Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?

Comment This is completely fair in principle (Score 1) 632

This policy comes from the fact that Social Security recipients die but often their next of kin do not report the event to the SSA, and a month or more of ineligible benefits wind up in the hands of some survivor in the family. This is not quite theft, but the fact is: Some family member fraudulently cashed a DEAD person's check that was legally void once the beneficiary was DEAD. This can only work due to the family's failure to notify the SSA promptly that the beneficiary was DEAD, which is the duty of the family. Now maybe everyone is too bereaved to be telling the government that grandma has passed, but in that case you ought not to be cashing a DEAD guy's checks. It is not reasonable to expect the government to make a federal investigation of who in the family wrongly took this money, whether it was deliberately or somewhat innocently taken. They take it back from the next of kin, and if the fault was with somebody else, well, you sort that out within the family. Sure that is messy, but it is a family mess and not a government mess. That the government takes it back from a refund, and doesn't affirmatively prosecute you, seems quite kind and gentle.

Comment Charging stations will have to absurdly multiply (Score 1) 357

Why no concern that if it takes 10x the time to charge an electric car as to fill a gasoline tank, then you're going to require 10x the number of charging stations as we have gasoline stations now, to the extent we replace gasoline with electric. Most places have more gasoline stations around than they want already. Where are we going to find the real estate in suburbia to put 10x that number of charging stations as we have filling stations today? And 10x is a kind estimate for electrics.

Comment Re:That's why (Score 1) 961

> And all you have to do is close the door and turn the valve on.

Idiot. A large cylinder holds perhaps 80 cubic feet at 1 bar. Calculate the volume of a small room and you'll see that dumping 80 cu ft of inert gas into a room would hardly affect oxygen concentration.

Consider why that large cylinder of CO2 for the soda fountain in every restaurant and convenience store is not treated like an asphyxiation hazard if it were to leak. And CO2 is not inert like N2.

Comment I blame Java for the healthcare.gov mess (Score 1) 577

The failure that is currently healthcare.gov is heavily based on a Java implementation for everything including the most trivial and simple things. I've spent two weeks getting nothing but inglorious errors, and the [mis]use of Java appears to be deeply involved.

It is an astonishing, breathtaking failure when viewed with any expertise in how things should be done. You log in and the screen just turns blank with no error message. Or you get an error message that literally just says "Error!" in red and nothing more. Or it gives an error that indicates what can't be the true cause. Or it says we're too busy ... at 3 a.m. Or try again later (but not how much later). Or a bunch of Java code gets splattered onto the page in literal text. Different errrors in every variety of browser. No evidence of version control or other error tracking. No indication of status, good or bad, no list of active problems, no advice when to expect resolution. There is no documentation or explanation of any of this from the authorities or the contractor. The authorities will not report usage statistics, how many succeeded, or how anyone succeeded if there is a specimen anywhere in the universe to copy. One has to question whether it has worked in a single instance.

Here is an acid-core example: every single user has to confirm via email. Yet the email is flat-out RFC-violating non-compliant, and can't be read in email readers that don't know how to handle this non-conformity. Specifically, this violation appears in the email headers:

Received: from . . . service.govdelivery.com
Content-Type: multipart/alternative;
. . .
Content-Transfer-Encoding: base64

This is not an example of bad Java, but it does show the kind of foolishness passing for system-building everywhere you look. This garbage came from govdelivery.com who are apparently the choke point for the entire system. If they fail, or if you fail to deal with them, you are SUNK.

They do have a pretty girl smiling at you from the home page. Puh-leeze.

These are not bugs or glitches or the overwhelm of success. This thing is utterly defective. A FAILURE. One must question whether it will ever work, and if it won't have to be abandoned for a do-over. Nobody expected a smooth rollout, but this is head-slapping incompetence.

And the law is, you must succeed with it, or ELSE! You cannot mail in forms, or call on the phone, to get this done. It all happens on the Web.

And when has the federal government ever appropriated non-government technology and property in this way, and used it as the sole means to enforce something against the citizenry? With the income tax, they at least give you the paper and a post office to send it back and forth. The government will depend completely on the Internet now to keep you from being fined or put in jail?

Comment Optical engineering explanation for purple fringe (Score 1) 472

The purple fringe problem is an old one when it comes to digital cameras. Most famously this spoiled the production models of the Sony DSC-F828 in 2004, which was supposed to be the ultimate high-resolution digital camera in its day with perfectly optimized Zeiss optics. The irony was that the very quality that made the lens superb in visible light made it that much more aberrant in infrared. Infrared blocking filters are not perfect, so out-of-focus infrared images appear whenever there is extreme contrast in the scene at extreme off-axis angles.

Purple fringes like this are not due to lens coatings or sapphire windows. Nor are they due to lens flare, flare being due to internal surface reflections, so it is wrong to call it a purple flare. Strictly speaking, it is a chromatic aberration, compounded with some coma effects.

The cause is simply infrared (IR) light being imaged by the image sensor. The lens is highly corrected to sharply focus visible light, but such corrections result in severe aberrations in focus for for any light outside the visible. These aberrations worsen with wider angles, that is, the farther out toward the edge.

Of course there is an IR blocking filter in the lens, but it is not perfect. A very small proportion of the IR does get through, but not enough to normally be imaged. However, when you have an severely bright highlight in the scene that is overexposed on the edge of the frame, the light itself will be "blown out" (pixels all white), but abberant unfocused IR rays will form a fringe. This fringe is purple because that is the false color that IR light yields in an RGB sensor. This fringe is not blocked by the IR filter because the highlight is far more intense (potentially by huge factors) than the exposure for the rest of the scene, so even 99.99 percent IR blocking filter lets through enough rays that when aberrated show up as a bright fringe.

Example from a Sony DSC-F828. Note the camera flash reflections from the shiny trophy at the edge of the frame have purple fringes, while the reflection off the glass near the center of the frame does not.

This problem only appears when you have a highly corrected lens, a high-resolution sensor, a high-speed-wide-angle lens, less-than-perfect IR filtering, and a scene of high spatial contrast at the edges. That's why it doesn't appear in most cameras, because few cameras are so high-performance in all of those areas at once.

Fixing the problem can be done by reducing the performance in one or more of those areas. Or you can design even better optics, but that is difficult to implement in a compact size like a phone requires, because it takes bigger bits of glass and more of them. You can also correct in firmware or software.

Comment Re:new owner may need to honour preexisting contra (Score 1) 443

No, not at all. Buyouts are structured typically to buy *assets* only (including the trade name), not obligations (debts, contractual obligations, etc). Warranties go extinct in such cases. No knowledgeable businessman would buy a whole business in the sense of buying into potentially unlimited liabilities. When you buy anything for a "lifetime" of support or warranty, you are only in effect buying a promissory note from a business that may or may not even exist when you come calling to collect.

Comment As a Florida homeschooling dad myself ... (Score 1) 294

I have to chuckle at reports like this, because I'm often told my homeschooled kids are not in the "real world" and aren't being properly "socialized".

Why would any child or parent object to fingerprinting?

Coercion and regimentation in government schools is just the *real world*, to wit:

Riding buses of a kind that nobody else rides except in prison or at boot camp, painted a special DOT color that other vehicles are not allowed to use, that travel under special traffic rules that nobody else is allowed to use.

Eating food provided by the government, served in a facility with famously odd personnel, none of which would be patronized if offered to people as a free choice.

Forced to endure unwanted company that inflicts physical assaults and social harassment that would be criminal or tortious if acted by adults.

Required to submit to government employees who are 100 percent unionized and paid twice what they're worth in the free market, who work 3/4 of each day for 3/4 of the year. (For those of you who learned public school math, that's 3/4 x 3/4 = 9/16, or about a half time job for double a full time annual wage.)


The real world?

Slashdot Top Deals

Computers are unreliable, but humans are even more unreliable. Any system which depends on human reliability is unreliable. -- Gilb