To say there is no market for antibiotics because they are used for only a short time is only part of the problem. It wouldn't matter if a lot of people were continually getting infected.

Sulfa and penicillin went into mass production because of WWII. The huge number of battlefield wounds and the resulting infections created a large demand and there was a crash program to manufacture these antibiotics.

After the war, mass production continued and in many countries antibiotics could be purchased over the counter. While this eventually contributed to bacterial resistance, there's no doubt there was a huge market for antibiotics and it prompted the development of many different antibiotic families.

So the question must be asked why aren't the same economic factors still causing the development of new antibiotics? While it costs more to develop a new antibiotic today, there's a lot more people in the world to justify it. The answer is that while certain diseases have become resistant, for the most part the old antibiotics still work and there just isn't enough people needing new antibiotics to justify the development expenses.

I agree with you that the real reason for RT was that Intel wasn't delivering low-cost, low-power chips that could compete with ARM (I have a friend who works at Microsoft who says the same thing).

But the point of RT wasn't that Intel wouldn't produce a chip for mobile, it's that they couldn't. Intel has always recognized the huge growing market for mobile, and they always wanted to produce procecessors for it.

But the x86 architecture has a lot of stuff in it (compatibility modes, security protections, etc.) that just aren't needed in mobile devices but it's hard to strip them out. It also needed a lot of power management stuff added on. As a result, for any given fab process an ARM chip would be cheaper and use less energy than a x86 chip. Intel's answer was to use their superior cutting edge fab technology that wasn't available to competitors to produce a competitive chip. However, their latest fab technology was delayed until just a few months ago, and their previous fab technology didn't give them enough of an edge.

As it is, ARM chips manufactured by Samsung and TSMC are only only one step behind Intel today in fab technology. Given how hard it is now to reduce the size (as evidenced by the unexpected year-long delay for Intel to get its latest fab technology ready), it's not clear that in the future Intel will be able to maintain its lead before everybody else catches up sooner or later.

Yup, during the digging of the Erie Canal in the early 19th century there were around 1000 deaths from malaria.

Some types of insurance such as home insurance still have a need for agents because when disaster strikes it's very helpful to have someone there to get you immediate aid and help you through the long process of rebuilding.

Other insurance such as auto insurance don't need local agents. While there are companies that do employ agents, there are plenty of low-cost auto insurance companies that don't and Google would be nothing new.

Some insurance such as life insurance takes sales people to use high pressure tactics. Most people would never buy life insurance on their own because nobody wants to think about dying. Google might make some headway here because all the data they collect would help them better figure out the risks for a particular person. But the life insurance industry is no slouch when it comes to data collection and analysis. And a lot of profits in life insurance come from investing the premiums, something Google would have no advantage in.

During the winter I keep my home at 55 degF and don't even bother with sweaters or other outer wear unless I'm not active (e.g., watching TV). Given that it was -11 degF last night, it makes a big difference in the heating bill. But mostly it's what I prefer.

Yes, farmers today spend some time in modern farm equipment. But most of that is only during the few weeks of planting in the spring and harvesting in the fall. The rest of the time it's still a very physical, demanding job. Try doing a major overhaul on a tractor or rounding up cattle for branding and vaccination if you think modern technology makes farming life easy.

But all the changes of modern agriculture happened too recently to have any evolutionary effect. Even with an ox or horse pulled plow, it still takes a great deal of strength to plow a field, and you walk just as far. Harvesting with sickles is still physically demanding. And at least 90% of people lived on farms until at least the middle of the 19th century.

The abstract says "Thus, the low trabecular density of the recent modern human skeleton evolved late in our evolutionary history, potentially resulting from increased sedentism and reliance on technological and cultural innovations."

The authors obviously know nothing about the history of agriculture. It wasn't until about 150 years ago that technology finally improved farming to the point where the dawn to dusk drudgery was reduced. Until then farmers easily walked as far as hunter-gatherers everyday, and probably butchered more animals (another task requiring great strength). Until the late 19th century almost everybody worked on a farm, so there wasn't nearly enough time for evolutionary changes to occur.

Oklahoma and Colorado share only a short border, and there are no major roads connecting them. What makes Oklahoma think Colorado is to blame for all its marijuana issues?

In the first generation of SMR drives there was no support for sector remapping and garbage collection in either the OS or the drives. Random write performance suffered a great deal as a result.

But the next generation of SMR drives will have an address translation layer similar SSDs which supports sector remapping and garbage collection. Much of the firmware will be adapted from what's in SSDs, but yeah, I'd be suspicious of first generation address translation firmware in SMR HDDs even though both WD and Seagate have worked on SSD firmware before and have experts in the technology.

The next generation of SMR drives will also have areas of non-SMR (lower density) tracks to support random writes without the read-modify-write penalty. A new ATA and SCSI command set is being worked out where the host can tell the drive which addresses will likely contain random write/read data, which will likely have sequential write/read data, etc. This will require cooperation from both the host and the drive to implement but should reduce the random write SMR penalty.

I'm sure you meant SMR, not PMR. PMR doesn't affect read or write times one way or another. But yes, SMR works a lot better for sequential writes than for random writes.

In the first generation of SMR, no garbage collection or sector remapping was done. The shingled tracks were separated into "bands" of a few tracks, and when a random write was done all the tracks in the band past that point had to be rewritten. These drives just bite the bullet on random write performance. Sector remapping and garbage collection are scheduled to be done in the next generation of SMR drives, I don't know if they're out yet.

Seek time is not the same thing as the time needed to actually read or write data. It's just the time needed to get the head to the track where the data resides. Once the head is there, the drive still has to read or write the data.

With these SMR drives, on writes there's the read-modify-write penalty of having to rewrite entire tracks, and this is not reflected in the seek times.

I have worked at both drive companies and large storage vendors and can back up everything you say. Among other things, drives for storage arrays often have larger RAM caches and have special vibration sensors because large arrays of drives produce unique vibrations.

And yes, large customers often get special attention from drive companies. They get samples long before everybody else. While part of it is to keep large customers happy, it's also because these customers spend the time and money on doing a lot of testing, which as you say includes environmental as well as reliability and performance. Large customers have staff experts in drive testing and can often pinpoint exactly what caused a problem, something retail consumers can seldom do. For these reasons it's worth it for drive companies to give large customers special treatment.

No, every drive company goes through models that are worse than others. You look back 20 years and you'll find reports of bad WD, Seagate, IBM, and other drives. It doesn't mean the company is cashing in. Drives are extremely complex devices and the technology inside them is rapidly changing. That means one drive model may use new and radically different technology than a previous drive even though from the specs the capacity increased only a little. On top of that, different drive model families at the same company are done by different teams. Some teams are just better than others, and often upper management doesn't know there's a problem until a bad drive is released. There is a valid point to be made that at least Seagate should have tested these drives better before releasing them from manufacturing. The problem is that if they totally messed up, it's obvious from the beginning. Beyond that, things become dicier. Engineering samples are not made on the same production lines as released drives. A failure during manufacturing that wasn't present during development may not show up until a lot of drives are made. Companies do testing on manufactured models, but there's still pressure to push the drives out. If they cut corners on testing at this point, a bad model still gets released to the public. But when it happens it's seldom deliberate. When I worked at a drive company I saw releases that were delayed by months when it became obvious there was a late-stage problem.

Drives in USB enclosures do support SMART data reporting using a protocol called SAT which embeds ATA commands inside of SCSI commands over the USB interface, it's just that a lot of system software doesn't support that. I have one of these drives, and I agree they're complete crap (mine died after a year). But the SMART data does say the drive is very bad.

What the article didn't factor in is that new car sales are currently almost at pre-recession highs, and still growing (http://www.autoalliance.org/auto-marketplace/sales-data ). Cars that would have been purchased in 2009-2011 are being purchased now. Given that most cars are on the road for at least 10 years, you have to factor in that in the next decade we are going to have a younger, more efficient automobile fleet than we would have without the recession.