Actually it had to do with the inability of the authors (and most commenters) to do simple math. The article said: "[...] were placed on Chicago exchanges 2-3 milliseconds after 2 pm." The speed-of-light-delay between the dc and chicago is 3.2 ms. So, no, not "physically impossible".

Yes, the technology is constantly improving and today it may be feasible to use repeaters. I'm just saying how it was a few years ago when I worked on something like that.

I admit I know very little about fiber but the difference in the effective speed that light travels at is quite substantial and adds up over long distances and based on what we were told at the time there was no way to avoid that, which is why the client was paying for the MW-based solution. Maybe they were misinformed, I don't know.

The price of the MW link is high but from what I've been told it does pay for itself and that's all that matters. There is some redundancy to improve reliability and of course the data is still transmitted over the fiber connection as well, so if your MW link fails you can try to salvage the situation but you only have a chance if your competitors' MW links also failed. Otherwise, you are left in the dust and the only thing you can do is go and yell and the people responsible for your MW link, which is a welcome break for the algorithms people who are usually the ones getting yelled at :) (or so I hear).

Of course you cannot get the "theoretical maximum" but my point was that you CAN do 0.2 ms or less of the theoretical minimum latency for the distances involved here if you use point-to-point microwave, optimized for very low latency, so 3.5 ms is certainly possible. It's expensive and can only handle very short messages, but it's good enough for this application. At least that is how the systems I worked on a few years ago performed.

I'm referring to communication between traders, who are part of the same company but in different cities. It is the latency of this communication that provides the advantage to the traders. Communication between the exchanges themselves is irrelevant.

I cannot speak for this particuclar company, I just gave it because it seems to do the same thing as the custom communication systems that I have worked on in the past and you asked for a link. The ones I worked on did work without regeneration over similar distances. Yes, it's very expensive and you are limited to very short messages but none of these things are a problem for this particular application.

Come on, my post wasn't that long, the least you could do is read it carefully before replying. I repeat: there is NO fiber involved. Fiber is too slow, even if it were point-to-point. Low-latency microwave is the name of the game in this business and yes, it is point-to-point.

This may be true in this case, I was only pointing out how HFT communication systems work, which you don't seem to understand. In many cases, the data is information from other markets which can be obtained very quickly automatically.

http://www.aviatnetworks.com/solutions/low-latency-microwave/

And you obviously have no clue how trading communication systems work. Hint: there is no switching or cables involved. Low-latency microwave transmission systems, designed specifically for such applications get the data there within 0.1-0.2ms of the speed-of-light-delay between the two locations given in the article.

If you bothered to read the e-mails from the actual hardware designers in the linked thread (I know, I know, I'm new here and I should know that nobody ever reads anything), you would have seen that the things you are saying in your posts have no relation to the reality of the situation. The hardware people said that they know very well the way they're doing things makes it difficult for the software and they gave their reasons for doing it this way. Whether we agree that the reasons are valid is a seperate question, but your claim (that they are not aware of the fact that this causes the software to be a mess) is simply wrong.

If it is a wormhole (or by using a "warp" drive) then the scenario I described may not work for producing a paradox. However, if you have two separate wormholes, one connecting points A and B and another connecting points C and D and you position them such that they are far enough from each other, not to affect each other's regions of spacetime but close enough that you can travel between A and C and between B and D at slower-than-light speeds then you can still violate causality and create paradoxes. There is a short description of that here: http://www.physicsguy.com/ftl/html/FTL_part4.html#sec:stmanipulation and there is a more detailed discussion in Everett's book Time Travel and Warp Drives: A Scientific Guide to Shortcuts through Time and Space (this covers both warp drives and wormholes, as well as other topics related to FTL and causality).

I may not have been clear but what I mean is that an ansible can allow any two other observers (one on each side) to communicate FTL in a flat space-time and these observers do not need to be at rest (even if the endpoints of your ansible must be at rest) and this is all that's needed to run into problems. Also, the speed of the observers does not necessarily need to be close to the speed of light and the farther away the two observers are the lower the minimum necessary speed for causing violations. A more concrete example:

Let's say observers A and B are at point (0,0,0) and C and D are at (1000ly,0,0). A and C are at rest relative to each other and have an ansible connection between them that will break if they are not at rest anymore. B and D are not in anyway restricted so they can travel and any slower-than-light speed that they want. So, B starts going at 0.5c relative to A and to the left of A and and D starts going at 0.5c relative to C and to the right of C. Thus B and D are going at a significant speed in each other's frame. A sends a radio message to B, which immediately relays it through the ansible to C, which sends it via radio message to D. Then it goes the other way from D to C by radio, through the ansible to B and by radio from B to A. So, what you have is A and D communicating much faster than light, while also moving thus easily allowing them to send messages into their past. In this case the fact that the ansible only works if the two endpoints are at rest relative to each other does not help you unless you also make up something that prevents A and C from communicating with nearby observers.

Another example with a somewhat different setup but still exploiting the same problem is shown here, with pretty space-time diagrams: http://www.theculture.org/rich/sharpblue/archives/000089.html. In this case you have A and B with an ansible and C and D with a separate ansible. A and B are at rest relative to each other. And so are C and D. However A and B are moving relative to C and D and are able to communicate by regular methods, which violates causality.

No, it has nothing to do with that (which is indeed garbage). You actually do influence your past. For example you can send your FTL message to a satellite and have it send it back to you and depending on the speeds of the message and the satellite you can receive the message before you sent it. This can create a paradox because you can setup your experiment so that you only send the message if you haven't received it at time t. But after you bounce it off the satellite it can arrive at time t-1, which would cause you not to send it, but if you didn't send it, then how did you get it in the first place?

FTL communication resulting in being able to send a message to your past follows from special relativity. Here are a few resources that show the details of how it works:
http://www.theculture.org/rich/sharpblue/archives/000089.html
https://en.wikipedia.org/wiki/Tachyonic_antitelephone#Two-way_example
http://www.physicsguy.com/ftl/html/FTL_part4.html (unlike the others this also covers the case of non-flat spacetime, such as warp drives or wormholes)

No, it will cause problems if *any* two objects communicate FTL, not just the endpoints. So, even if your ansible only works if the two endpoints are at rest relative to each other, you can still easily cause problems by having the endpoints communicate (using standard EM waves) with other objects which are free to move at any (slower than light) speed. We know very well that there is no problem with two objects in different frames communicating with each other using EM waves, even in cases where the time dilation is significant (e.g. satellites communicating with Earth). So, your scheme would not only have to require keeping the two endpoints at rest but it would have to prevent any standard communication between the endpoints and other objects that are not at rest relative to the endpoints. Now, *that* is contrived and would also make the whole system not very useful.

Regardless of whether it is possible to do anything faster than light, this statement demonstrates a profound ignorance of modern physics.

Hint: it has nothing to do with light itself. The c constant is a fundamental property of space-time in the universe we live in. Light happens to travel at that speed in vacuum because photons have no mass, but the real problems with communicating at higher speeds has nothing whatsoever to do with light or with anything else we perceive. Also, physics does not actually say that you cannot communicate faster than light. It "only" says that if you manage to do that you can violate causality and create paradoxes.