G+ fits my desire for social-networking perfectly: I hardly have to spend any time on it to get what I want out of it. I spend no time whatsoever on the other systems, because they're more cumbersome and demand my time in ways I'm not comfortable with. G+ is the only system that lets me contribute the little amount of time I'm willing to contribute, without being useless. So maybe its users *do* use it for fewer minutes a month -- but isn't that okay? Is there not a market for that? Lots of people probably watch crappy TV -- should we judge other channels based on the fact that they have a few, well-targeted shows, that a segment of the population watches (but nothing else)? Maybe it should be our goal to use these systems less, not more! In that respect, G+ represents an increase in efficiency -- which is a driver of GNP. So it's a good thing. Go G+!

Spoiler alert: Deus ex machina, and he's back in business again by the end of the movie. Doesn't get the girl, though.

So, if it rained on the blimp, would the water fall all the way through? When it flies through a cloud -- is the cloud really flying through the blimp?

My wife's family has a lot of Celiac disease going around. I can promise you that, as patients, they know far more about it than any of the general practitioners they visit (or, in one case, the specialists who went 30 years without properly diagnosing it), who probably haven't read about it since they had a few-minute lesson about it in medical school -- assuming they're young enough. The older ones, I've personally observed, tend to just nod wisely, then leave the room and go look it up. And that's something that (more or less) 1% of the population has -- I feel sorry for anyone with a rarer condition. The system of trusting doctors to somehow "know everything" and everyone else to "know nothing" can't work, the limits of our individual knowledge don't scale that way. (That said, it'd be a good start for doctors to acknowledge their own limits. As engineers, we're quite willing to look stuff up when we run across it, but I don't see that so often in the medical field. Is that the result of being told, right or wrong, that medical school was where they'd learn everything? An over-developed skepticism of anything they read? I really would like to know. Because it seems far too common to have doctors who don't take the time to learn and become experts in their patients' less-common conditions.)

The clinic I worked with (as programmer) once announced that they HAD to destroy records after 7 years. As I recall, HIPAA only required them to keep the records for at least that long -- not keep them then destroy them at 7 years + 1 day. But with that kind of misunderstanding, we can't trust our doctors' offices to be the keepers of our data.

Also, this isn't just about us keeping and reading our own data: it's also about making it easier to go from doctor to doctor. Sure, a hospital/ER might willingly give out data to patients -- but is it in a format that's easy to import at all other doctors' offices? No? Well damn. How does one doctor know what else you're seeing doctors for? You ... tell them. In that painful intake form, that's hand-written, every time you come into the office. My sister went to the trouble of taking all her records from other doctors (painful to get) to a new doctor, only to have him dismiss all of it and start his diagnostic path from scratch. Why bother to read through someone else's notes? Pshaw! Just start over. No reason to save time by not repeating experiments, when a patient's well-being is on the line, and they came to you because previous stuff didn't work.

We should have our data. And we should be able to present it when we go to any doctor, to bring them instantly up to date on your history -- generally as a patient and specifically to the conditions you want treated.

The closest I've seen is this: vets' offices. Maybe it's thanks to a monopoly in the vet industry, but quite a few of the vets we've seen over the years (as we moved from state to state) used the same software. They could read the print-outs we brought from other vets' offices (we had to keep all that in a folder ourselves, nothing electronic) because it was in exactly the format they were used to. That's at least a step up.

France, by the way, uses (used?) a nationally-issued "carnet de sante" in which all your doctors would write their notes, at least for kids. Standard format, all vaccinations and other procedures and observations recorded, dated, with doctor information in case a phone call was necessary. And it was carried by the patient. I don't recall french doctors' office having large shelving units full of old patient data -- they relied on the patients themselves to bring that back in each time. And they took the time to re-read old notes, both from themselves and from other doctors. It IS doable.

Let's not forget that Siri uses Wolfram Alpha, though clearly not all of it (having tried them side by side.) That the two share some quirkiness isn't unexpected.

You know what? Just for you, I'll apologize. I assumed that, like kids these days, you didn't know the difference between those data-structures, and call everything a "hash table" when you mean, at best, "associative array", usually backed by a tree. [We've been interviewing recently, candidates haven't exactly been passing with flying colors.] Congrats on knowing your data and having an efficient solution to the specific problem. Pot and kettle are done, though.

"If you find something is already in the hash"

How do you think that algorithm works, exactly? The reason it's efficient is because it's stored in an index (b-tree / red-black tree), which involves non-linear (though fast) insert operations. Whether you sort as you insert, or sort after inserting, it doesn't matter: it's not linear time. Also, your algorithm fails to consider hash collisions, unless you're doing extra processing on all "similar" items (multiset or multimap) to verify equality, not just equivalency. That algorithm would still involve quadratic/2 (no need to compare a vs. b and b vs. a) comparisons between hashes-to-same-value entries, though at least you'd only have to do it for the few collisions you come across.

Maybe if you had ever been forced to *implement* some of those containers, from scratch, you'd know how they work. Some of us actually had to do that, thank you very much. If you do ever come across an implementation of 'DISTINCT' that works on unindexed data in linear time, please let all the major database vendors know. They'll be *very* interested.

5. Goto 1.

Wait, I think I found the *real* problem ...

(Example: I found a quadratic algorithm used to deduplicate an arbitrary large table in there, when a hash-table, as available in the Java libraries, would do the job in linear time)

Err, no? You could gather the hashes in linear time, but you'd need to sort them to efficiently find the duplicates, and sorting isn't linear-time. It'd be faster than quadratic, yes, but let's not get carried away. Also, please don't pull data into memory in JAVA for this, if you can have the database do it.

Morale of that is that 100 incompetent people
Moral.

[evidence needed]
[citation needed]
[explicitly stated allegations needed]
[ad hominem needs review]

Why is it OK for Google to steal their ideas and then give the resulting product away?

Misunderstanding of patents: you can acquire a patent even if you're not the inventor, and you can negotiate (without actually litigating) with it even if there exists prior art that would theoretically invalidate it on the basis of the cost of legal action; two people can invent the same thing independently without looking over each others' shoulders (it's quite common). The fact that company A has a patent for X, and company B does something resembling X, doesn't mean that B has, in any way, stolen from A. In fact, if you want to copy off of someone else's work, patents are possibly the most difficult way of going about it.

The companies involved have plenty of other sources of revenue with which to recoup the costs of developing technologies like the mouseless mouse click. Invalidating patents doesn't mean invalidating software licenses for entire software products, nor does it make hardware sales disappear. They'll be okay, I promise.

Patents were well-intentioned, but not designed to perform to expectations. They still don't protect small inventors. There's no useful definition of "reasonable" license fees. We're moving in the direction of first-to-file, which screws independent inventors who don't constantly troll the patent lists to see if their idea was previously patented.

Agreed. My reading is that the GP is narrowly correct: CSS can't "completely" reskin a site, because it depends on the underlying html which does have some structural semantics to it. XML data + XSL templates (possibly several, for different end-devices and/or user agents) + CSS would more closely fit his goals, but the XSLT approach wasn't nearly as widely publicized as CSS, so I don't completely blame him for missing that boat. Even today's popular toolkits encourage "backing pages" for data, but the layer above that is generally straight to html, not to XML for later transformation.

While Firebird supports update or insert, who else does? Yes, the merge syntax is common, but that's a lot more annoying to use for day-to-day single-row updates in an OLTP environment. Also, what does "agile" have to do with it? Or did you not mean it in the project-management sense?

That's a lot like the PL/SQL-compatible variant of Firebird, called Fyracle, which is still based on Firebird 1.5, not the current 2.1 or 2.5 series.