L3 cannot deliver the data to Verizon since there is not enough connectivity between L3 and Verizon to hand the data over at the interfaces where L3 is attempting to do so.

Verizon does not want to put all their bandwidth eggs in L3's basket just to accommodate Netflix so they want Netflix to either peer directly or force L3 and its other CDNs to re-route traffic through other Verizon peers.

Depending too heavily on a single upstream provider is not sound business practice and Verizon wants to avoid getting tied up in that sort of relationship with L3 mostly due to Netflix.

It would also add ~2mm to thickness and 10-20 grams for the sliding mechanism, the keyboard, stiffening structures and bottom cover.

And there is the sliding mechanism as an additional mechanical and electrical point of failure.

I prefer physical keyboards over on-screen as far as typing goes but the design and cost compromises, not so much.

Verizon's subscribers would be able to get the content they want if Netflix routed traffic to Verizon through other peers than L3.

Verizon upgrading their connectivity with L3 to infinity and beyond would not be good business practice since Verizon would be screwed the second Netflix decides to change their transit mix to move away from L3 and then Verizon would have to start over.

It makes sense that Verizon would want to force Netflix to diversify its peering.

The problem with the 'fastest' route is that it may not be the CHEAPEST route.

If L3 really wanted to relieve pressure on their bottlenecked links to Verizon instead of trying to turn this into a PR exercise to make Verizon cave in, they could re-route traffic through Verizon's other peers with under-loaded links but that could cost L3 more money and possibly cause peering disputes with those other peers.

Only problem with that is Verizon has TONS of under-used transit capacity with other networks - when Verizon posted their thing about peering points with Netflix's partners, they also mentioned that their transit to other networks at times where Netflix was hitting 100% was only ~40% on average.

So, Verizon would have plenty of transit capacity if it was spread more evenly across all the peering Verizon has.

And the reason why using VPNs or other work-arounds works is because they cause traffic to pass through one of Verizon's under-used routes instead of the overloaded routes between Verizon and Netflix preferred by Netflix's CDNs and transit providers.

If the Netflix traffic distribution was more evenly spread both geographically and across available peers, Verizon would have much fewer reasons to object to upgrades.

The overlap region between ideal track centers is still somewhat of a gap; albeit not a dead/silent one.

There will still be some residual information in there due to head deviations from the ideal path and when solving complex puzzles like reconstructing overwritten PRML blocks, every little extra hint counts.

I have little doubt it is possible to recover at least some data from PRML drives that have been erased once, maybe twice. But the process would probably require the precision and sensitivity of something like an atomic force microscope, which would be a "little" too much time and effort for the casual identity thief or creep.

If you want to use sonar to map the ocean, you measure multi-path delays, phase shift, attenuation, etc. using a receiver array too.

Radar does much of the same as well: you need the delay to calculate the distance, phase/doppler shift to calculate the speed and heading, signal strength to estimate the cross-section, etc.

You can go way beyond just measuring drift from nominal values. With a distributed receiver array, they could probably use multi-path delays, reflection, attenuation, etc. from thermals, air currents, moisture, etc. to calculate temperature and other parameters almost anywhere within the network's airspace.

There is still a gap between tracks in today's drives; just nowhere near as much so whatever signal might be available on the fringes will be much weaker.

The real killer for PRML-based drives is that to cope with the amount of noise the head receives from nearby tracks, the coding itself relies on statistical analysis to reconstruct the data. Whatever signal might be on the fringe will be some blend of the old data under the current track, the new data, data on the tracks to either side, the previous data on the tracks to either side, etc. There may not be enough signal left in-between tracks after a full PRNG erase or two to recover anything useful.

That's what magnifiers and optical microscopes are for.

Even if you cannot individually identify pixels, you can still notice the marginally sharper text/line-art edges, smoother gradients, reduced stair-casing along polygon edges in 3D applications, etc.

But beyond 300dpi at typical tablet/smartphone reading/playing distances, I doubt that many people would really care about difference between 300-350dpi and 400-450+dpi.

There isn't much of a point in pushing display densities much beyond 300dpi for hand-held applications since most people can barely tell the difference but Apple was the first one to make a big deal out of it. Announcing higher resolutions sold many of the previous models and will sell future ones too - with the progress stagnation that has hit smartphones and tablets for much of the past two years, higher resolutions and the IGP power to drive them are almost the only two things that have improved by a fair amount.

Personally, I would be far more interested in seeing 1920x1200. 2GB RAM and 32GB SSD become the norm below 10"/$300. Right now, the N7-2013 is still practically the only tablet below $300 with those specs and price point. Most other devices below $300 launched in 2013-2014 have specs closer to the more ancient N7-2012 or sometimes even worse, which is sad, a shame and ridiculous. If analysts and manufacturers are surprised to see tablet sales dropped, they have not been paying attention to the crap the market has been flooded with for most of the past two years; no surprise sales are dropping when there is nothing worth upgrading to on the market.

The only differences being that with cell towers, the receivers and transmitters are controlled by two independent parties and the signal itself was not intended for that particular purpose.

You know exactly where the tower is, you can easily reconstruct the RF signal as-transmitted by receiving the bitstream yourself and calculating the original signal as-sent. All that is left to do is compute the correlation between measured signals across your receiver network and weather along the receiver-transmitter path and their immediate surroundings.

This is a bit like a radar version of passive sonar.

Cameras can fail, the communication links between the camera and whatever serves video stream can fail, the displays themselves can fail, power to the displays can fail, etc.

If you do a simple HUD overlaid on top of the conventional glass window, at least you are not completely screwed even if there is a compounded instrumentation failure as long as you can still get visual cues somewhere such as the moon, stars, city lights, runway and slope guidance beacons when landing, etc.

As long as you still have some degree of flight control left, line-of-sight, skills and luck, you can land a plane even with extensive instrumentation failures - there have been many seemingly impossible yet successful landings.

Having cockpit displays/HUDs to aggregate, complement and supplement existing instruments is fine but for completely replacing direct line-of-sight as a backup? I don't think so.

Geoblocking and all the unnecessary middlemen that try to use it to secure their artificial geographic monopolies need to die if they refuse to compete globally.

To be fair to local online vendors though, there would need to be an international standard for sales taxes such as one harmonized rate per country so international vendors would at least not need to deal with the countless regional variants within countries when charging foreign taxes. Another possibility would be to let financial institutions charge domestic taxes on the taxable part of electronic purchases since they are well-versed in the tax codes of whatever regions they do business in so vendors would not need to worry about managing international taxes at all.

If developers do not want to worry about the underlying hardware, all they need to do is stick to Google's developer guidelines and use Java. Let the JRE and native recompiler abstract all the hardware-dependent stuff. Not quite as compute/power-efficient (at least in theory) but from what I have seen, there seems to be tons of developers who waste tons of cycles regardless of portable vs native anyway.