What-If 0031: "FedEx Bandwidth"

What if there was a forum for discussing these?

Moderators: Moderators General, Prelates, Magistrates

speising
Posts: 2363
Joined: Mon Sep 03, 2012 4:54 pm UTC
Location: wien

Re: What-If 0031: FedEx Bandwidth

Postby speising » Wed Feb 06, 2013 6:31 pm UTC

In this vein: since switching to digital data transmissions, our em pollution becomes increasingly difficult to differentiate from random noise for any listening aliens.

User avatar
orthogon
Posts: 3099
Joined: Thu May 17, 2012 7:52 am UTC
Location: The Airy 1830 ellipsoid

Re: What-If 0031: FedEx Bandwidth

Postby orthogon » Wed Feb 06, 2013 10:24 pm UTC

speising wrote:In this vein: since switching to digital data transmissions, our em pollution becomes increasingly difficult to differentiate from random noise for any listening aliens.

Excellent point. Is that why SETI never found anything?
xtifr wrote:... and orthogon merely sounds undecided.

User avatar
Max™
Posts: 1792
Joined: Thu Jun 21, 2012 4:21 am UTC
Location: mu

Re: What-If 0031: FedEx Bandwidth

Postby Max™ » Thu Feb 07, 2013 2:22 am UTC

ijuin wrote:
Klear wrote:
keithl wrote:And of course every bit of information on the planet, natural and artificial, is hurtling at 600 km/s towards Abell 3627. Beat THAT, FedEx!


Everything is moving nearly the speed of light somewhere if you pick the right inertial frame. Nothing special about that...


I believe that the 600 km/s figure is relative to the Cosmic Microwave Background, which is the closest thing to a "universal" frame of reference that we have. That is to say, the CMB is blueshifted by about 600 km/s in the direction that we are moving.

http://calgary.rasc.ca/howfast.htm
Spoiler:
Image

Image


Neat site.
mu

ijuin
Posts: 1148
Joined: Fri Jan 09, 2009 6:02 pm UTC

Re: What-If 0031: FedEx Bandwidth

Postby ijuin » Thu Feb 07, 2013 5:02 am UTC

rmsgrey wrote:
ahammel wrote:Once it's sequenced and assembled it's just bits (even if slightly corrupt bits). Figuring out that those bits contain a hologram is probably not super difficult compared to, say, using a hard AI to travel faster than light while having a hot tub party with Helen of Troy and Neil Patrick Harris in your holodeck.


Okay, there are tricks used when designing messages intended to be beamed out to the stars in order to try to make the message as readable as possible - if the message in the DNA were intended to be read by "aliens", not just by people who already knew how to decode it, then making sense of it would be fairly easy. If it were encoded for efficient storage, then it becomes a serious cryptography problem...


Consider that the human(oid) genome is something like 4 billion base pairs (call it 8 gigabits/ 1 gigabyte). Let's be generous and say that 50% of this data space can be allocated to our message. That gives us half a gigabyte (512 MB). At Blu-Ray level resolution and currently-standard encoding methods, that's good for maybe ten minutes of video, tops. I can only assume that a 3D volumetric hologram would require more data for a given resolution than a 2D image. This implies that the data in a hologram message would indeed need to be highly compressed. Given the bulk of all of this data, we are left to wonder why it was left for us in the form of a hologram, possibly the most data-intensive medium, instead of just an audio and text message with a few photographs.

User avatar
ahammel
My Little Cabbage
Posts: 2135
Joined: Mon Jan 30, 2012 12:46 am UTC
Location: Vancouver BC
Contact:

Re: What-If 0031: FedEx Bandwidth

Postby ahammel » Thu Feb 07, 2013 3:37 pm UTC

ijuin wrote:Consider that the human(oid) genome is something like 4 billion base pairs (call it 8 gigabits/ 1 gigabyte). Let's be generous and say that 50% of this data space can be allocated to our message. That gives us half a gigabyte (512 MB).
The human genome is actually six gigabases, which works out to 12 gigabits (at two bits to the base) = 1.5 gigabytes, but your point is taken.

I was assuming that the message was in a long insertion, or a few artificial chromosomes. Probably not practical, it's true.
He/Him/His/Alex
God damn these electric sex pants!

rmsgrey
Posts: 3652
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: What-If 0031: FedEx Bandwidth

Postby rmsgrey » Thu Feb 07, 2013 4:37 pm UTC

ahammel wrote:
ijuin wrote:Consider that the human(oid) genome is something like 4 billion base pairs (call it 8 gigabits/ 1 gigabyte). Let's be generous and say that 50% of this data space can be allocated to our message. That gives us half a gigabyte (512 MB).
The human genome is actually six gigabases, which works out to 12 gigabits (at two bits to the base) = 1.5 gigabytes, but your point is taken.

I was assuming that the message was in a long insertion, or a few artificial chromosomes. Probably not practical, it's true.


Depending on how many generations you want the message to survive for, you not only need to embed the data, but also some machinery for preserving the data - if all humanoids in the galaxy come from the same original stock, then their current diversity (and relative uniformity within each type) points to considerable mutation and genetic drift having occurred since the common ancestral gene-pool. Without either massively redundant encoding, or some robust error correction each generation, the message would have become thoroughly garbled...

User avatar
ahammel
My Little Cabbage
Posts: 2135
Joined: Mon Jan 30, 2012 12:46 am UTC
Location: Vancouver BC
Contact:

Re: What-If 0031: FedEx Bandwidth

Postby ahammel » Thu Feb 07, 2013 7:06 pm UTC

Yeah, putting the DNA in anything alive is a silly way to do long-term data storage.
He/Him/His/Alex
God damn these electric sex pants!

rmsgrey
Posts: 3652
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: What-If 0031: FedEx Bandwidth

Postby rmsgrey » Fri Feb 08, 2013 10:01 am UTC

ahammel wrote:Yeah, putting the DNA in anything alive is a silly way to do long-term data storage.

I dunno - more of it's likely to survive seriously long-term that way than if you just bury a single copy somewhere. There are advantages to letting your data go viral...

webgiant
Posts: 252
Joined: Mon Aug 17, 2009 5:36 pm UTC

Re: What-If 0031: FedEx Bandwidth

Postby webgiant » Fri Feb 08, 2013 10:07 am UTC

Klear wrote:
rmsgrey wrote:
PolakoVoador wrote:Indeed very interesting numbers, though I also would like some ridiculousness.

The Halo bit reminded me of correspondence-chess. Can we consider it the very first form of "online" multiplayer?


I wouldn't be surprised if play-by-mail gaming turned out to pre-date chess.


I wouldn't be surprised if chess turned out to pre-date mail.

Technically "chess" didn't predate play-by-mail. Modern chess didn't officially exist until the end of the 15th and/or beginning of the 16th century (when the queen and bishop gained their modern system of moving), but King Henry I of England and King Louis VI of France in the early 12th century played a match of proto-chess by mail during the 12th century.

Thus play-by-mail gaming pre-dates modern chess.

As for mail itself, proto-chess didn't exist before the 6th Century C.E., while mail dates back to the Persians of the 6th Century B.C.E. . So mail predates even proto-chess.

ianai
Posts: 6
Joined: Sat Feb 09, 2013 7:13 am UTC

Re: What-If 0031: FedEx Bandwidth

Postby ianai » Sat Feb 09, 2013 4:38 pm UTC

jacobraccuia wrote:Would it still be faster to to mail a hard drive than send it online, when you take transfer times into consideration?

To transfer 500gb to an external, mail it, then transfer it vs. sending source to source..

I probably don't know enough about computers to be able to make an argument, but I feel like source to source would be faster.


The WhatIf was interesting, but clearly not apples-to-apples for the reason jacobraccuia mentions. The time it takes to pull data off whatever storage you have in Point A, put it onto an SSD or whatever, then pull it off the SSD to put it on the storage unit in Point B is non-trivial. When calculating the Internet transfer time, those operations are part of the total time, unlike calculating FedEx's bandwidth. Randall makes a small concession to this when he talks about the need for massive card processing center.

It is still faster to very large amounts of data, but not as much faster as Randall calculates in the WhatIf. Also, sending just SSDs or MicroSDs has another issue: The human time to connect all the drives. Having been FedEx'ed a slew of 500GB external drives (this was quite a while ago), connecting up 6 or so of them at once, running a script to transfer all 6 individually, physically hook up another 6, lather, rinse, repeat many, many times is a lot more annoying than typing a command and walking away for a week.

ianai
Posts: 6
Joined: Sat Feb 09, 2013 7:13 am UTC

Re: What-If 0031: FedEx Bandwidth

Postby ianai » Sat Feb 09, 2013 4:44 pm UTC

rmsgrey wrote:Local connections are generally faster than going via the internet - apart from anything else, transmissions over the internet have to resist degradation over longer distances, which means a lower transmission rate over equivalent hardware.


Transfer speeds can lower with distance because TCP is latency sensitive. However, there are non-TCP transfer methods which are latency insensitive. Also, you can transfer bits over the very long distance with no packet loss, so I am not sure if "degradation" is precise.

It is still faster to FedEx a bunch of drives if the amount of data to be transferred is very large. And in most cases a lot cheaper too. :)

ianai
Posts: 6
Joined: Sat Feb 09, 2013 7:13 am UTC

Re: What-If 0031: FedEx Bandwidth

Postby ianai » Sat Feb 09, 2013 4:45 pm UTC

Wnderer wrote:This sort of relates to a similar question I had. How much memory would it take to download Wikipedia and how much space would it take up?


Google for "All Of Wiki", an iOS app that downloads all the text on Wikipedia to your iPhone / iPad.

Currently 5.27 GB for text only, English version. Picture probably make that go up at least an order of magnitude. But I would be surprised if it is over a few TB.

ianai
Posts: 6
Joined: Sat Feb 09, 2013 7:13 am UTC

Re: What-If 0031: FedEx Bandwidth

Postby ianai » Sat Feb 09, 2013 5:01 pm UTC

Bounty wrote:
jacobraccuia wrote:Would it still be faster to to mail a hard drive than send it online, when you take transfer times into consideration?

To transfer 500gb to an external, mail it, then transfer it vs. sending source to source..

I probably don't know enough about computers to be able to make an argument, but I feel like source to source would be faster.


One of the cited articles lists the best case internet transfer rate as 1 terabyte per day, so conceivably you could move that drive over the internet in 12 hours. For this kind of thing though we're not concerned about your .5TB HD, we're looking at the 120TB from the Hubble Telescope.


I have personally transferred multiple TB over the 'Net in a single day.

The article assumes an office with a 100 Mbps connection. To put this in perspective, my house has a 100 Mbps connection. A business which needs to do large data transfers should have a larger connection than my house.

Most people doing data transfers measured in terabyte (or petabytes) colocate servers in datacenters. The largest datacenters on the 'Net - including every datacenter Google owns - have capacities measured in multiple terabits. A Tbps is 10,000 times the 100 Mbps limit mentioned. Individual servers in these datacenters have at least 1 Gbps links, and frequently 10 Gbps link - 100 times the 100 Mbps limit mentioned. (There are servers with Nx10G or 40 Gbps or even 100 Gbps links, although that is not typical.)

Obviously I have skipped over about a million tiny details, but I've probably already bored most people.... :) Point is, 120 TB is not a lot by today's standards. Which is not surprising, the link from the Hubble to the ground is not exactly modern or blazing fast.

ianai
Posts: 6
Joined: Sat Feb 09, 2013 7:13 am UTC

Re: What-If 0031: FedEx Bandwidth

Postby ianai » Sat Feb 09, 2013 5:08 pm UTC

mudlock wrote:Storage density increases by ~15% annually, which is less than the 29% internet bandwidth-usage growth. At these rates, the internet would still eventually surpass FedEx, but it would be closer to 2070 than 2040.


Not apples-to-apples. Randall is comparing the possible capacity of FedEx based on best case scenario (e.g. planes filled to the max with MicroSDs) compared to typical bytes transferred on the 'Net. Ditto for the increase.

The actual capacity of the 'Net is MASSIVELY more than is actually transferred. Especially if you use best-case scenarios, e.g. transferring things only locally (inside a city as opposed to across a continent or ocean).

But that doesn't make the WhatIf any less fun

rmsgrey
Posts: 3652
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: What-If 0031: FedEx Bandwidth

Postby rmsgrey » Sun Feb 10, 2013 9:52 pm UTC

ianai wrote:
rmsgrey wrote:Local connections are generally faster than going via the internet - apart from anything else, transmissions over the internet have to resist degradation over longer distances, which means a lower transmission rate over equivalent hardware.


Transfer speeds can lower with distance because TCP is latency sensitive. However, there are non-TCP transfer methods which are latency insensitive. Also, you can transfer bits over the very long distance with no packet loss, so I am not sure if "degradation" is precise.


For transmission of a waveform along a pipe, the maximum frequency of signal that can be transmitted without being garbled depends on the length of the pipe. You can get around that to an extent for digital signals by using repeaters, but unless you're prepared to power a repeater every foot or so, you're looking at longer hops for internet traffic than for dumping the data through a USB port onto an external drive...

You can get around the latency issue, but you can't practically get around the frequency one.

ianai
Posts: 6
Joined: Sat Feb 09, 2013 7:13 am UTC

Re: What-If 0031: FedEx Bandwidth

Postby ianai » Tue Feb 12, 2013 4:57 am UTC

rmsgrey wrote:
ianai wrote:
rmsgrey wrote:Local connections are generally faster than going via the internet - apart from anything else, transmissions over the internet have to resist degradation over longer distances, which means a lower transmission rate over equivalent hardware.


Transfer speeds can lower with distance because TCP is latency sensitive. However, there are non-TCP transfer methods which are latency insensitive. Also, you can transfer bits over the very long distance with no packet loss, so I am not sure if "degradation" is precise.


For transmission of a waveform along a pipe, the maximum frequency of signal that can be transmitted without being garbled depends on the length of the pipe. You can get around that to an extent for digital signals by using repeaters, but unless you're prepared to power a repeater every foot or so, you're looking at longer hops for internet traffic than for dumping the data through a USB port onto an external drive...

You can get around the latency issue, but you can't practically get around the frequency one.


Every foot? Repeaters on typical fiber links are 100s of KM apart, and bit error is targeted at in in 10^12.

I have a feeling I'm being trolled....

rmsgrey
Posts: 3652
Joined: Wed Nov 16, 2011 6:35 pm UTC

Re: What-If 0031: FedEx Bandwidth

Postby rmsgrey » Tue Feb 12, 2013 11:59 am UTC

ianai wrote:
rmsgrey wrote:
ianai wrote:
rmsgrey wrote:Local connections are generally faster than going via the internet - apart from anything else, transmissions over the internet have to resist degradation over longer distances, which means a lower transmission rate over equivalent hardware.


Transfer speeds can lower with distance because TCP is latency sensitive. However, there are non-TCP transfer methods which are latency insensitive. Also, you can transfer bits over the very long distance with no packet loss, so I am not sure if "degradation" is precise.


For transmission of a waveform along a pipe, the maximum frequency of signal that can be transmitted without being garbled depends on the length of the pipe. You can get around that to an extent for digital signals by using repeaters, but unless you're prepared to power a repeater every foot or so, you're looking at longer hops for internet traffic than for dumping the data through a USB port onto an external drive...

You can get around the latency issue, but you can't practically get around the frequency one.


Every foot? Repeaters on typical fiber links are 100s of KM apart, and bit error is targeted at in in 10^12.

I have a feeling I'm being trolled....


Typical local data links - to internal hard drives, or USB ports - are on the order of a foot in length. Those short links can support higher frequency data than typical internet links. You can improve the data rate by allowing multiple connections in parallel, or by using different materials for the link, but if the only difference between two connections is that one is thousands of times longer than the other, the maximum rate of data transmission for the shorter connection will be significantly higher because it can transmit higher-frequency signals without degradation, so you can fit more bits into any given time period.

User avatar
3rdtry
Posts: 152
Joined: Sat Feb 16, 2013 1:46 pm UTC

Re: What-If 0031: FedEx Bandwidth

Postby 3rdtry » Sat Feb 16, 2013 1:54 pm UTC

If only we hadn't abandoned the pneumatic tube postal systems, we could actually have an "intertubes" nowadays, and send many GBs of data across the city in a few seconds.

[damn, I can't even post pictures on my first post]


Return to “What If?”

Who is online

Users browsing this forum: No registered users and 4 guests