Isn't the whole point of progress getting more for about the same money ?
Indoor plumbing, running water, and proper sanitation cost more but get better quality (and length) of life. Everything doesn't always boil down to cheaper.
You are spinning TB not as a problem solving solution , but as a competition thing. Competition isn't progress.
TB is supposed to be (future) mainstream, not exotic niche technology .
As PCs generally go smaller TB probably will be mainstream. Laptops took over from desktop of being dominate form. Laptops are continuing to shrink and all-in-ones are largely the only growth category in desktops. As space is squeezed out the role that full sized PCI-e cards played will be a solution space that Thunderbolt steps into.
The latter can justify higher prices for better performance, and catering to unique user demands .
Better performance is a justification in itself for higher prices where it makes a business difference. Long term the price will go down over time, but it is relatively rare faster and cheaper instantly in the same iteration. With most PC technology at this point in the evolution things are faster just to tread water on price. So can continue to charge the same amount. So no it isn't much cheaper.
Regardless of potential, I believe TB needs to beat USB 3.0 in performance (done),
You are twisted on the what the speed is being used for. TB
transports the data of other protocols. If TB was slower than those other formats inserting it would slow that data traffic down. That would not create any value. So TB needs to be faster so the "overhead" of encoding-decoding and transport in TB protocol format is transparent to the underlying protocol. Speed is being used to solve a distance-without-penalty problem. Solving that problem is what the speed is for. It is not some marketing crotch grabbing hoo-haa to boast about "my bandwidth is bigger then his bandwidth". The point is to solve the problem of remote location or current stuff with little to no perfomrance impact. That's what TB does.
The point is not to generate the cheapest box. It is to couple multiple boxes together with multiple data streams with little to no performance loss.
When USB 3.0 goes to 10Gb/s then Thunderbolt will have a problem if it can't go to something fast enough to transport it. Having to stay out in front of USB practically guarantees that Thunderbolt is
always going to be more expensive given both technologies have the same access to the underlying fabrication technologies. Staying out in front of the other protocols (going faster) and holding costs about the same would be progress.
There is no way Thunderbolt is going to deliver value if it is not tackling the problems it was designed to solve. "As cheap as USB or Firewire" is
not what TB was designed for. Never was. All the "one port to rule (and replace) them all" and "Its a USB klller" stuff was always and still is total FUD.
Whether TB costs drop dramatically depends upon how broad the adoption goes. It is a bit artificially high now because Intel is still turning the spigot on to allow device implemenntors into the ecosystem. There is little need or desire for a race to the bottom with crappy/flawed implementations. That is a mentalit that attempts to push the focus on areas where TB and USB overlap a bit. That is just a battle TB just can't win due to USB legacy deployed footprint. Billions of devices is a ton of inertia.
but also come with the pricing, usability and availability of Firewire or even eSATA (not even trying).
Firewire and eSATA don't transport multiple protocols. Neighter one is particular good to hauling data for the other. Nor for multiple instances of itself ( FW800 can't handle two FW800 streams. eSATA isn't going to do concurrent SATA 6Gb/s accesses. )
And for the new MacPro, it has to happen now, considering all performance crucial data storage is forced to go external .
The storage of the Mac Pro isn't zero.
Progress over the next 2-5 years the storage capacity inside the Mac Pro is only going to go up. So the 'enough' treshold will cover a larger number of folks. [ if can optionally put PCI-e SSDs on the back of both GPUs that will only happen faster. ]
People can talk all they want about the storage being in the basement anyways - for maximum drive performance, where required, that's still rather uncommon .
It isn't about all storage being in the basement. It is about bulk and collaborate storage being in the basement. Once bulk gets to certain thresholds putting it all inside of one box increasingly doesn't make sense. Throw collaboration on top it is makes less sense even faster.
Once bulk, typically unchanging, files are separated from temporary/ephemeral working space files that need to be close to memory/CPU these "I need multiple TBs of workspace" claims are those of the relatively few.
