How computers are made... Via friend, reader, and former colleague Simon M.
Younger electronics types may not recognize the components in the photo. The package at the upper left is an eight pin DIP package. Before surface mount took over the world, they were the standard package for most integrated circuits (with various numbers of pins). The other five components are resistors, with one of their leads cut off and the other twisted up. They were the standard package for most resistors before surface mount's rise...
Monday, September 21, 2015
Stuffed in Paradise...
Stuffed in Paradise... You might not be able to see it, even upon close inspection, but the bowl at right is full of salad. Underneath the raspberries, kiwi, strawberries, feta, apple, and candied pistachios is a bed of baby spinach leaves (one of the few kinds of greens I can eat raw without suffering from horrible digestive disorders). Debbie made a wonderful dressing for this from strawberry yogurt, mayonnaise, vinegar, poppy seeds, sesame seeds, and onion flakes. Very tasty stuff! We polished off this entire bowl, just the two of us. Our tummies are very happy right now!
Why do I think AI is overblown?
Why do I think AI is overblown? Take a look at what it takes for the relatively simple problem of extracting a list of ingredients from a recipe. It's a very nice example of something that is trivially easy for normal humans, but intractably difficult for computers.
I especially enjoyed the declaration of victory with an 89% accuracy rate per sentence. That means a recipe with 20 sentences in it would contain, on average, about two errors. That's victory?
I especially enjoyed the declaration of victory with an 89% accuracy rate per sentence. That means a recipe with 20 sentences in it would contain, on average, about two errors. That's victory?
Drugs and capitalism...
Drugs and capitalism... One of the signature human accomplishments is the development of effective therapeutic drugs. The scientists and doctors who were the brains behind this accomplishment have the well-earned respect and admiration of the world's much healthier and long-lived population. Less well understood, and certainly less respected and admired, are the capitalists who made this accomplishment possible – by their pursuit of that awesome incentive called “profit”.
Here's an article that lays out one example of the conflict that constantly arises in this industry. In this case, the price of a particular drug increased dramatically (literally) overnight. Given that it appears in that bastion of Progressivism – the New York Times – it is surprisingly even-handed.
It always surprises me when people are offended by the price of drugs. They'll pay the price anyway, if they need them. That high price is what motivates the pharmaceutical companies to develop the drugs in the first place. Without that incentive, none would bother, and we'd all be worse off.
What to do when someone needs a drug, but really can't afford to pay for it, is a separate question altogether. The problem isn't the high price of the drug, but rather the inability of some people to pay for them. There are numerous experiments going on right now, worldwide, on to deal with that. Here in the U.S., the most common mechanism currently is charitable programs from the drug companies themselves (as mentioned in this article). It will be interesting to see which of these turn out to be most workable. My bet would be on those programs that don't involve the government :)
Here's an article that lays out one example of the conflict that constantly arises in this industry. In this case, the price of a particular drug increased dramatically (literally) overnight. Given that it appears in that bastion of Progressivism – the New York Times – it is surprisingly even-handed.
It always surprises me when people are offended by the price of drugs. They'll pay the price anyway, if they need them. That high price is what motivates the pharmaceutical companies to develop the drugs in the first place. Without that incentive, none would bother, and we'd all be worse off.
What to do when someone needs a drug, but really can't afford to pay for it, is a separate question altogether. The problem isn't the high price of the drug, but rather the inability of some people to pay for them. There are numerous experiments going on right now, worldwide, on to deal with that. Here in the U.S., the most common mechanism currently is charitable programs from the drug companies themselves (as mentioned in this article). It will be interesting to see which of these turn out to be most workable. My bet would be on those programs that don't involve the government :)
Afternoon in Paradise...
Afternoon in Paradise... As we'd planned, we drove the four-wheel drive road from Avon to Mantua yesterday afternoon. It was a beautiful afternoon, and we saw some nice fall color – but very little wildlife. It wasn't until we got most of the way to Mantua that we saw even any birds! But ... what we saw there more than made up for the dearth until then. First we saw a dozen or so kestrels (not my photo at right), which we had previously misidentified as peregrine falcons. They are beautiful birds, and we had some extended viewing of them as close as 75' or so to us.
Then we saw a dozen or so of a bird that was completely new to us: pinyon jays (photo at left is not mine). Sibley describes their color (the males) as “dusty blue”, with the females a similar gray. The photographer in me would call the females a desaturated version of the males :) It was a treat to see them!
Then we saw a dozen or so of a bird that was completely new to us: pinyon jays (photo at left is not mine). Sibley describes their color (the males) as “dusty blue”, with the females a similar gray. The photographer in me would call the females a desaturated version of the males :) It was a treat to see them!
Wind turbine owners paid Texas to take their electricity...
Wind turbine owners paid Texas to take their electricity... What? How could this possibly make any sense?
If you have my sense for the consequences of government involvement, you probably instantly suspected subsidies. And you would be right!
If you have my sense for the consequences of government involvement, you probably instantly suspected subsidies. And you would be right!
Two billion lines of code...
Two billion lines of code... That's the estimated amount of code that powers all of Google's web sites.
“Lines of code” (LOC) or thousands of lines of code (KLOC) are, these days, standard measures of the size of a software project. As any programmer would tell you, these are very imperfect measurements – somewhat akin to measuring a farm by its acreage. Some farms pack the plants more densely, have better soil, more sunlight, more water, etc. – so the number of acres is a very crude measurement of a farm's productivity. Similarly with software, LOC is a crude measure of how much a particular software program does, or how complex it is. But it's better than nothing, and nobody has yet come up with something that's widely considered to be better (though I'm partial to using the compressed size of code, rather than LOC).
No matter how you measure it, though, Google's code base is simply enormous. It is almost certainly the most complex single thing that mankind has ever produced – though that statement may not be at all obvious to non-programmers. By comparison, the design of an automobile is an utterly trivial effort – even when you include the design of the automobile's components. In fact, even in automobiles today, the amount of engineering that goes into software (including firmware) is larger than all the other engineering put together.
It fascinates me how the engineering effort put into software is invisible to a large segment of the world's population. They simply have no idea what went into, say, the design of the smartphone they use every day. As the number of people exposed to at least basic software concepts increases (and it is, most certainly!), that proportion will change. Has any other technology ever had this sort of disconnect between actual effort and perceived effort? I don't know of any...
“Lines of code” (LOC) or thousands of lines of code (KLOC) are, these days, standard measures of the size of a software project. As any programmer would tell you, these are very imperfect measurements – somewhat akin to measuring a farm by its acreage. Some farms pack the plants more densely, have better soil, more sunlight, more water, etc. – so the number of acres is a very crude measurement of a farm's productivity. Similarly with software, LOC is a crude measure of how much a particular software program does, or how complex it is. But it's better than nothing, and nobody has yet come up with something that's widely considered to be better (though I'm partial to using the compressed size of code, rather than LOC).
No matter how you measure it, though, Google's code base is simply enormous. It is almost certainly the most complex single thing that mankind has ever produced – though that statement may not be at all obvious to non-programmers. By comparison, the design of an automobile is an utterly trivial effort – even when you include the design of the automobile's components. In fact, even in automobiles today, the amount of engineering that goes into software (including firmware) is larger than all the other engineering put together.
It fascinates me how the engineering effort put into software is invisible to a large segment of the world's population. They simply have no idea what went into, say, the design of the smartphone they use every day. As the number of people exposed to at least basic software concepts increases (and it is, most certainly!), that proportion will change. Has any other technology ever had this sort of disconnect between actual effort and perceived effort? I don't know of any...
Laugh of the morning...
Laugh of the morning...
Definition of a Finnish extrovert: during a chat, he stares at your shoes.