Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror

Comment Re:flavor? (Score 1) 124

Here's an exercise for you.

Grow a tomato plant. When you have large, hard, green tomatoes, pick a few and put them in a brown paper bag.

Eat those tomatoes when they ripen. Compare them to tomatoes which ripen on the vine. As well, hold onto a tomato picked at optimal ripeness, and a tomato picked when green; time how long before each begins to soften, wrinkle, and rot.

You want tasty tomatoes? Live closer to the farm.

Comment Re:Perfect Tomato? (Score 3, Interesting) 124

Tomatoes are harvested green and shipped. They don't develop the sugars that make them sweet because they're not ripened on the vine. Were they vine-ripened, shipping them to remote states would land you with rotting tomatoes.

In practice, tomato flavor is related to the distance shipped from the harvesting operation. The logistics to get tomatoes to your table with less time between picking and purchasing are responsible for providing better flavor.

Comment Re:Differential and management are not the same. (Score 1) 200

So, why going back to a physician once you have the best diagnostic possible?

Because humans are able to access and interpret a wider array of information. We see patterns in how people walk, how they breathe, and how the scientific analysis of data provides a particular diagnosis in detectable situations where a different diagnosis is correct.

Sometimes it's 0.01% likely that other diagnosis is correct, given the data you understand; and some available data you can't yet name or quantify identifies when it's almost-guaranteed the alternate option is the correct one. We call that "intuition" until we can identify the variables, quantify them, and feed them into a machine or training course to include them in the documented diagnostic criteria.

Doctors and nurses will begin to notice patterns of the AI giving a certain list of likely diagnoses when specific symptoms are present when some different diagnoses is correct. They'll notice that if the diagnosis list or the symptoms are different, then the alternate diagnosis is not correct. In other words: they'll start recognizing error conditions and failure modes, and identifying not just when not to trust the machine, but what the machine should have selected for.

Working rigidly from a procedure means ignoring failure modes no matter how consistently you can spot them. Humans do the equivalent of modifying the deep-learning algorithm on the fly.

Comment Re:Differential and management are not the same. (Score 1) 200

is sufficiently difficult that mere error isn't generally considered a matter of culpability unless it's accompanied by negligence or recklessness

In other words: medicine is based on imperfect information, it's impossible to correctly diagnose and treat anything, and people should stop expecting doctors to get everything right and focus more on getting things less-wrong.

I'm a fan of exploratory pharmacology, although I don't know if that's considered ethical. I don't particularly care, so long as it's safe.

I went in for psychiatric care due to ADHD, and eventually discovered my insomnia was severe (I thought I was getting 6.5 hours; I was getting ~2 hours for over a year). I don't have attention issues when well-rested; I'm fidgety and have impulse-control problems. That means my ADHD is inattentive-type when sleep-deprived and hyperactive-type when well-rested. I've also determined the original issue (starting things, working for a few days or weeks, then never finding the motivation to continue) is rooted in anhedonia, and is textbook major depressive disorder even though I don't feel depressed.

So here's the fun part.

My first try was Modafinil, because I did not want amphetamine, because I knew I had sleeping issues. Modafinil worked great, and then messed me up bad (extreme depression) after two weeks. I went back and checked out how I was really sleeping for the prior several months and determined my FitBit was reading me as asleep when lying in bed for hours awake; switched to the Sensitive tracker those days and it read my sleep time accurately--at damned near nothing. Yeah, don't use Modafinil to stay up for 2 weeks straight; and don't use Modafinil if you otherwise don't sleep for 2 weeks straight.

I had tried Phenylpiracetam (not scheduled, not approved; NDRI) to no success prior to getting a psychiatrist, but only did that for a week because it made me really high. A talk with my psychiatrist determined that "really high" was pretty much "there are these feelings when good things happen and I've never felt this before and it's a euphoriant and I'm high as shit!" I told him I needed to get some counseling and figure out if that's anhedonia or just me being high, and he asked a bunch of questions and determined ... it's anhedonia. Put me on Amphetamine.

Did not like Amphetamine.

Amphetamine hits me really hard. At 10mg XR it makes me anxious and depressed; at 20mg XR it causes severe overdose symptoms (I pissed brown and lost 6 pounds in one day, including muscle mass); at 15mg XR the anxiety goes away and I feel mildly depressed. If I take one, I don't sleep for at least 26 hours. I stopped taking them while taking Belsomra (Suvorexant), which allowed me to sleep but didn't make me tired.

Belsomra is hard to get covered by insurance, so I tried Eszopiclone. I was high as shit 24/7 and nearly drove my car into another car 20 hours after the last dose, but it didn't help me sleep. 12 days in I stopped taking it, went through really bad withdrawal for one day, and decided GABA drugs are not for me. To hell with that.

It goes on and on. I've determined serotonin drugs are not a thing for me--that means all those SSRI anti-depressants are a no-go. SAM-e (at 800mg) and Atomoxetine (at 80mg) both cause serotonin mania; Atomoxetine at 60mg causes serotonin-related problems (tachycardia, fatigue) that go away at a split 25mg dose. Atomoxetine is an SNRI that primarily occupies NET; once NET is 98% occupied, an increase in dose rapidly occupies SERT and jacks up the level of Serotonin in your brain. I like Atomoxetine at lower doses, as it eliminates the excessive poor behavioral impulses and leaves me with something I can control; it also allows me to sleep, so I don't need any sleep drugs.

I've been pushed into suicide-grade depression, driven insane, and outright poisoned. As a patient, I can handle it: I'm extremely psychologically-resilient and have a strong tendency to dissociate, so even side effects like suicidal impulse (as opposed to just ideation or fantasies of self-harm) are generally harmless to me. That means I can take a rapid battery of switching out various drugs, while less-resilient patients will need a higher degree of caution and monitoring--which is how psychiatric care is approached now anyway.

My point, however, is that these limits don't just outline what to do in earnest attempt at treatment. If the doctor wants to give me a drug as an exploratory measure to get a better understanding of what's going on in my brain, I'm up for it. I've learned serotonin is no good for me. I've learned dopamine will keep me awake, but small amounts of the right NDRI will enable my rewards mechanism and operant behavior so I can self-activate. Each failure reveals a little information; information is valuable in a field where literally nobody has any clue what they're doing.

The doctor can't crack my head open, point out where I need dopamine and norepinephrine, and give me a drug to give exactly the right amount of dopamine and norepinephrine in the right places. One NDRI will give me depression and anxiety; another will make me respond to rewarding stimulus by reinforcing my behavior (having almost no reinforcement response is really bad, by the way). They don't actually know precisely what's wrong with you; they have a list of patterns that tell them what might be wrong with you and what might help, along with what those things do. Can't blame them for getting it wrong before they get it right.

Surgeons can't know what your CNS is going to do when they poke you; you might go into cardiac arrest at any time. You need sedation. You might have an allergic reaction to a drug and die. You get cancer and you appear to have the flu, but it gets worse. Biopsies and other screens show false-negatives and may say you're clean for cancer. It's really hard to separate out one set of conditions from another when the information you have tells you it's 37% likely X is wrong, 29% likely Y is wrong, 21% likely Z is wrong, and so on.

Some doctors are in way over their heads and need a second opinion, and should know better. Some doctors just don't give a shit, and should be hanged. Pretty much every doctor you encounter actually has no earthly idea what's going on, and is doing the best job possible with the full breadth of collected human medical knowledge. Unless you have a broken bone, don't expect them to identify the problem correctly 100% of the time.

Comment Re:Programming is like.... (Score 1) 355

Yeah, this is another elevation of someone's personal hobby to a personal fantasy. The opener got me:

For starters, the profile of a programmer's mind is pretty uncommon. As well as being highly analytical and creative, software developers need almost superhuman focus to manage the complexity of their tasks.

Becoming a programmer requires putting in effort. You have to put in the effort to study, learn, and understand the field of computer science and the application of programming processes. It's not just "I read a book on C# and now I can code!" It's not just knowing design patterns and other trivia. It's knowing why those things are there.

I got a project management certification. I break down deliverables into hierarchies of deliverables. In programming, we have all these design patterns and concepts like encapsulation so you can break down large, complex bodies of code into isolated blocks. The whole microkernel thing separates processes physically and forces the use of a common communication protocol; modern OOP methods attempt the same with code living in the same source file right next to each other. These are all the same concept: you're localizing each piece to a small scope so you don't make a change internal to one part and have something far, far away break as a result unless you broke the part you changed.

Planning takes a lot of time. Most programmers don't seem to appreciate architecture or management; they all just want to sit down, write a bunch of stuff, get a result that pushes their joy buttons, and keep making changes as they organically grow a complex system. It only works as long as they can keep the relevant parts in their heads and manage to not break the other parts they've already forgotten.

Good project management sharply reduces the complexity of execution (getting all the work done) by breaking down and structuring the work, making it easy to track. In the same way, good program architecture sharply reduces the complexity of implementation by reducing the number of connections between different parts of the code. If your code is tightly-coupled, then a change in how you perform a certain operation can affect the way other parts of the code flow due to changing states besides the output state. If your code is encapsulated, then all client code uses an interface and receives a result without being affected by everything that happens along the way: any change that could affect other parts of your program is a bug fix or a refactor.

Engineers don't build cars such that the rear differential is affected by changes to how the piston rod connects to the crank shaft or the steering pinion is affected by how the power steering module transfers power in response to the driver's input. Engineers build engines, transmissions, power steering modules, and differentials, and plug them together like legos. You replace the vacuum-driven, hydraulic power steering module with an electrically-driven, drive-by-wire power steering module and your rack-and-pinion steering system still works because the electric motor's drive shaft plugs into the same spot the other thing hooked up to.

We build programs the same way now, or else we complain that programming is super-complex beyond the limits of rocket science and automotive engineering.

Comment Re:Misdirection (Score 1) 153

No, it's a stupid point. If it's advantageous for all the big players to take the same kinds of throttling action, then what? A small player enters the market and... what? The market won't just turn over and give them influence; they still have to go through someone else's fabric, and can face rate controls and the like by proxy. Building their own infrastructure is expensive.

Why hasn't MintSim caused everyone to abandon Ting and T-Mobile?

Comment Re: Sue Islam for killing innocents. (Score 1) 136

There is a Constitutional law which says Congress shall make no law establishing a State religion or infringing upon the free exercise of religion. That means Congress cannot make laws declaring a religion of the United States, or banning a religion, or whatnot. Congress also can't give any regulatory body the power to enforce any such thing.

Congress also cannot produce a legal basis for removing bibles, torah, qu'ran, or the Tao from the curriculum of public schools. Congress can neither prevent nor produce a legal basis to order a Federal courthouse to adorn itself with religious displays. Congress cannot produce any sort of law allowing a legal gag of the practice of religion by state or Federal legislatures, either, meaning that the Senate can vote for its proceedings to open with prayer, and any Senator may abstain from said prayer without legal repercussion.

The "Separation of Church and State" doesn't exist. The uneducated make an argument that, somewhere, there's a line in the Constitution declaring that the Government must be 100% secular, agnostic of all religion, and neutral in all aspects of its behavior. This has been used to found arguments against police departments which place crosses at sites of roadside deaths, even though you can't make any such legal argument against the practice even via the Incorporation Clause of the 14th Amendment (applying the 14th would make it illegal for the State to stop the police department from doing any such thing).

Comment Re:Money creation and savings (Score 1) 172

Not true at all as long as you hold your savings in a financial institution. Put your money in a savings account or similar and that money will get lent out to do other useful things in the economy.

Uh, hold on. My money in my bank accounts is spent by sending it to other bank accounts. Billionaires don't carry billions of dollars in cash; it sits in bank accounts, then gets transferred to other bank accounts. If that money is spent, it stays in the banking system, doing exactly the things you said; if it's unspent, it may well still stay in the bank account and provide a basis for fractional reserve lending, and it itself goes unspent.

It seems that the difference between spending $40 million and saving $40 million is that $40 million is whether that $40 million becomes part of some business's revenue stream. In either case, that $40 million remains in financial institutions. So, again, savings that goes unspent is essentially removed from circulation.

This is a direct contradiction to your assertion that savings = removal of money from the economy

It's not. When you leave money unspent, it stops acting as revenue stream, thus stops supporting wages, thus stops affecting prices. The Fed adjusts the money supply to maintain 2% inflation. How does that affect the money supply?

If you're spending your money, then it's being moved between accounts at financial institutions. It contributes to the loan basis for fractional banking. It also itself is a revenue stream for businesses, and acts as... well, money. It feeds into revenue and supports wages.

If you're not spending your money, it's sitting in an account at a financial institution. It contributes to the loan basis for fractional banking. It is not itself a revenue stream and doesn't support wages. Over time, the Fed issues more currency in order to keep inflation at 2% per year, and in the process compensates for the reduction in spending caused by your money idling in a bank account.

In that second situation, the Fed has issued more money; then, ten years later, you take your money out and spend it. It's as if your money (and not all of the loans based on it) was removed from the economy, new money was printed up to replace it, and then you magically poofed your money back into existence.

Your argument has one huge, glaring flaw: spending your money doesn't take it out of the banking system. Money spent is still in the banks to act as a basis from which to issue loans.

Comment Re:Why? (Score 2) 190

It's a complex economical issue that demonstrates that common sense is a logical fallacy, especially when dealing with large and complex systems. As such, it gets stupid people screaming for protection from something they think will hurt them.

Slashdot, like everywhere else, is full of self-indulgent idiots. Even rednecks have complex skills and knowledge other people don't; those of us with more-refined careers, better education, and all kinds of justifications about how smart we are like to forget that anyone without our particular technical skill has some other difficult skill (accounting, etc.). All sides think the others are full of morons and that their own ideals are perfect; we all then proceed to mindlessly throw out baseless and damaged opinions about things we don't understand.

I like economics because it's complex. Economics isn't about having read a book on economic systems; it's about thinking, analyzing, and conjecturing. Generally, in all non-economist circles, economics becomes a political topic where people grab a few choice axioms to demonstrate their position--or just shoot from the hip.

Look at the two opposing arguments. If you replace a man with a machine, then you've eliminated the man's job (forever). Alternately, if you replace a man with a machine, then you've created jobs for people to make machines. Neither is correct.

If you replace a man with a machine, then you've replaced a set of labor-hours to produce a thing with a set of fewer labor-hours to produce the same things. You get fewer machinists and machine operators than people replaced by the machines. The jobs lost to this re-distribute, typically such that you get some machinists, more machines producing more of the things for a lower cost (and price) than before, and jobs doing other things entirely.

That's also complex, takes time to fully render (there aren't and can't be replacement jobs the day jobs are eliminated by new technology because the full of downstream economic effects have to cut through all pressures holding them back), and can happen in a bunch of different ways. Slow technological replacement causes negligible unemployment increases before job replacement kicks in and puts us back on stable footing; rapid replacement causes rapid unemployment build-up and recessions.

Nobody cares. What they care about is the group of people on their side--the luddites and the technophiles. They care that a bunch of people agree with whichever wrong-headed, overly-simplistic ideal they're clinging to.

As a media outlet, if you can scare people, you can get them to come back to you for more news. The ability to see threats is the ability to avoid threats. News outlets that frighten you with terrifying news are perceived as protecting you from harm. They're your look-outs. Bad analysis and sloppy reporting to imply a threat gets you more viewers and more ad revenue.

Comment Re:Horrors! (Score 2) 190

I know nobody likes to read 90-paragraph essays and would rather do knee-jerk reactions either forwards or (as you did) in satire, but I still prefer to do a full accounting to justify the current position.

The tl;dr is that a rapid technological deployment will cause a terrible recession due to unemploying several percentage points of the workforce and driving the remainder to tighten their wallets, creating further unemployment; while a slow technological deployment will cause hardly a whimper, and just drive up consumer wealth.

25,000 jobs exchanged per month is a slow, gradual, easy deployment. We probably won't even notice. We'll notice so little that we'll get a few percentage points richer and still complain that we're poorer and the rich are taking all the money, even as we buy more junk, buy bigger houses, get better healthcare, and generally push up the living conditions of the middle- and lower-classes.

(Seriously, nobody cares about the explosive growth of middle- and lower-class wealth in the past 30 years; whenever I've backed someone into a corner pointing out all the things we buy now and the sheer quantity of consumption on which we live, they defect to an argument that material purchasing power isn't wealth in a conversation about the rich supposedly taking all the material purchasing power. I'm too stubborn to accept that complex, abstract concepts are hard to argue because people can't be forced to look directly at broad effects; if you can't hold up a rock and a grape and drop both to show the rock doesn't fall faster than the grape, they'll just ignore what you're saying and substitute their own imaginative fantasy for facts.)

Slashdot Top Deals

Where there's a will, there's a relative.

Working...