Slashdot is powered by your submissions, so send in your scoop


Forgot your password?
DEAL: For $25 - Add A Second Phone Number To Your Smartphone for life! Use promo code SLASHDOT25. Also, Slashdot's Facebook page has a chat bot now. Message it for stories and more. Check out the new SourceForge HTML5 Internet speed test! ×

Comment Re:Beta testing self-driving vehicles... (Score 1) 47

Well, eventually they will figure it out how to make self driving cars safer than more than 99% of human drivers. When that happens, I'm not sure, but it will happen. Now, if you introduce them too early, a very risky and unsafe version of self driving cars that is maybe safer than 20% of the human driver population, but less safe than 80%, then anybody of those 80% using a self driving car would mean a safety risk.

Except that's not really how it happens, you don't need to be a race car driver to be a good street driver. A good street driver is merely consistent, appropriate speed, paying attention, obeying the traffic rules. It's not a skill level, it's a fail rate. You do things right for a year or five years or twenty years and then for some reason you fuck up. As in failed to yield, ran a red light, didn't see the pedestrian, fell asleep at the wheel, didn't check their blind spot, lost control of the car fail. I can guarantee you that all the SDC test vehicles are better than 100% of humans at not rear-ending anyone.

If it's not coming officially it's coming unofficially with all sorts of assistants where technically you drive yourself. And people will ignore it, but we'll dismiss them as Darwin awards.

Comment Re:No good if people don't have the cash or jobs. (Score 1) 280

That's the "rate" part of the equation: time is a factor.

Imagine if the FAA and the DOT hold back drone and self-driving vehicle deployment by not providing appropriate regulation for 10 years. The tech matures, heavily. Everyone is ready to go on it. It would provide immense cost savings at little risk. Then: they set requirements, and open the flood gates. Tens of millions of jobs vanish in six months; unemployment jumps by 15%.

Does that sound like a good economic situation to you?

Now imagine FAA and DOT get their asses moving now, start permitting some early deployments on contingency of TLA oversight and detailed reporting, and work on defining regulations to enable this stuff. The technology is young, risky, potentially-profitable, but potentially-disastrous. Tens of thousands of jobs go away in a year, becoming millions over the next three years, and tens of millions over the next decade.

That's actually a better situation.

It only takes weeks in the best case for prices to respond: the delivery fee and driver tips for pizza vanish, and that $16 order becomes an $11 order. That's $5 that can be spent elsewhere for each pizza; and it's an extra pizza ordered wherever someone was willing to pay for a pizza but not willing to drive or pay for delivery. Between these, you're going to need more pizza makers, more retailers, and more shipping for whatever other stuff you're buying with that $5 (although the pizza makers will shift in part from whatever that $11 was previously spent on instead).

Over the years, taxis give way to something Uber-like, because the regulations for a driverless taxi don't include background checks on the driver. Shipping costs exclude the driver's salary, instead only involving the electricity or fuel cost and the vehicle maintenance. More stuff is bought, so there's more shipping, meaning more shipping vehicles built, more mechanics, more electricity or diesel, and the like. That also means more retail, so more cashiers, inventory specialists, merchandising, and loss prevention, as well as infrastructure support for the retail centers. A lot of low-end and high-end jobs.

In the end, about 3.8 shipping and taxi jobs vanish, plus millions of delivery jobs. More retail jobs appear; some other service jobs appear; business management jobs for logistics to control all this shit opens up; if we buy new IT services (e.g. Spotify, Netflix, high-speed Internet), the support staff for those get fueled by those displaced jobs. A span of low-skill and high-skill jobs proliferate.

That doesn't take long in small bites, or in growing markets. Once you've started the economy shifting, it can move faster and pour workers from one class of jobs to another smoothly. If displacement accelerates over years, replacement will accelerate, too, and unemployment takes a small bump upwards--and comes right back down soon after the change-over. If that displacement happens all-at-once up-front, though, you get a massive loss of jobs.

You're always going to have transitional unemployment. That's what welfare is for. You can't make things cheaper and increase wealth at all income levels without bumping people out of jobs, because you have to pay everyone's wages, and you can't lower the price below the wage cost. Cut half the wage-hours out of a product and it's suddenly half as expensive, and off go half the people working to supply it. We either bundle more (e.g. cars, internet, cell phone service--always coming with more features, more speed, latest tech, pour on the new stuff and keep the price high) or we put people in the unemployment line until somebody finds out they can't sell us all the other shit we're now buying unless they hire more workers.

The real trick is to get it to span the risk gradient, and to span wide enough to not displace workers too much faster than you replace their jobs.

Comment Re:Nothing could go wrong here (Score 5, Interesting) 163

There are a lot of problems with "fact-based news", the biggest one being identification of actual "facts."

Look at ProPublica as an example. Their MO is generally to take facts and build a giant lie without ever actually lying, technically. I've given them thorough dressings-down for their blatant attacks on the American Red Cross and Amazon, but nobody actually cares because ProPublica has a better hook: take something people trust and convince them that trust has been violated. There are a few good examples here, though.

The familiar American Red Cross attack article on their handling of Haiti claimed ARC lies about the amount of overhead because they hire independent contractors. The reasoning is that ARC keeps 9% of their revenue stream as operating expenses, but their real overhead is around 40% or higher because they hire contractors who also have operating expense--never mind that the contractors are more-efficient than any non-professional, non-expert option, or that the materials have "overhead" because they need to be mined, shipped, and sold. Things aren't magicked into existence, and ARC isn't a vertically-integrated organization with expertise in everything; they generally try to bring the most-efficient solution to a problem, and that means hiring the best contractors they can find, that being the ones who perform at the highest return per cost invested.

ProPublica has repeatedly published ARC internal documents and loudly shouted that ARC is hiding and ignoring serious defects in their organization's handling of major disasters. This one's even simpler: the documents they published were Lesson's Learned documentation. They discussed what problems they had, why they had problems, and any potential methods for avoiding those problems in future disaster scenarios. Many are marked for further review and discussion. The documents ProPublica published are explicitly for the purpose of identifying problems encountered and preventing them in the future, yet they managed to claim ARC is "hiding and ignoring" all of these problems.

Their article on Amazon's "Buy Box" claims they always put Amazon first, even if they're more-expensive. What actually happens is Amazon (almost) always displays the lowest price-plus-shipping option for a particular product by default; and Amazon uses the lowest-price shipping option for that, which is Amazon's Subscribe and Save shipping. You can get free shipping by having $25 of items in your box or having Prime; ProPublica unilaterally applied a non-free shipping option to inflate the total cost. They also nitpicked about Amazon always listing Shipped by Amazon options first in the full list of sellers, even when these aren't the lowest price options; if Amazon didn't do that, they could have instead attacked them for advertising "free shipping" but making it "difficult to find the Amazon-shipped items to actually get it".

Notice the facts. Facts, facts, facts. ARC spent $500 billion, built 6 houses, was going to build 50 but gave up (never mind that the project was determined wasteful and pointless, and people were dying of a cholera epidemic that ARC stopped instead). Amazon shows you their option first and doesn't count shipping in their prices (never mind that free shipping is an option but alternate sellers don't offer it). ProPublica gives facts and tells you what to think about them.

It gets worse.

Jimbo Wales thinks he can fix this sort of un-news. Does he think he can identify and gate out finicky reasoning and spin? Can he identify when facts are missing, or induce others to do so? For that matter, can we identify who has the most-correct and most-complete set of facts, and if they're disclosing them all without ordering them to create an alternate narrative?

It takes some inherent bias to break fake news. I tear down fake articles I understand, and I hit economics pretty hard because I like economics. Fake news isn't just about the facts; it's about telling people what to think about the facts. Many journalists (unlike the folks at ProPublica) have actual journalistic ethic, at least enough that they report what they honestly believe. You get people reporting about David Dao's history because it's true, even though appropriate journalistic ethic would throw that story in the bin because it's victim-shaming and irrelevant to the news at hand; and they at least believe the facts they report are true. You get people reporting on politics with different interpretations of the implications, and they believe those things are true.

You're going to go up against these people because you believe something else about the facts.

Are you ready for that?

Comment Re:Robots are good (Score 1) 280

You have to understand that most (then quickly virtually all) menial jobs will be automated away

A large number of jobs babysitting machines will be created. You know, the same thing that happened all through history: we got rid of highly-skilled, heavily-trained, expensive craftsmen and replaced them with assholes who can operate a lever after five minutes of instruction. Then, we reminded those people we can replace them easily, and paid them less.

I'm mostly ok with your proposal but minimum wage is insufficient to live on so that part is still a stick which is not what UBI should be. Universal social security should be at a minimum, a livable wage.

The 2013 number wwas $546/month per single-adult.

That used a 224sqft single-individual living space, comparing to low-income apartment rents at an average of $1.00-$1.06 per square foot rent, measured in Baltimore, New York, Seattle, and various areas of California, as well as spot checks across the country. The cost of constructing such an apartment was assessed against published lists of materials and replacement intervals, and compares favorably because the kitchen and bathroom fixed costs are roughly $3,000 out of a $26,000 construction cost. Because individuals with a basic income have a known income which can't be lost by termination of employment, loss of working hours, or loss of welfare benefit, the cost of risk in scaling down living spaces shrinks.

With that consideration, the viable monthly rent for such a living space is $237; I budgeted $300/month. That left $246.

Food, using retail prices checked across high and low income areas, was originally specified as $100/month. I've modeled complete food plans in 2016 as low as $25 per 2,000kcal/day over 30 days, but that's rigorous and fragile; the additional buffer is required to control risk.

I also allocated $35/month to clothing and $35/month to personal care. These expenses are more-flexible--clothing obviously can be held onto longer; and soap, tooth paste, laundry care, and the like are overbudgeted--and so I eventually modeled onto a combined $170/month food, clothing, and personal care budget. That gave me enough flexibility for a $45/year Sam's Club membership, utensils, and kitchen tools in one model, even getting so far as purchasing a $200 bread making machine in the fourth month on savings.

Utilities come to $35/month in this model. I used to live in a 750sqft apartment and pay $57/month for utilities; it was poorly insulated, with brick, 2x4 air gap (no batting), and 3/8 drywall on three sides (two long, one short). That includes a $13/month gas customer charge and a $7/month electric customer charge; the landlord can split a single account across multiple tenants to reduce these charges via metering-on-site, although we could theoretically apply regulation to specify a building charge for multi-tenant residences to try to reduce them administratively. Some buildings have shared utilities, but I don't like one tenant's overage to cost other tenants; if the landlord is using on-site metering, they have the same responsibility, except they account for utilities by actual use instead of by equal responsibility.

This leaves $46.49 unbudgeted, with each budgeted expense overbudgeted as a risk control. All in all, roughly 45% of the $549 is a risk control; it's technically possible for a single individual to survive on around $300/month, barely, under perfect conditions. That's unacceptable risk.

The 2014 number is $552/month; the 2015 number was $583/month; the 2016 number is $602/month. That's $331 rent; $188 food, clothing, and personal care flexible budget; and $33 utilities. This leaves $50 unbudgeted. Note that the income actually grows faster than inflation, so these budget numbers are higher than the difference in cost--food costs increased by about 2/3 as much as the food budget between 2013 and 2015, for example, meaning the 2015 budget was $182 but the actual cost was $177, versus $170 in 2013 (food prices increased by ~4.2% from 2013 to 2015; income-per-capita increased by ~6.8%).

So yes, it's enough for one person to live on.

There are also considerations in there for aid for children of low-income households, since they don't receive a UBI, being that we'd have to provide a UBI sufficient across the entire variation--that is, almost every household would have to end up with money left over after childcare expenses, thus pumping out and neglecting children is profitable, or else the childcare welfare isn't actually enough to care for children. Because the risks of a public aid welfare for childcare are known and actually quite low, scaling that model down to specialize in childcare welfare works well.

I've actually put numbers to this shit, not just feel-good ideas about what is and isn't enough to live on. Do you want something that actually works, today, and in the future; or do you want idealistic bullshit that we can't afford, that won't work, and that will cause economic collapse if attempted? People die for unrealistic ideals--real people, with lives that could have been supported. We actually can end homelessness and hunger in the United States today. Bullshit fantasies about a world where nobody can remember how basic physics works because the machines have become our lords and masters and we have become the cattle have no place in civilized discussions.

Comment Re:We already had this sales pitch... (Score 1) 137

There are a few things wrong with your analysis. The first is that disk writes tend to be bursty for desktop users. You write a few hundred MBs (or a few GBs) and then drop down to an average of a few tens or hundreds of KBs per second. Spinning rust can easily keep up with the average write throughput of a typical user, it's the bursts that it has problems with. If you can buffer a few hundred MBs of writes, reorder them to reduce head movement, and then write them out behind the user, then you'll get much better performance. Obviously, this won't help for server workloads where you're I/O limited all of the time, but it will help a lot with desktop / laptop use.

The second is that one of the big bottlenecks for modern filesystems is the wait until data is safely in persistent storage. System RAM doesn't help here, because it goes away with power failure. To ensure consistency, you have to pause writing parts of an update until you've received confirmation that the previous part is written. In a conventional journaled FS, for example, you don't start writing the updates until you've confirmed that the journal has been committed to disk. With NV cache, you can get this confirmation practically instantly. If there's a power failure, then the drive just has to replay the transactions from NVRAM.

Comment Re:Like what? (Score 1) 280

I was trying to extrapolate from sword-making to something more-recognizable. Maybe these aren't as related as I'd considered. The process for making e.g. a katana involves a hell of a lot of folding, repeated quenching, clay coating, and so forth; and all the folding is done largely to work impurities in the metal to the surface, producing an anti-corrosive coating and making the metal more uniform and less prone to fracture, which is probably not important when you have access to modern refined feed stock.

People still charge $800 for a traditionally-constructed, hand-made chef's knife, and there's a waiting list.

Comment Re:Synonyms being used (Score 1) 109

That sentence doesn't say it sold receipts; it says it sold "anonymized data". They could be collecting Lyft receipts, reading the locales, aggregating statistics (which produces data that doesn't have any idenitfying information), and selling Uber statistics on age demographics, lengths of trips, where trips started and ended, times of day, etc.

If you get a block of data that tells you that 20-25-year-old males in Boston are traveling from one block area to another block area, average trip lengths of 4 miles, peaking around 7:30am and 3:30pm, with histograms per 15 minutes, what does that tell you about the users whose data went into it? Does it let you identify them as individuals? Can you somehow factor out someone's name from that?

Comment Re:Like what? (Score 1) 280

Someone's designing all this new stuff, someone's operating all of these ultra-low-labor factories. We'll need 10 people instead of 300 to run the factory, but what then? We've suddenly got the income to buy 30 times as much stuff--and we need 30x the production, meaning 300 people running 30 factories instead of 1.

Don't be daft.

Comment Re:Robots are good (Score 1) 280

Robots won't "take all our jobs" in the same way that machines didn't "take all our jobs" already. The labor required to produce most goods and services today is extremely-fractional. Like, it used to take 200 times as many workers to produce iron; and making a shirt in 1800 was 479 labor-hours (at minimum wage, that's $3,950) versus a total of under 3 hours today (including the fiber, spinning, weaving, dying, and construction). Food production requires less than 6% as many workers for the same yield.

We'll end up with 200 times as much shit.

Comment Re:Ting (Score 1) 195

My next Mintsim renewal for 2GB/month is $160 for the year. The standard price for 2GB/month LTE, unlimited voice, unlimited text is $199/year + 3% taxes ($205). For 5GB it's $299/year +3%; 10GB is $399/year +3%.

What's Ting? $6/month for the line, plus $3 for any voice at all, plus $3 for text, plus another $5 for data, plus $7 in taxes, plus any usage? That's like $24/month minimum; if you're eating more than 100 messages and more than 500MB data, you're already around $38/month. I'm paying $17/month.

Sounds like you're getting soaked.

Slashdot Top Deals

When it is incorrect, it is, at least *authoritatively* incorrect. -- Hitchiker's Guide To The Galaxy