"I'd guess that usb3 latency is on same level than fw and usb3.1 will be even better. So actually the latency differences might be about 1ms. Is this so very important for so many?"
Can you back this claim with any evidence? Or were you speculating?
Intel have already announced Thunderbolt1 cables will work with Thunderbolt2.
Ok, then the cables work, maybe not next time anyway.
Do you have any evidence that usb3.1 have worse latency than usb3.0?
Or any reason to suspect so?
If someone can tell any latencies for usb3 connected pro(sumer) audio devices, I'd be happy to learn some more.
So your complaint is that Thunderbolt is too expensive, yet you think they should pick optical which would make it significantly more expensive. At 100Gb/s it's too expensive to take of even at places like Facebook with current technology, these optical modules are currently hand made.
If you are talking about 100G ethernet, it is expensive, but not that expensive. Or is Facebook in some kind of economical trouble now?
http://www.eetimes.com/document.asp?doc_id=1281026
;P (Ok, $10k per port is pretty steep...)
But really 100G switches are expensive, but TB or LP switches are not. Since they just don't exist. And probably never will.
Ethernet developers made a big mistake by going from 1G straight to 10G. That was far too expensive when they did it to become widespread enough. Now we've been waiting for 10Gether to become affordable about an decade.
So that's not what I suggested, what they should have made is
real optical Light Port with 10Gps. Or even 5Gps if it would have been much cheaper. After that there could have been wide demand for optical modules and slowly the prices would have gone down. If after about a 5 years there would have been 100 million optical modules sold, they would really be cheap. Now TB still has zero optical modules sold, so it will be long time before prices drop and when even more expensive optical TB version is introduced, lots of people start asking that can't we just continue using this old cheaper version, because it's fast enough anyway.
S/pdif is remarkable accomplishment, since it's the only time in industry that new media(optical) has been managed to make cheap.
So what I really tried to say, is that I think that optical 5th gen TB would be cheaper if they would have started with optical. Now the jump from electrical to optical when 3rd or 4th gen will arrive, will be almost impossible.
For 1st gen TB adopters, I'd guess it would have been no issue, if devices would have been $100 more expensive. To get TB wider acceptance, it would have and will be issue, but maybe Apple isn't interested in wider acceptance.
Since it now uses x4 PCIe v2, going to v3 would increase bandwidth to 40Gb/s with the same amount of lanes. Adding lanes could mean a need for new cable and sockets, unless there are spare and unused connectors in the socket.
You got the problem opened. Adding lanes will be very hard for maintaining backward compability. Also making one lane faster in copper will be very difficult. And going to optical now will also be very hard for backwards compability. So my best guess is that they don't dare to jump to optical yet next time and they can't double the speed in copper lane. So that's why I'm guessing they will try to increase speed by 50% and if needed make new bigger connector that has more lanes, but you can still use old connector (just like what they did with mini-usb3).
Longer lifespan, how so? You can continue to use gear at it's rated spec even if newer faster standards come along, it's pretty much a cost of doing business though as things don't remain constant. I have seen you previously think the omission of PCMCIA was a bad choice, why don't you just use USB 3 then? The fact is that Thunderbolt is everything and more than both PCMCIA and FireWire.
I do think that removing expansion that is widely accepted in industry is mistake and I also think that replacing that with new expansion that is not widely accepted does not correct the mistake.
Apple could have kept the EC (yep, EC is not the same than PCMCIA, if you don't know) slot at the same time when adding a TB port. This things can and should overlap. This would be most beneficial to users, since they can choose when they like jumb to some new standard.
Same way Apple could offer MP that has same external expansion than new MP will, but at the same time have the same internal expansion that the old MP had. But they chose not to.
Same way they chose in 2011 offer TB without usb3. Although the latter would have been far more useful for far more greater audience.
Optical interconnect would have longer lifespan if it would have been chosen from the beginning, because then there would be no need to change the connection ever. No adding lanes and changing the shape of connector according to that. Now it will be very interesting to see if there is any convenient way to do that when they have to change to optical. Will they stay with active cables having media conversion on both ends on all cables? What will the prices be then? If they make real optical connection with passive cables, how do you connect older TB devices to that? I see no easy roadmap ahead. But I do see some professionals asking, why other than Apple's laptops can daisy chain two 4k monitors, but macs can't.