whatever is above them in downtown Seattle.

Five city blocks (low-rise -- doesn't look to be anything over eight stories tall), a quarter-mile of the Alaska Way Viaduct, the entrance to the downtown ferry and water taxi docks, and two entrances to one of the larger docks at the Port of Seattle. If the tunnel is deeper than I think, or the soil is more liquid, add another seven city blocks (also low-rise), one park, the ferry docks themselves, part of the Port of Seattle dock, and maybe the football stadium.

Yes, disassembling from behind would be cheaper, if it worked. If anything went wrong, it would be far, far more expensive.

I calculated from the other end: assuming that the Bitcoin market represents an efficient market (ie. sale prices are only slightly above production prices), and that the marginal cost of mining hardware is 0:

- A bitcoin sells for $584
- Mining a block generates 25 bitcoins
- 144 blocks are mined per day
- 63000 transactions per day

584 * 25 * 144 / 63000 = $33.37 per transaction.

And do you perhaps think point-of-sale systems and credit/debit card systems or wire transfers require none?

The Bitcoin network uses about $35 worth of energy to process a single transaction. Now, I don't know how much energy a single credit card transaction uses, but given the transaction fees that processing companies charge, I'm willing to bet that it's far, far less than $35 worth.

OTOH, the traditional CPU/mobo setting is a little problematic; first you put the most heat-concentrating element in the middle of everything, and then later you realize it needs cooling.

The theory is that the exhaust air from the CPU heatsink spreads out to parts that are more heat-tolerant but still need active cooling, such as the voltage regulators. A VRM that can operate at 100C without trouble can be cooled just fine with a slow flow of 50C exhaust air from the CPU cooling system.

In practice, people have found that a front-to-back airflow, preferably ducted, is quieter and more effective than a mix of back-to-front, blow-down, and turbulent airflows. It does, however, require actual engineering work, rather than just attaching a bunch of fans to everything.

*Exterior* temperature? My current heatsink can manage a CPU temperature below that, with the fan at idle. Why would I want to downgrade to this thing?

Don't tell the HURD people -- they'll change which microkernel they're building around, again.

How big is big enough that nothing but Oracle will do? Facebook is on MySQL, Wikipedia is on MariaDB and Google is using Bigtable.

It's more the nature and size of access rather than the sheer volume of data. Facebook and Wikipedia both act on small portions of the overall dataset, Wikipedia additionally is a read-mostly workload, and Google's access patterns aren't suitable for a relational database.

I don't know about Los Angeles or San Francisco, but Phoenix's water usage has been dropping as a result of replacing citrus orchards with subdivisions.

There was a brief period (roughly 1993 to 1995) when copy protection worked to stop small-scale piracy: around the time when CD-ROM drives first became popular. If you could stuff a CD full of game files, you had a game that could not be economically pirated, because copying the CD required either a dedicated hard drive to store the data (hundreds to thousands of dollars), a hugely expensive CD recorder (tens of thousands of dollars), or a CD stamping facility (millions of dollars).

The kit includes eight 8 x 8 x 5 inch Enphase inverters weighing 6 lbs each

Which gives a power density of 0.78 watts per cubic inch. The Google challenge calls for a minimum power density of 50 watts per cubic inch.

Starting on page 12 of the report is a series of maps showing the changes since the 2008 report. Of note:

* The South Carolina seismic zone has been displaced southward by about 50 miles.
* The New Madrid zone has changed shape, with some areas seeing a substantial reduction in estimated earthquake risk.
* The risk zones in California are more sharply defined.
* The risk for the central Rocky Mountains area is higher, but still relatively low.
* The earthquake risk estimate for coastal Oregon has been reduced.
* A new seismic zone is present in Oklahoma, reflecting whatever is causing the massive increase in earthquake rates there.

The majority of passenger trips in the US are either less than 50 miles, or more than a thousand, with almost nothing in between. At the short end, the flexibility of car travel beats the cost reduction of rail; at the long end, the speed of air travel beats rail.

The only exception to this is the BosWash area, where -- guess what? -- Amtrak is able to provide profitable rail service. It's not motivation that keeps the US from having good passenger rail service, it's geography.

The random number generator should not be seeded only by a PID.

The PID is used as an absolute last-ditch fallback in the case that no other sources of randomness are available. In order for this to happen, /dev/urandom needs to be inaccessible, the KERN_RANDOM sysctl needs to be unavailable, gettimeofday() needs to fail, and clock_gettime() needs to fail.

If you're running on a system that crippled, you've really only got two choices: try seeding using the PID, or use an unseeded RNG. Or follow Theo's advice and get yourself a real operating system.

Soviets made a safety experiment with a nuclear reactor in Chernobyl. It didn't go wrong at all.

Correct. The 1982, 1984, and 1985 tests of using rotational energy of the turbines to power the emergency pumping system all went just fine. The 1986 disaster happened when the operator ignored the test procedure (specifically, the instruction "Reduce reactor thermal output to between 700MW and 800MW. If reactor thermal power output drops below 700MW, abort test and shut down reactor" -- the operator reduced the power to 30MW, raised it to 200MW, and attempted to perform the test).

More significantly, if you see left-to-right motion and say "forward", what percentage do you get right? I suspect there's a bias in videos towards left-to-right motion of subjects (or conversely, right-to-left motion of backgrounds), and I don't see anything in the paper about controlling for it.