video, we'll just send a random stream of bits across and hope it makes sense!

There are protocols which specify at each layer how the headers are structured.  For example in an IP packet (layer 3), the protocol IPv4 specifies that the first 4 bits must indicate the version number.  The receiving device reads the first 4 bits and knows this is the version.  However, the device does not know beforehand what will be in that field.  However, the recievers are not layer 3 devices and therefore do not understand IP protocol.  From their perspective it is just a bunch of 1's and 0's.

See RFC 791 for the IP packet header specifications:  http://tools.ietf.org/html/rfc791#section-3.1

A better throughput? You mean it would seem like the signals were travelling faster than they really are, almost as though the cables were laid over a longer distance than RET would have us believe?

Accept the device would be wrong 50% of the time at the physical layer.  At higher layers this would be a problem since with reliable transport there is a checksum used to verify that the header received is not corrupted.  If it doesn't match then it would cause a protocol such as TCP to send massive amounts of retransmits.  This would increase exponentially until hardly anything would ever get through to the other side.  Routers at one end would not be able to see their neighbor at the other, which would cause BGP to shit itself.  Congratulations to the conspiracy, they just destroyed the internet.

Quote from: zeroply on November 07, 2008, 01:00:41 AM

Funny how the cables aren't even close to straight lines in this map. Almost as if they were laid down using faulty navigational data...

Just saying, is all...

Or it could be that map is not official, or the fact that because the earth is round and that is a flat map, distortions would have to be made.  The map below is accurate with respect to the cable runs.

So you post a map which has "straight lines"  but is obviously just a schematic and does not reflect the shape of the cables. I don't think they actually go out 500 miles into the ocean in a straight line, make a 45 degree turn, and then go in a straight line for 2000 miles. How about giving me some real data instead of a bunch of assumptions? Things such as:

1. Map of undersea cables to scale and with accurate positions.
2. Latency data of the connections.

You are assuming that the latency will reflect a round earth because you have already convinced yourself that the earth is round.

So now all information being transmitted everywhere in the world is random? Forget organised text, music and video, we'll just send a random stream of bits across and hope it makes sense!

I didn't suggest that all the data was random. There will, however, be plenty of data where, from the receivers point of view, the next value will be unpredictable.

A better throughput? You mean it would seem like the signals were travelling faster than they really are, almost as though the cables were laid over a longer distance than RET would have us believe?

You don't understand much about compression algorithms, do you? They remove patterns, making the data less predictable than it is. They improve throughput by reducing the size of the data being sent though, not by making it go through faster.

Quote from: ragnarr on November 07, 2008, 04:35:20 AM

Quote from: zeroply on November 07, 2008, 01:00:41 AM

Funny how the cables aren't even close to straight lines in this map. Almost as if they were laid down using faulty navigational data...

Just saying, is all...

Or it could be that map is not official, or the fact that because the earth is round and that is a flat map, distortions would have to be made.  The map below is accurate with respect to the cable runs.

So you post a map which has "straight lines"  but is obviously just a schematic and does not reflect the shape of the cables. I don't think they actually go out 500 miles into the ocean in a straight line, make a 45 degree turn, and then go in a straight line for 2000 miles. How about giving me some real data instead of a bunch of assumptions? Things such as:

1. Map of undersea cables to scale and with accurate positions.
2. Latency data of the connections.

You are assuming that the latency will reflect a round earth because you have already convinced yourself that the earth is round.

1.  To scale would require it to be drawn on a globe, not a flat map.  Representing a spherical object on a flat surface will always cause distortions.  The map I provided is the official one.
2.  I don't really care about latency, there are amplifiers about every 100km along each of those connections anyway.  The fact is those cables had to be manufactured and run to a specific length.  If you were contracted to run those cables, and you purchase or manufacture cable for a distance assuming RE, then they would have run out of cable before it was run.  The lengths of these cables are available.

You could have found this latency data yourself:

http://www.southerncrosscables.com/public/Network/default.cfm

Segment C:  Takapuna New Zealand - Spencer Beach Hawaii

Length:  8002 km
Bandwidth:  10gbps  WDM
delay:  40.01ms

If it helps, when we last discussed this, RoboSteve believed that...

Quote from: Robosteve on October 11, 2008, 09:05:37 AM

Those cables expand when they reach the ocean floor due to geothermal energy.

Yeah that would definitely break them, also, not all of them are on the ocean floor. The Polyethylene that is on the outer part of the cable would melt at the temperatures that it takes for steel to expand that far.

And they would only need to predict one bit at a time. Hence, 50%.

Unbelievably stupid, or troll? Or both?

Quote from: paddy on November 06, 2008, 11:27:30 PM

By that logic, you could create a blank binary string large enough to fit the next winning lotto numbers, flip a coin for each binary digit, and have a 50% chance of getting it right.   

No, because you have to get them all right together for that to work.

Quote from: paddy on November 06, 2008, 11:27:30 PM

It's 50% to predict a single bit, not binary code.  It's the same with a coin flip:  1/2 heads.  Getting heads twice:  1/4, 3=1/8...1/16 1/32 1/64 1/128 1/256 1/1024... sensing a pattern here?

And they would only need to predict one bit at a time. Hence, 50%.

And you don't need to get bits to all work together - just lotto numbers??? 

Do you understand there is no difference between a series of letters, a series of numbers, and a series of bits - they all all just in a different base (26*,10,2) but can all represent the exact data?  A single digit 2 can be written as 10 in binary 3 as 11 and 4 as 100,etc... but you would need to predict the complete sequence for it to all work together - period.  A binary representation of a sequence of lotto numbers is no easier to predict than decimal representation of a sequence of lotto numbers. 

If you cannot see your error, I'll make a prediction (50%/50%) to determine you are a troll or very numb in the brain.  Just admit you are wrong and move on... it happens.


* base 26 if we are just talking about A-Z one case, no spaces or punctuation.  ASCII would be 'base 256' but it is never written as that, it is written as one letter per byte, aka an eight bit sequence.

Quote from: Robosteve on November 07, 2008, 06:02:59 AM

Yes. Videos get streamed all the time; it's possible to statistically predict what one frame will be based on the previous one. If the person picks a random number between 1 and 100, the receiver will look at all the cases where people have ever had to pick a random number, and guess which number is going to be picked.

Learn what a frame is, I think you mean datagram.

Streaming video and VOIP uses UDP in the transport layer, which is not a reliable protocol like TCP.  With UDP, there is no checking at the session layer, it is not needed, if a single datagram in a stream is lost, or corrupted, it doesn't matter to the end user because it looks like a momentary blip in the video or a crack in the sound (for VOIP).  The exception for this is streaming video from webpages, which runs over TCP.  However in that case the video is buffered.

Robosteve's error is so big that got you confused. Robosteve is talking about the frames of pixels that form a single image in a video; The basis for mpeg compression is, in part, that every frame has the same background than the last in a big percentage of the frames, so only the changes in the background are transmitted in the new frame. The whole image is only transmitted every fifty frames.

This is a very curious use of the word "prediction", and you could say it is correct (sort of). But the ISPs have no idea which of the TCP or UDP sessions are graphic, and anyhow they do not have the alternative of achieving better compression ratios than those achieved by the end users. They also have to send the right information, not what they predict might be what is expected. A 50% chance is totally unacceptable. Bit error rates of one in a billion (or 99.9999999% accuracy) are typical and bit error rates of one in a million are sometimes considered unacceptable.

Robosteve's error is so big that got you confused. Robosteve is talking about the frames of pixels that form a single image in a video; The basis for mpeg compression is, in part, that every frame has the same background than the last in a big percentage of the frames, so only the changes in the background are transmitted in the new frame. The whole image is only transmitted every fifty frames.

Thanks, I didn't know that about mpeg. 
When we are talking about intermediary devices, we are not talking about video frames, we are talking datagrams, packets, and switched frames.  Though I should have realized he was speaking of the application itself.

This is a very curious use of the word "prediction", and you could say it is correct (sort of). But the ISPs have no idea which of the TCP or UDP sessions are graphic, and anyhow they do not have the alternative of achieving better compression ratios than those achieved by the end users. They also have to send the right information, not what they predict might be what is expected. A 50% chance is totally unacceptable. Bit error rates of one in a billion (or 99.9999999% accuracy) are typical and bit error rates of one in a million are sometimes considered unacceptable.

Routers and switches would have no idea, however, a content filter would. Here's another flaw in his reasoning, what about data that is encrypted.


Most leased lines come with a 99.9 percent uptime guarantee.

Not to mention, to get around the latency issue, you end up having to predict new bits based on the last bits you received, that were also just guesses. 

I really don't think anyone could believe such a system could ever work, I think it was just an attempt to derail this topic and get away from the real issue - you would need to exceed the speed of light in a vacuum to compensate for the FE distorted map, where the speed of light through glass works perfectly for a RE model.

They use negative refractive index meta-materials to get around this problem.

They use negative refractive index meta-materials to get around this problem.

and that is supposed to solve which problem?

They use negative refractive index meta-materials to get around this problem.

This still can not make Light Travel faster than it does in a vacuum. All negative refractive index materials do is refract the light in the opposite direction than it would in normal materials. So that if you had a prism of normal material that was set up to refract the light to the left, a Negative Refractive index material would refract it to the right.

SO it would not get around the problem.

Optical fibres work because the amount the light is refracted is enough to bend the light back into the fibre. So using Negative refractive index materials would actually destroy the required properties needed by the optical fibre to be able to carry a light beam from one place to another. 

Of course, this all ignores that there are people that actually lay the cable on the seabed and ensure that it is buried.  

The cable is loaded into the cable tank on the cable ship.  This ship then lays out the cable, based on RE charts, in the location detailed by the owner of the cable.  A trenching machine ensures that the cable is buried in the seabed to prevent damage from fishing nets and other potential sources of damage.

It wouldn't take a transoceanic cable in the southern hemisphere to create noticeable differences in distances.  Even cables between the islands of Indonesia, or between Australia and New Zealand, would have significant differences in distance that would be noticed by the crew detailed to lay these cables.

Without going into the differences between the speed of light inside a fiber optic cable, or the necessity for a "predictive algorithm," there are a lot of problems with the comparison with accepted RE geography and assumed FE geography.  I would say accepted FE geography, except there isn't anything that is accepted when it comes to FE geography.  You have the polar projection, you have Tom Bishop's "Columbus/Magellan" map, and the admission that there really isn't a FE map because that haven't been able to figure out how to make a FE map that matches real world observations.

Rig, you're forgetting...

Damn, you are right.  I was forgetting all of those.

I was counting on actual observations to have some validity in explaining real world phenomenon.

"You've all got it wrong - the negative refractive index material is used as a seeding core for a photonic crystal fibre, generating superluminal group velocities in order to overcome the greater distance. Erbium doping ensures that the fibre needs much fewer booster stations, preventing the laying ships from becoming suspicious that they are laying more in the south than in the north. The dopants are periodically fed power by ultrashort THz frequency pulses fired from UAV's overhead, overcoming the difficulties associated with discrete deep-sea power transfer."