I've noticed a fair number of very attractive short women often end up with short men.

It may be that generally speaking people prefer a mate who is similar to them in size -- perhaps there's even some evolutionary biology explanation where small women prefer a smaller mate because it reduces the risk to her of having a large baby that is difficult to birth. Maybe it's some kind of social psychology, a small woman may believe a large man will be unpleasant to mate with because of his bulk or that somehow big men have big penises and would be painful to have sex with.

But whenever I notice it, I find it strange that if male height is some kind of marker for desirability why a very attractive woman who could otherwise gain a taller mate who would come with the all the social and perhaps even physical advantages of height actively choose mates who are not just closer to them in height but below average in height.

I think static linking makes a lot of sense, but you will get a lot of resistance from people who say that it makes patching harder because some vulnerabilities will now require more patching (eg, SSL) because every application will have their own copy.

I think this is debatable in some ways, because it assumes every security issue affects a shared library and not part of the core executable. It also ignores the applications that merge shared library code or provide a system available function internally (often to avoid weird version mismatches or system incompatibilities). These may or may not even have the vulnerability based on what code was merged.

I think to some degree some VM "appliances" start to approximate a kind of application virtualization. Several greatly strip a distribution of unused kernel and system components down to the bare minimum necessary to run the application.

I'm sure some product does this, but it would be interesting to see a virtualization system that could take an application and generate a bootable VM, merging into the VM only the parts of the operating system actually used by the application and resulting in a lighter weight VM that was still standalone and didn't require a specific host OS.

Secret pay agreements create an imbalance of power between employers and employees because they eliminate pricing transparency and allow for non-rational pay inequity, such as racial, gender or merely office politics reasons.

There's no reason people need to be paid the *same* -- different people bring different skills and experience to the table as well as being hired in different points in time when skills may have been more or less in demand.

Two people may do the same job and make different wages, but they're not the same person, either. With transparency, the onus shifts to *management* to justify why people are paid differently, and management has to be cognizant of pay differences being transparent -- they can't deviate significantly without real reasons.

Do you think there's no scheming/bitterness/morale problems *now* relative to pay? Making it secret is much worse.

There was a Planet Money story about this.

A company decided to make everyone's salary open knowledge, posted on the wall for everyone to see.

This would better solve gender pay equity than Pao's no-negotiation strategy. It puts more pressure on management to limit pay decisions to something defensable, prevents employees from pitting against each other for pay and minimizes management's ability to overpay or underpay. Employees know where they stand relative to other employees (and what they may need to do to make more). It motivates better paid employees to show they're worth it and makes it harder for well-paid employees to goldbrick.

The problem with no-negotiation is that for any given hire there are a finite number of employees available to take the job and the best candidate is likely to either be a little better or a little worse than average. Without the ability to negotiate, the better candidates will be less inclined to take the job because it only offers average pay and the below average ones will be more likely to take the job because it pays above what they're worth. You'll end up trending towards below average talent for more than they're worth.

Transparency allows for positioned to be negotiated for and if a given hire has an above average skillset and experience, you can agree to pay them more and won't have to worry about justifying it. The same is true the other way around -- it's justifiable to pay below average, too when you have legitimate reasons of skill or experience.

Pao's strategy is right out of the socialist playbook -- arbitrary price controls, and it destroys the free market's ability to seek efficient pricing. This isn't a political complaint, but an economic one. Most current job markets with "secret" pay agreements now are also bad because they create an imbalance between seller and buyer by eliminating pricing information.

It's also pretty sexist because it attributes a behavioral attribute to gender. I'm pretty sure Carly Fiorina, Meg Whitman, and other Fortune 500 CEOs don't have a negotiating weakness.

The thing is, I think it will be the kind of thing that is declared life by a committee of microbiologists, virologists, chemists and physicists on a split decision.

It will leave plenty of room for the usual cast of religious nutjobs to say it isn't life and it will be the kind of thing that will open the door for endless debate as to whether it really is life.

I think it would take an organism much more recognizable as life and/or intelligent to really be groundbreaking.

Life outside the solar system seems more significant than inside it.

I have a feeling that a find of life inside the solar system will end up being something weird along the lines of a virus or bacteria that chemically might qualify as "life" but is so marginal that it only excites a biochemist.

Plus there's the notion that whatever caused life on Earth might have contributed something to life elsewhere in the solar system.

Life outside the solar system seems more likely to be intelligent (given that we'd have to detect meaningful activity, not electron micrographs or chemical assays you'd pick up on site) and statistically unrelated to life on earth because of the distance.

What exactly is so special about installing solar panels? It sounds to me like pretty conventional electrical and construction work.

Even recreational marine electrical systems can be more complicated, with a mix of solar, wind, grid, generator, battery (12/24/48V) and mixed loads (native, 12v, AC).

Right now? That's been going on for decades. I dated a woman briefly who was trying to get a commercial pilot license and her boyfriend (it was a complicated relationship...) was a pilot for a regional feeder airline and he was making less than I was as a low-level civil servant at the University. And this was circa 1991.

His hours were crazy, too, the kinds of work patterns you'd swear wouldn't be allowed if you asked the random person on the street if pilots should work those kinds of shifts.

The airlines were able to get away with it because they held out the golden carrot of a pilot job on the big planes.

I'm sure this automation idea is being floated by airlines because they want to cut costs. I can't help but believe they see the writing on the wall when oil prices surge again and air travel at the scale we have it now simply becomes economically impractical for the bulk of passengers.

I wonder how much of this objection has nothing to do with the vasty overstated risks but instead is of a commercial nature. Alcoholic beverages are extremely expensive in a lot of places (stadiums, bars, restaurants, events) and sneaking your own in is inconvenient or impossible.

I woner if the real opponents of this aren't people who make money charging $10 for cocktails to captive audiences. How much money do they stand to lose when people start bringing a half-dozen packets to the big game?

How is the drinking control regime threatened when you can't restrict alcohol because of its bulk and liquid nature?

Some idiots will no doubt overconsume it, but they are probably the same idiots that do it now.

They're merely reclaiming their heritage as a penal colony.

And many of those places ended up being good places for underage kids to buy weed or alcohol.

Buy it? If they get utility status within a city, I would assume they would string their own backhaul.

I can't see Google tying themselves to Comcast for any purpose, plus there's probably some strategic long-term value to owning their own backhaul network in a city for future services like wireless or cellular.

What do you suppose Google actually uses for backhaul in its municipal fiber? OC-192, 10GBASE-ER? Depending on how many strands they light I would bet the backhaul capacity is probably way less 100G although I'm sure it's engineered so they can light more as usage would dictate.

I'm sure all rollouts probably assume a ton of oversubscription because the greediest average household consumer is going to be what, 4x video streams with random downloads on top of it?

Did you miss the SSD endurance tests where they abused the hell out of SSDs and found them to be way more durable than the skeptical wags like to say they are?

Given normal precautions like backups, they seem good enough to me, at least reasonable brands like Samsung/Intel. I plan to make my next NAS/SAN box totally SSD based, which, by the time I get around to doing it in a year or so will be even more affordable.

Even if the risk of single disk failure is higher than SSD, performance is so overwhelmingly better that it outweighs the assumed marginal increased risk. Getting the equivalent performance out of spinning rust just isn't practical without high powered controllers with huge memory and deep stripe depths.

I'd actually like to see the economics of consumer-grade SSDs in large, commercial-style SANs given the endurance test results. The money charged for SLC flash disks is crazy expensive from SAN vendors. I have the suspicion that the failure rate of decent MLC disks is probably outweighed by their low cost relative to the upfront cost of SLC.

I think these PCIe flash drives win for raw performance because they have access to the entire device at PCIe bus speeds.

Maybe some ideal 16x card with a dedicated SATA controller per connection would give you RAID-0 performance of 30 Gbits/sec for five disks, but something tells me you'd be limited by the individual SATA limit of 6 Gbit/sec. In the real world, I don't know that any $100 8x RAID card would do that.

But I would also bet that most workloads are IOP bound, not mass throughput bound, and with reasonable amounts of RAID cache on the card there would be little practical difference between the two.