I used to always custom build all my own boxes and carefully tune my Linux installs. For certain things I still do but it's rare now. I use MacOS for my desktop because I used Linux as my primary desktop for over a decade and it always sucked. If anything, I'd say it got worse with time. To much work on the look and not enough work on solid underpinnings. And MacOS has solid developer tools and a Unix command-line. For servers I mostly use cloud services such as Amazon. I went through the stages of having my own server clusters, then virtualized server clusters, and pretty much eventually ended up with a custom solution very close to what Amazon now offers but with less hardware available and at much greater cost. Usually I'm still running Linux instances but I prefer when I don't have to know what the OS is at all. For most my personal computing I actually use my iPad. I even prefer coding from it. Unfortunately I've mostly moved to C# for development and I've yet to find a decent programming environment on the iPad for it. May end up writing my own.

Note to self: get more sleep before commenting....it's losing rotational energy to pushing the moon farther away. Gah.

No, the Earth really is slowing down very, very gradually. The tidal forces from the moon is slowly leeching off rotational energy from the Earch (as heat). See here: http://en.wikipedia.org/wiki/T...

This excellent blog article describes a technique developed by Judea Pearl decades ago to do exactly this. Would be interested to understand how this is different/better.

So long as programmers feel the need to write C to show off what studly coders they are we'll be stuck with C. We'd all be better off if we could spend less time fixing C-related bugs and concentrate on making sure safer languages were just as fast and functional. I'd suggest C# as a better alternative but recently I've been discovering how stupid its handling of byte order is. It's not a bad language except the amount of idiot Microsoftisms it has. C++ is just as bad as C. Objective-C is a mix of genius and insanity. Java is its own set of kludges. Python is nice but slow. Go isn't enough of an improvement. ... Not sure we're ready to replace C yet but we should get ready.

I've often wondered about the possibility of not re-burying the trench: make the trench shallower, cover it with a walkable grate, and just leave it that way. Sure, the grate will get covered by leaves, and the trench will fill with water (have to have a way to drain that), but those seem like minor problems. The cable would be shielded from the vast majority of problems (falling branches, cars hitting poles, squirrels). And since it's just a grate covering, it's just as easy to find problems & service as if they were on a pole. I'm sure I'm missing some reason why this isn't feasible, though...

"Science data" as opposed to "telemetry data". It's a bit of a jargon term, but makes sense to me.

If a person discusses their own medical history with someone else, HIPAA does not apply. If they talk about it in public and someone overhears it and somehow uses that information, including a marketer, somehow, HIPAA has nothing to do with that.

Now, there may be an expectation of a certain amount of privacy when discussing something over email, but if that information is somehow obtained -- even by a breach of the email servers, and assuming neither server/individual is a hospital/doctor/insurer/etc or an employee of such -- HIPAA does not somehow magically apply. Just because it is medical information, it is not immediately protected by HIPAA.

I'm not sure why you're hung up on PHI, I didn't see anything related to medical information mentioned yet. But that's what business agreements are for: to share or pass the blame.

Even the Beats knock-off STREET ANC cans from SMS have the noise cancelation that is reviewed as being as good as the QC line, while being cheaper, and having a different mix of connectivity options and styling choices.

Bose has got to start differentiating themselves or innovate instead of leaning on brand inertia.

This is true with one big caveat: the kernel still comes from the cromeOS partition, not the linux partition. I learned this the hard way with my chromebook....I could never get it to a 2.6 Kernel (never mind 3.x) because the system had actually booted the kernel from the chromeOS partition, but the rest of linux from my ubuntu partition.

I'm beginning to think that the lack of difference between the party policies isn't that they're the same party...I think the institutional attitudes of various agencies doesn't change with government rotation because most of the employees of the agencies don't change. That can be as good (if the party you disagree with is in power, it's hard for them to gut an agency they don't like), and it can be bad (an out of control agency can almost do whatever the hell they like, since they know they can outwait any mangement they disagree with).

I'm not sure how to solve this one, though...if you clean out the entire upper echelon of an agency at administration rollover, then you risk seriously politicising even the most bland agencies. On the other hand, some of these agencies clearly need an attitude adjustment, and I really do think the attitude problem is endemic to the entire culture of the agency, not just their leadership.

Maybe a max term for any federal employee that they can't work for any one agency for more than 10 years?

Cron has specific semantics about batch scheduling of tasks or periodic, non-overlapping tasks. It runs them in a particular execution context, and I like knowing that it logs it an very identifiable way (both through the audit log and cron logs). Syntax of a cron job file is very low on the totem pole of things I care about when it comes to batch or periodic tasks. This is not a trivial task, as you say, and it deserves a closed system, especially if it must be targeted by cross-platform products needing such a facility.