Make sure the machine remains visible at all times. Put plastic tape over any access hatch so you know if it's been tampered with.
Lexan case, locks, sensors, alarms--visual and audio.
If the paper is then counted by machine, then fair enough
Analog failure: machine misreads ballot. Direct-recording machines can display the ballot as it is understood.
Paper is easy to compromise: you can manufacture, alter, and lose it. The same is true of all data, really. You only have integrity inside the polling center.
Your main weakness is recounting. Let's work backwards from electronic for this one.
With electronic voting machines, once you've proven the integrity of the machine, you can prove the ballot set. Everyone votes, the machine displays some kind of computation, and only then do you enter the machine to remove the ballots (digital media copy, whatever).
For provisional ballots cast in-person (wrong polling center, etc.), you load the provisional ballots into your hash set by the same rules, marking them as provisional--and you don't include identifying information in that set. This still exposes identity to the public if your polling center collects only a few provisional ballots (if it's only one, we know whose is provisional), which is the same problem as very few people voting (one polling center here had 67 people vote in the primary, with only three Republicans who all voted for the same candidates, so I can give any of them their exact ballot back. Imagine three people voting in the General by provisional in a pile of 1,600 votes). The solution is to obscure who is voting provisionally.
Under this scheme, ballots are traceable: the central authority publishes each ballot set with its polling center. We can all recompute the hash and prove the ballot set is untampered. They can strike any provisional ballot, so long as they leave its contents up in the ballot set and mark it rejected.
As you note, absentee ballots come in the mail, are paper, are not observed, and so represent a weakness. Disenfranchisement represents a larger threat, so we accept this weakness. On military base and ship, we can actually run the same kind of public election as here, observed by the public of the men on ship--who of course will leak if they're not allowed to record election ballot traces.
So what about paper?
Well, you can slip a piece of pencil under your fingernail and spoil ballots when counting paper in full public view. You can miscount. Election judges can collude, despite the whole reliance on people not colluding (we've seen this in other places).
We perform our recounts days or weeks later, with ballots kept in a secure location. Again: how do I know anyone in the chain-of-custody isn't just a group of 3-5 people who can print a new seal and are fully-willing to collude? Buy a few people off, get them an election pen (remember what I said about pencil?), and they can create a few new spoiler ballots.
In an election with 100,000 ballots, one state here recounted with 3,500 fewer, and claimed they must have just found more spoilers (without explaining which were probably not originally spoilers). We just accept that. Three percent of votes.
See how many avenues of attack we have? Now look at our electronic voting machines. Yeah, hell no. The stuff out there is horrible. That doesn't mean paper is safe; it just means we damned well better move to stronger EVMs if we're going to move off paper.
Candidate wins by 2000 votes and only 1000 advanced votes, why bother counting
Computing whether a candidate won can be...difficult. Take any Smith-efficient system. You have to compute all the one-on-one pairings: if a ballot ranks Bill 1, Anne 2, and Jim 3, then Bill vs Jim is 1 vote to Bill, Bill vs. Anne is 1 vote to Bill, Anne vs. Jim is 1 vote to Anne. These pairings are important.
To get the Smith set, you find out who wins these.
Let's say Bill, Anne, and Jim each get a majority of rankings above Dave and Mary. Dave and Mary categorically lose against these other three. Dave and Mary are out.
Now we have Bill, Anne, and Jim. Bill beats Anne; Anne beats Jim; Jim beats Bill. O
Diebold made ATMs. They're money-changers. They see a need for a system that collects and transmits votes.
When you build a microwave oven, you don't load it up with fork-detection hardware. You tell people to never microwave a fork. In an election, people are intentionally trying to microwave a fork, and you need defenses against that. Diebold, ES&S, Hart, these people built a microwave to accomplish microwaving food, not to accomplish microwaving food in hostile territory where even the people who bought the microwave are trying to sneak forks into the microwave.
I'm getting into the business. It starts with security. Gonna be a while before I'm ready to play, but you know what? Everybody hates all of these people. They have no idea what they're doing. I'm going up against giants, but you gotta realize: I'm not competing with them; they're competing with me.
So here's the thing: computers can be made impenetrable; physical locks can't. Computers, however, are immensely difficult to validate: we can ensure a small code body (maybe 4,000 lines) is correct, but we can't ensure the whole OS stack, the enormous application in its entirety, all libraries, and so forth are correct.
You know that thing where you can't hack into a computer that's unplugged and unpowered? You can totally hack into a lock by getting a large enough hammer. Any lock. You can manufacture a duplicate lock, a duplicate seal, a duplicate ballot box.
A computer that's powered on, only accessible via port 443 with TLS, and configured to reject a client not presenting an appropriate authentication certificate has a relatively-small code path to deal with. Unfortunately, at that level, you're facing the operating system's network stack, its firewalling stack, parts of the TLS library, and anything in the application that tries to identify the user by reading the data (which should just be using the TLS library to ask for the principle). That's a lot of code. If all of that is secure, holes everywhere else don't give hackers access unless those hackers are also valid, authorized users trying to exceed their access.
We would have to isolate the device from the network entirely.
And there you go.
Voting machines only need to collect and package votes. They need to exist for one day, and they only need to release their contents in a batch--or in a one-way stream such as by FEC light transmission. Folded Reed-Solomon at 5,000% correction, a flashing LED, and a photosensor on the thing across the room would let you safely transmit votes during voting, if you really want. Thing is you still need to batch everything up at poll close and provide proof of the ballots in public view to confirm the vote tally.
So we're down to a single approach: unplug the machine from the Internet entirely. You vote, it collects votes. At poll close, it computes some kind of reproducible hash of the ballot set, which it displays for all observing. We all write it down, take pictures, whatever. Then it transmits votes--possibly by someone opening the locked door, plugging in a USB device, and copying the ballot set off physically.
With no surface to attack, it's secure.
You still need to prove the integrity of the machine at poll open. Remember: software manufacturers, hardware manufacturers, and all elections staff and authorities are also threats.
First, with aren't they using smart cards with passwords on the keys?
Because voting machine manufacturers are money-changers, not security experts. They make ATMs, they make voting machines.
Second, why did the software permit weak choices?
Made to order.
Third, why are infosec officers not replacing those pages in the manual, training users in proper procedures, rejecting the products at user acceptance or running tools for weak password detection?
The same companies who provide the machines also provide voter outreach and elections consulting. They'd be the ones deciding if you should reject this shit.
Why do you think I'm starting a business and breaking into that industry? Paper ballots look kind of like the Internet to me (from thousands of polling places, through the hands of small groups of trustees, flowing to central State offices... looks just like all the online banking traffic), and the electronic voting industry is run by people with no background in IT security or risk management. Everything is run wide open and vulnerable to all kinds of attack. Somebody needs to fix it.
Let's put aside the issue of using a password based system in the first place.
No, let's not do that. It never even occurred to me to use passwords when I stared writing the SAFE VOTES guidelines (elections procedures and standards specifically for elections run via direct-recording electronic voting). I didn't even require 2FA--largely because using the credentials to do anything permanently takes the machine out of commission, causes alarms, and generally draws a whole lot of attention and stops your election, but also because the credential is created at poll open and is destroyed at poll close that day.
the official manual will say use a secure 10 character, upper, lower, special, & number password.
Assuming the accounts are persistent and long-lived, I would have said to use either TOTP, PKI (FIDO U2F, Smart Card), the two as a 2FA pair, any password plus TOTP or PKI as a 2FA pair, or a single-factor password of at least 14 characters and preferably generated as four random dictionary words (49 bits with 5,000 word dictionary and no variations, vs. 52 for 4-class 8-character).
TOTP has a huge problem: if you demonstrate the installation of the software image on the machine, such that you can prove an untampered image, then the public observes the TOTP credential. You can't keep it secret. This is also true of entering a password.
State level, you need a LOT of corruption across a highly distributed network of independent voluntary organizations to impact a vote
The votes move from hand to hand from thousands of locations through small groups of trusted parties. That's tens of thousands of weak links--millions in some states.
The paper based voting system in the US that has been used for decades is pretty good.
It is. It's a complex mess with enormous integrity problems, but it's only severely-abused in a few places--and that severe abuse is relatively minor. A strong, decisive victory or even just huge voter turn-out is usually enough to overcome the level of tampering even in states where it's rampant.
We estimate about 30,000 ballots just tossed out in Florida every year. Ohio did a recount once and came up 3,500+ ballots short, with the excuse that "it's probably spoiled ballots" but no trace to show which ballots were spoiled (yeah, "We're not sure, we guess it's X, it's probably fine, trust us" fuck you).
end voter manipulation via social media that has been their only real threats.
You should see paper ballot security considerations. Many of these are built on the idea that you can trust some party by putting other parties around them, under the theory that they're adversarial and will act as checks on each other; unfortunately, we have Democrats endorsing Republican candidates now, and besides that you can just join a party and carry out espionage (really, vote tampering is an enormous offense and an attack on our national security).
You can fix that with voting rules that reflect the voter's preferences better. The combination of Single Transferable Vote for multi-winner elections, the Big Party Rule (a party with 25%+ registered voters can nominate two candidates for single-winner elections, but only by primary election using STV), and a strategy-resistant Condorcet method like Tideman's Alternative Smith would provide pretty good protection. Electoral Fusion helps, too.
Big Party Rule upends the base voter issue: a ton of people will vote R or D, and now they have two R and two D candidates. You get Trump vs. Rubio, Bernie vs. Hillary. Combine this with any Condorcet system and you now need a mutual majority favor.
Under the Big Party Rule, you would probably get Bernie from the left and Hillary from the moderate Democrats; interfere with a huge pro-Bernie propaganda campaign and you get excitement for Bernie, but who is the second choice? It's... still Hillary. The same is true of Trump and a more-moderate candidate like Rubio.
Hit the General Election and now what?
Well the entire moderate independent vote is going Hillary-Rubio or Rubio-Hillary. The moderate R goes Rubio-Trump or, for the Never Trump crowd who voted Hillary, Rubio-Hillary. The Pro-Trump crowd will go Trump-Rubio-Johnson-Hillary because they want to shoot down Bernie Sanders by putting Hillary above him (they might or might not rank Stein above or below Bernie at the tail end). The Bernie crowd is going Bernie-Stein-Hillary-Rubio, Bernie-Hillary-Stein-Rubio, or so forth.
In this scenario, a huge amount of the population is trying to NOT elect Trump or NOT elect Bernie. That makes those two unelectable. The voters on the left have said they'll concede to the Republicans if they take Rubio (that's what actually would have happened in this line-up, based on the data I've been able to dig up); the voters on the far right have said they'll settle for Rubio if not Trump; moderates are taking Rub
Think about the Internet. There are thousands of points where data leaves, moves through intermediate routers, changing hands, until it reaches a location. Your bank, for example.
Why do you use encryption when sending your credit card details? It moves from your PC to your ISP, through several core routers. Random people can't sniff it. Your traffic isn't exposed to untrusted third parties; arguably, all of the systems through which your data flows are more-trusted than the merchant to whom you're sending your credit card number!
"Computers can be hacked", right? What if the "hacker" is a technician at an ISP? Remember that guy who did that with the lottery?
Paper ballots begin at polling centers, where we trust staff to not manufacture additional ballots in the back and swap out ballot boxes with duplicate seals. They transfer through a chain of custody, always in the hands of 3-5 people who we assume aren't colluding, or stored in a secure location we assume isn't intruded upon by someone with a key and admin access to the security logs. They change hands several times.
That doesn't even get into the error rate of counting--with the implication that you can manipulate the error rate. People freak out about that when it comes up: Kris Kobach in Kansas was giving instructions about what ballots to not count (spoiler ballots, in late, etc.), and he was also a candidate in the GOP primary for Governor. You get a lot of individual judgment with hand-marked paper ballots especially--is that a stray mark, or just a little outside the circle? Is it enough to flag to an observer that the purchased vote has been cast faithfully? Does the signature sufficiently match the one on file? All up to personal taste.
Paper ballots are a complex, flimsy, and insecure network.
We could put a white OLED over the camera lens to act as a backlight for LCD; or put the camera lens behind the OLED screen itself. Either way, turn off the screen in that area and you've got a transparent cover over your lens, same as today.
It wasn't sound in 1960...the rest of your argument hangs on that fallacy.
No it doesn't. It relies on something called "technology", whereby the amount produced per working-hour increases.
You think SS payments haven't also gone up?
The COLA-adjusted Social Security payments in 2016 are 7.83x as many dollars as the benefit in 1960.
The number of actual dollars of income per American over the age of 65 in 2016 is 11.77x as many dollars as the benefit in 1960.
In other words: for every $7.83 that Social Security needs to pay out in 2016, there is $11.77 available at the same FICA tax rate--IF applied to the same portion of all income. Put another way, we have $1.50 available taxable dollars for every $1 of taxable dollars we need to cover COLA-adjusted Social Security benefits starting in 1960, assuming we're taxing the same portion of all income,
It turns out that, yes, there's more available every year. What we do: in 1980, take 8% of 80% of the money. In 1990: take 10% of 60% of the money. In 2000: take 12% of 40% of the money. So in 1980 we had 6.4% of the money, in 1990 we had 6% of the money, in 2000 we have 4.8% of the money, etc.
The PCB is actually pretty cheap, likely around 10 cents to a quarter. The MicroUSB port is a 26-cent part; the MicroHDMI connector is around 93 cents (yeah). Looks like you can get a MicroSD slot for around 96 cents in 1,000 quantity, around 74 cents in 2,000 quantity. $2.19. Not bad...prices have improved. If pressed, you might be able to get the MicroHDMI down around 40 cents; the MicroSD has springs and other mechanical bits. Call it maybe $1.50?
The Pi has 62 components, including SOC, plus 5 connectors, at an average 7.4 cents each. Given the above, estimate 5.6 cents each. The SOC seems to be around $1.50 in volume, leaving 3.2 cents per small component.
A TV tuner is an immensely-complex device with multiple large circuits, and has to decode color channels separately. There are a lot of parts in there. That's... actually changed since I last bothered to look: not only is TV now digital, but people came up with CMOS tuner ICs--expensive to set up, but cheap to manufacture in quantity, eliminating much of the cost.
You don't seem to understand anything about estimating. Pathfinder gave NASA ground data to correlate with telescope data, enabling them to model the Martian atmosphere--that was the point. The biggest miss on Opportunity was predicting a continuous coating of dust on the solar panels, which didn't happen at all.
A month is enough to get a read on insolation at ground versus in space versus weather, by the way, so NASA has a pretty good estimate of year-long insolation data, and can work out charging rates from that to relatively-high accuracy because they have a lot of data about how atmosphere interacts with insolation. They also knew a lot more than you do about batteries, and estimated a shorter battery life and higher wear.
As to extrapolation versus interpolation, most estimating isn't exactly that. A lot of estimating is "A is like B, and X is like Y except smaller/bigger and made with B. We've been using a new technique to do C, which is completely-unlike B, but should apply to parts of the process of doing B, so B should be improved in budget and time by these factors. Doing X should thus come out in this manner. Here's the time, budget, and resources necessary." We've never done anything like this, but we know about how much it should cost and how long it should take--and what the variation is in our estimates (i.e. risk).
If you expect anybody to approve increased spending and cuts in the future, you are crazy.
This actually reduces tax burdens without cutting programs. It's revenue- and deficit-neutral. Yes, it sounds like black magic; it's an accounting trick. There are further reductions in the future because of downstream effects; I don't account for them first-pass.
Also fund the SS trust with good assets _first_.
The Trust has good assets. It holds Federal Treasuries and they perform their function quite well. The problem is its FICA source is broken. The amount of income available per retiree has increased year after year much faster than inflation, yet the proportion of income exposed to FICA taxation has decreased. In other words: 2x as much income is being made, and 40% of that is exposed to FICA taxation (=80% as much being taxed, thus falling revenue).
Increase taxes now for spending cuts in the future should NEVER be believed.
The actual proposal in 2016 was a 33.5% CIT (vs 35% at the time) and a blunt 3.2% reduction in the top personal tax rate. I don't have the data to tweak personal income brackets further or I'd have set it back up to 39.6%.
Under the 2016 rough model as such, a 2-adult household making $400,000 would have taken home an extra $1,674/year. At $50,000 of income, that household is taking home $8,070/year additional after taxes and Dividend. Meanwhile a household with $25,000,000 of income is taking home an additional $841,117 after taxes.
The proposal is actually to restructure programs now to pay out cash to a lot of people, offsetting tax burdens and acting as an effective tax cut in order to reduce costs, increase revenues at current tax rates, and lead to further tax cuts later while we roll out even bigger government services and pay off our debts.
So let's divide the Gross National Income for the United States by the number of Americans age 65 and older. Then, let's compare that to the same value in 1960 plus inflation since 1960.
Without a cap on the FICA, a 6% FICA rate in 1960 would have needed to come down to a 4% FICA rate in 2016 to avoid simply loading the Trust up with money it was never going to spend.
The treasuries are as good as cash--better, really, because they avoid the economic drain of simply taxing money out of circulation and totally avoid interest. The Treasury could tax $10 billion in 1995 and spend it; but the Social Security surplus of $10 billion is already here. If we spend that $10 billion in 1995 and then put it back in 2005, then we're theoretically only spending $5 billion of 1995 money.
Moreover, if you raise taxes to pay it back, you only need to raise them 40% as much as you would have to fund the $10 billion originally. It's not just inflation: the GNI/C is higher by more than inflation, and the number of people is also higher. The longer the US holds debt, the easier it is to recover that debt with fractionally-small taxes.
Think of it this way: Inflation at 2% over 15 years is 35% (yes, I used a rough 50% number up above). Productivity growth of 1% means the number of inflated dollars per person, 15 years later, grows by 56.25% (1.02% * 1.01% each year). Population growth of 3% per year means you have an increase of 55.8% population. Take the 156.25% GNI/C and multiply that by 155.8% C and you get 243.45% GNI increase.
That's 2.4x as much money out there to tax, and 1.35x as much to recover with inflation. You're way ahead. If you raise taxes now to pay for your debt, you only have to raise them about 56% as much as if you'd gone the revenue route initially, assuming your interest rates are in-line with inflation.
So your nation needs to not bury itself in debt; it can leverage debt pretty damned well, though, and should absolutely favor being indebted to itself over holding piles of cash.
A working UBI doesn't exist. A universal dividend provides a basis for social security and welfare, while keeping your economy functional. You need your social insurances and welfares because the standing income necessary to stabilize temporarily-displaced workers is huge, and the standing income necessary to eliminate all recessions and end the existence of poor inner cities as a phenomena is small.
There's no reason Social Security should have ever been insolvent. In practice, it should have been able to reduce the FICA rate over time while increasing benefits.
Most UBI proposals are for about $500/month, and even that requires dismantling Social Security, which would deprive tens of millions of people of their retirement income, generating a firestorm of political opposition.
I drew up a Universal Dividend proposal that pays $500/month to each adult while making social security permanently-solvent. I dismantled nothing, raised no taxes, and created no additional deficit.
There's a $900 billion increase in Federal outlay (I restructured $1.1 billion into $2 billion). When you remove the amount of the benefit offsetting its own funding from this, you find a $1,200 billion reduction. That's net $300 billion reduced taxes.
For example, if you pay $4,000 into the FICA and receive $6,000, that's $4,000 offset. If you're wealthy and you pay $100,000 into FICA and receive $6,000, that's only $6,000 offset.
Because that offset exceeds the additional outlay, we can adjust the general tax brackets to draw more income at various levels and create tax cuts across the board. The base model even includes a 1.5% corporate income tax cut (from 35% to 33.5%).
You provide welfare and social insurances on top of this. The Dividend keeps economies healthy (yeah, huge job creator, breaks recessions before they begin) while your aid programs keep families fed, clothed, and housed. I didn't dismantle the welfare system when I ran the numbers.
Think about everything you would buy if you had unlimited money. You'd buy food, clothing, rent, car insurance, games, fast cars, rocket ships.
If we cut your money back, you start to prioritize. You must eat. Rent is pretty important. You need your car to get to work. The tighter it goes, the more you refine your priorities.
One of your priorities is savings.
Extremely-rich people will buy financial securities such as stocks, bonds, and commodities on the secondary securities market. When we issue stocks and bonds, money goes from the buyer to the business (or government); when we trade them on the secondary securities market, money moves back and forth between traders. These people are essentially sucking money out of your 401(k).
Poor people can't afford all that, and the middle-class don't make much money trading stocks. Middle-class may have a modest emergency fund; the poor don't.
Those things don't produce economic activity. They're just idle cash. When you move some of the income down (please no asset taxes), it moves into the hands of people whose marginal opportunity is in spending, creating revenue and jobs.
The second factor is distribution.
Distribution is simple: a poverty-stricken area is full of people who can't work because people are too poor to buy, thus there isn't revenue to pay wages. Those people are unproductive (and they cost welfare).
Moving spending power into high-unemployment, low-income areas creates jobs. That means people work, produce, and draw taxable income. That income is linked to production, so your nation's overall wealth and wealth per person increases (these people are working and producing).
You know how salt bridges form in water softeners and you have to open them up and break them apart? It's sort of like that with capitalism. The system figures out how to operate itself, but it does become stuck. It develops defects and needs some maintenance.
Bernie is more of a Democratic Socialist posing as a Social Democrat. I dislike him.
but only because it -- like Qatar, Brunei, Kuwait and UAE -- exports massive amounts of fossil fuels
Norway also has a massive fund and planning for decommissioning its petroleum economic machine.
I haven't seen any rankings of after-tax per-capita GDP, but Norway would have a much poorer showing in such a ranking
They tend to measure their median incomes in after-tax income, and it tends to be competitive with the US. The Lorenz Curve doesn't account for taxes and benefits, when it really should. This is a sticking point, and one you're right to question; your prediction is incorrect, though.
The other Scandinavian countries have significantly lower before-tax per-capita GDP than the U.S
Of course. They're at about 85% of US. They also have lower crime rates, lower poverty rates, and better resilience to economic disruption. Sweden, as you point out, looks like the US, nearing 10% (year average around 8.8%) after the 2008 crash; Norway moved from 2.5% to 5%; and Denmark got up as high as 6%. These nations also had strong balance sheets during the recession--unlike Greece, Spain, Italy, and Portugal.
there's absolutely no reason the U.S. should want to emulate the Scandinavian Social Democracy model.
The most interesting component of these nations's economic model is their egalitarian social insurances. Their aid programs, unemployment insurances, and other such things pay large, long, and well above the poverty line. This drives their higher taxes, and offsets with some government assistance even for middle-income families. Their taxes are flatter because the effective tax rate for lower-income households, when deducting assistance from tax burdens, is lower.
That model is quite confused. We can do better than that.
Every gain in productivity--every per-labor-hour GDP increase, every per-capita income increase--comes from structural change. Assume you can make 30 donuts each hour. I get you a machine that stamps out the dough, which costs 500 hours of work to produce and maintain and which operates for 5,000,000 hours, or 0.36 seconds per hour. With the machine, you can make 300 donuts per hour--that is, 300 donuts per 3,600.36 working seconds. That's 9.999 times the production per labor-hour.
Great, but we all understand technology.
To make 300 donuts, I needed ten workers. Now I only need one. There's a single extra machine worker per 1,000 donut makers, so let's ignore them because they're statistical noise.
Now, donuts become cheaper. People aren't stupid: they economize. I'm enwrapped in price wars with the other bakers and we end up lowering the cost of donuts to a comfortable 10% profit margin. People are buying three times as many donuts, so we have about 7 of 10 workers displaced, and we're all making great profits.
People aren't buying ten times as many donuts. That's structural change. Folks have the money, but they're not buying more of your product. They're over there buying TVs or Netflix. Boston experiences an unemployment uptick and Silicon Valley becomes super rich as the money that was supporting Boston businesses floods into the tech market instead.
Real-life example: Baltimore. Detroit. Flint. Blackwater. Kansas City, Montana, soon, since a Harley factory--getting money from everywhere--is moving to Pennsylvania. Those 800 factory jobs draw income from everywhere, spend it locally, and support thousands of other local jobs. When that factory moves, your economy collapses, and it's not getting back up on its own: if your local economy had the entrepreneurial capacity to do something else, it would be doing both in the first place.
So what do we do?
Welfare pays until you get a job. The inflow of welfare money creates local spending--effective deman
The whole problem with your premise is that it requires omnipotence and premonition.
Dude I have certification in a field centered around this, and I don't have certification in the specific KA for people who took extended study in estimating. Yes, there's actually certification specifically for estimating, and for risk management, and a number of things.
In the specific case of the rovers they didn't know the severity of dust storms or how badly it would degrade battery charging.
The Rover should have had a more-detailed analysis. NASA was actually pretty accurate with their predictions of its lifetime; they had anticipated dust buildup on the solar panels, and the dust storms were actually cleaning the solar panels, which caused the discrepancy.
Space vessels are a different matter: NASA has plenty of experience getting them into space and operating them in space. They know the power drain and the conditions of operation. It'd be nice if there were some fuel left over after launch, except the launch stages drop off the probe and so that extra fuel is lost. This leaves NASA with a pretty good idea of exactly how much fuel they have once they get into space--and, besides, once they're in space they know how much fuel they have and can re-estimate from there.
That means NASA should be able to predict space vessel lifetimes pretty well. They're built to exhaust their fuel source, and NASA knows how long fuel lasts. NASA knows the operating conditions and the fuel demands. At some point, NASA should be able to determine the probe's lifetime, rather than calling it at 5 years and having half the fuel left at 10.
The same is, again, not true of things in orbit which we repair. We end up extending their lifetimes by intervention. This is also why nobody can predict how long a car will last.
Over-engineering is when you pay $350Bn but you needed $200Bn to finish the mission. That's waste.
Efficiency is when you need to pay $200Bn to finish the mission and you pay $210Bn to get twice as much out of it.
NASA has a record of claiming that things like the Mars Rover are performing amazingly, and that they don't understand how it's been able to keep going so long. They should have a risk profile that shows e.g. 50% likelihood of meeting mission objectives at $200Bn spend, 95% at $250Bn spend, 98% at $252Bn spend; and also, a graph of likely mission opportunities (e.g. extended operating lifetime) at each spend level.
In other words: it's perfectly okay to say you built this to have a 98% chance of completing the mission and that there's a 50% chance of going an extra 2 years and a 20% chance of going an extra 5 years. It's deceptive to say you built this with an expected lifetime of 5 years and holy shit it's still going strong at 15 years! An expected lifetime of 5 years could be that you had a 6-month variation, and could be 50% likely to make it to 5 years, 90% likely to make it 4.5 years, and 10% likely to make it to 5.5 years. It could also mean that you had a 50% chance of failure by 5 years and a 10% chance of failure by 6 months, so you built it to have a 5% chance of failure by 5 years--and it has about 50% likelihood to make it 8 years and a 10% likelihood to make it 12 years.
That all sounds fancy, but it's standard. More than that, it misses the elephant in the room.
NASA didn't build Kepler or the Rover to last a few years before breaking down from component wear.
NASA built them to outlast their mission, and provided fuel expected to exhaust in the mission time.
NASA claimed this truck got 12mpg, fueled it up with a 35gal fuel tank so it could go 420 miles on a tank, and got 1,050 miles out of the tank because it actually goes 30mpg. WTF?
Type of Demogrant. UBI tries to pay out enough to live; a Citizen's Dividend pays a dividend.
I designed one that would have been an effective $300Bn tax cut in 2016, paying $6,000 per adult in 24 payments throughout the year, based on 1/8 of personal income and corporate profits as the FICA source. You build Social Security retirement and disability on top of this, along with a welfare system.
The Dividend doesn't supply enough for anyone to survive; rather it tends to strongly increase buying power and effective demand in impoverished local economies. That creates jobs around those who need jobs where unemployment is high. You supply welfare to ensure access to needs of individuals, and a dividend to couple the economy to itself and ensure that any wounds heal. Quickly.
A Dividend as such grows faster than inflation. It grows with the GNI/C trend, which is always faster than inflation in medium terms (1-3 years).
People propose UBI as an alternative to work and welfare, absolving the government of responsibility. A Citizen's Dividend isn't an alternative to work and welfare, but rather a system to ensure people get a fair share and that the economy stays healthy and provides employment where employment is needed.
Slashdot Mirror
Make sure the machine remains visible at all times. Put plastic tape over any access hatch so you know if it's been tampered with.
Lexan case, locks, sensors, alarms--visual and audio.
If the paper is then counted by machine, then fair enough
Analog failure: machine misreads ballot. Direct-recording machines can display the ballot as it is understood.
Paper is easy to compromise: you can manufacture, alter, and lose it. The same is true of all data, really. You only have integrity inside the polling center.
Public view?
Your main weakness is recounting. Let's work backwards from electronic for this one.
With electronic voting machines, once you've proven the integrity of the machine, you can prove the ballot set. Everyone votes, the machine displays some kind of computation, and only then do you enter the machine to remove the ballots (digital media copy, whatever).
For provisional ballots cast in-person (wrong polling center, etc.), you load the provisional ballots into your hash set by the same rules, marking them as provisional--and you don't include identifying information in that set. This still exposes identity to the public if your polling center collects only a few provisional ballots (if it's only one, we know whose is provisional), which is the same problem as very few people voting (one polling center here had 67 people vote in the primary, with only three Republicans who all voted for the same candidates, so I can give any of them their exact ballot back. Imagine three people voting in the General by provisional in a pile of 1,600 votes). The solution is to obscure who is voting provisionally.
Under this scheme, ballots are traceable: the central authority publishes each ballot set with its polling center. We can all recompute the hash and prove the ballot set is untampered. They can strike any provisional ballot, so long as they leave its contents up in the ballot set and mark it rejected.
As you note, absentee ballots come in the mail, are paper, are not observed, and so represent a weakness. Disenfranchisement represents a larger threat, so we accept this weakness. On military base and ship, we can actually run the same kind of public election as here, observed by the public of the men on ship--who of course will leak if they're not allowed to record election ballot traces.
So what about paper?
Well, you can slip a piece of pencil under your fingernail and spoil ballots when counting paper in full public view. You can miscount. Election judges can collude, despite the whole reliance on people not colluding (we've seen this in other places).
We perform our recounts days or weeks later, with ballots kept in a secure location. Again: how do I know anyone in the chain-of-custody isn't just a group of 3-5 people who can print a new seal and are fully-willing to collude? Buy a few people off, get them an election pen (remember what I said about pencil?), and they can create a few new spoiler ballots.
In an election with 100,000 ballots, one state here recounted with 3,500 fewer, and claimed they must have just found more spoilers (without explaining which were probably not originally spoilers). We just accept that. Three percent of votes.
See how many avenues of attack we have? Now look at our electronic voting machines. Yeah, hell no. The stuff out there is horrible. That doesn't mean paper is safe; it just means we damned well better move to stronger EVMs if we're going to move off paper.
Candidate wins by 2000 votes and only 1000 advanced votes, why bother counting
Computing whether a candidate won can be...difficult. Take any Smith-efficient system. You have to compute all the one-on-one pairings: if a ballot ranks Bill 1, Anne 2, and Jim 3, then Bill vs Jim is 1 vote to Bill, Bill vs. Anne is 1 vote to Bill, Anne vs. Jim is 1 vote to Anne. These pairings are important.
To get the Smith set, you find out who wins these.
Let's say Bill, Anne, and Jim each get a majority of rankings above Dave and Mary. Dave and Mary categorically lose against these other three. Dave and Mary are out.
Now we have Bill, Anne, and Jim. Bill beats Anne; Anne beats Jim; Jim beats Bill. O
Diebold made ATMs. They're money-changers. They see a need for a system that collects and transmits votes.
When you build a microwave oven, you don't load it up with fork-detection hardware. You tell people to never microwave a fork. In an election, people are intentionally trying to microwave a fork, and you need defenses against that. Diebold, ES&S, Hart, these people built a microwave to accomplish microwaving food, not to accomplish microwaving food in hostile territory where even the people who bought the microwave are trying to sneak forks into the microwave.
I'm getting into the business. It starts with security. Gonna be a while before I'm ready to play, but you know what? Everybody hates all of these people. They have no idea what they're doing. I'm going up against giants, but you gotta realize: I'm not competing with them; they're competing with me.
So here's the thing: computers can be made impenetrable; physical locks can't. Computers, however, are immensely difficult to validate: we can ensure a small code body (maybe 4,000 lines) is correct, but we can't ensure the whole OS stack, the enormous application in its entirety, all libraries, and so forth are correct.
You know that thing where you can't hack into a computer that's unplugged and unpowered? You can totally hack into a lock by getting a large enough hammer. Any lock. You can manufacture a duplicate lock, a duplicate seal, a duplicate ballot box.
A computer that's powered on, only accessible via port 443 with TLS, and configured to reject a client not presenting an appropriate authentication certificate has a relatively-small code path to deal with. Unfortunately, at that level, you're facing the operating system's network stack, its firewalling stack, parts of the TLS library, and anything in the application that tries to identify the user by reading the data (which should just be using the TLS library to ask for the principle). That's a lot of code. If all of that is secure, holes everywhere else don't give hackers access unless those hackers are also valid, authorized users trying to exceed their access.
We would have to isolate the device from the network entirely.
And there you go.
Voting machines only need to collect and package votes. They need to exist for one day, and they only need to release their contents in a batch--or in a one-way stream such as by FEC light transmission. Folded Reed-Solomon at 5,000% correction, a flashing LED, and a photosensor on the thing across the room would let you safely transmit votes during voting, if you really want. Thing is you still need to batch everything up at poll close and provide proof of the ballots in public view to confirm the vote tally.
So we're down to a single approach: unplug the machine from the Internet entirely. You vote, it collects votes. At poll close, it computes some kind of reproducible hash of the ballot set, which it displays for all observing. We all write it down, take pictures, whatever. Then it transmits votes--possibly by someone opening the locked door, plugging in a USB device, and copying the ballot set off physically.
With no surface to attack, it's secure.
You still need to prove the integrity of the machine at poll open. Remember: software manufacturers, hardware manufacturers, and all elections staff and authorities are also threats.
First, with aren't they using smart cards with passwords on the keys?
Because voting machine manufacturers are money-changers, not security experts. They make ATMs, they make voting machines.
Second, why did the software permit weak choices?
Made to order.
Third, why are infosec officers not replacing those pages in the manual, training users in proper procedures, rejecting the products at user acceptance or running tools for weak password detection?
The same companies who provide the machines also provide voter outreach and elections consulting. They'd be the ones deciding if you should reject this shit.
Why do you think I'm starting a business and breaking into that industry? Paper ballots look kind of like the Internet to me (from thousands of polling places, through the hands of small groups of trustees, flowing to central State offices... looks just like all the online banking traffic), and the electronic voting industry is run by people with no background in IT security or risk management. Everything is run wide open and vulnerable to all kinds of attack. Somebody needs to fix it.
Let's put aside the issue of using a password based system in the first place.
No, let's not do that. It never even occurred to me to use passwords when I stared writing the SAFE VOTES guidelines (elections procedures and standards specifically for elections run via direct-recording electronic voting). I didn't even require 2FA--largely because using the credentials to do anything permanently takes the machine out of commission, causes alarms, and generally draws a whole lot of attention and stops your election, but also because the credential is created at poll open and is destroyed at poll close that day.
the official manual will say use a secure 10 character, upper, lower, special, & number password.
Assuming the accounts are persistent and long-lived, I would have said to use either TOTP, PKI (FIDO U2F, Smart Card), the two as a 2FA pair, any password plus TOTP or PKI as a 2FA pair, or a single-factor password of at least 14 characters and preferably generated as four random dictionary words (49 bits with 5,000 word dictionary and no variations, vs. 52 for 4-class 8-character).
TOTP has a huge problem: if you demonstrate the installation of the software image on the machine, such that you can prove an untampered image, then the public observes the TOTP credential. You can't keep it secret. This is also true of entering a password.
State level, you need a LOT of corruption across a highly distributed network of independent voluntary organizations to impact a vote
The votes move from hand to hand from thousands of locations through small groups of trusted parties. That's tens of thousands of weak links--millions in some states.
The paper based voting system in the US that has been used for decades is pretty good.
It is. It's a complex mess with enormous integrity problems, but it's only severely-abused in a few places--and that severe abuse is relatively minor. A strong, decisive victory or even just huge voter turn-out is usually enough to overcome the level of tampering even in states where it's rampant.
We estimate about 30,000 ballots just tossed out in Florida every year. Ohio did a recount once and came up 3,500+ ballots short, with the excuse that "it's probably spoiled ballots" but no trace to show which ballots were spoiled (yeah, "We're not sure, we guess it's X, it's probably fine, trust us" fuck you).
Small numbers.
My concern is we can't prove integrity. I can fix that with electronic voting machines, so long as we follow some god damned strict standards. Give me a little time; we have to do this right.
end voter manipulation via social media that has been their only real threats.
You should see paper ballot security considerations. Many of these are built on the idea that you can trust some party by putting other parties around them, under the theory that they're adversarial and will act as checks on each other; unfortunately, we have Democrats endorsing Republican candidates now, and besides that you can just join a party and carry out espionage (really, vote tampering is an enormous offense and an attack on our national security).
You can fix that with voting rules that reflect the voter's preferences better. The combination of Single Transferable Vote for multi-winner elections, the Big Party Rule (a party with 25%+ registered voters can nominate two candidates for single-winner elections, but only by primary election using STV), and a strategy-resistant Condorcet method like Tideman's Alternative Smith would provide pretty good protection. Electoral Fusion helps, too.
Big Party Rule upends the base voter issue: a ton of people will vote R or D, and now they have two R and two D candidates. You get Trump vs. Rubio, Bernie vs. Hillary. Combine this with any Condorcet system and you now need a mutual majority favor.
Under the Big Party Rule, you would probably get Bernie from the left and Hillary from the moderate Democrats; interfere with a huge pro-Bernie propaganda campaign and you get excitement for Bernie, but who is the second choice? It's ... still Hillary. The same is true of Trump and a more-moderate candidate like Rubio.
Hit the General Election and now what?
Well the entire moderate independent vote is going Hillary-Rubio or Rubio-Hillary. The moderate R goes Rubio-Trump or, for the Never Trump crowd who voted Hillary, Rubio-Hillary. The Pro-Trump crowd will go Trump-Rubio-Johnson-Hillary because they want to shoot down Bernie Sanders by putting Hillary above him (they might or might not rank Stein above or below Bernie at the tail end). The Bernie crowd is going Bernie-Stein-Hillary-Rubio, Bernie-Hillary-Stein-Rubio, or so forth.
In this scenario, a huge amount of the population is trying to NOT elect Trump or NOT elect Bernie. That makes those two unelectable. The voters on the left have said they'll concede to the Republicans if they take Rubio (that's what actually would have happened in this line-up, based on the data I've been able to dig up); the voters on the far right have said they'll settle for Rubio if not Trump; moderates are taking Rub
Think about the Internet. There are thousands of points where data leaves, moves through intermediate routers, changing hands, until it reaches a location. Your bank, for example.
Why do you use encryption when sending your credit card details? It moves from your PC to your ISP, through several core routers. Random people can't sniff it. Your traffic isn't exposed to untrusted third parties; arguably, all of the systems through which your data flows are more-trusted than the merchant to whom you're sending your credit card number!
"Computers can be hacked", right? What if the "hacker" is a technician at an ISP? Remember that guy who did that with the lottery?
Paper ballots begin at polling centers, where we trust staff to not manufacture additional ballots in the back and swap out ballot boxes with duplicate seals. They transfer through a chain of custody, always in the hands of 3-5 people who we assume aren't colluding, or stored in a secure location we assume isn't intruded upon by someone with a key and admin access to the security logs. They change hands several times.
That doesn't even get into the error rate of counting--with the implication that you can manipulate the error rate. People freak out about that when it comes up: Kris Kobach in Kansas was giving instructions about what ballots to not count (spoiler ballots, in late, etc.), and he was also a candidate in the GOP primary for Governor. You get a lot of individual judgment with hand-marked paper ballots especially--is that a stray mark, or just a little outside the circle? Is it enough to flag to an observer that the purchased vote has been cast faithfully? Does the signature sufficiently match the one on file? All up to personal taste.
Paper ballots are a complex, flimsy, and insecure network.
I don't see why the screen isn't rectangular with software round corners.
We could put a white OLED over the camera lens to act as a backlight for LCD; or put the camera lens behind the OLED screen itself. Either way, turn off the screen in that area and you've got a transparent cover over your lens, same as today.
It wasn't sound in 1960...the rest of your argument hangs on that fallacy.
No it doesn't. It relies on something called "technology", whereby the amount produced per working-hour increases.
You think SS payments haven't also gone up?
The COLA-adjusted Social Security payments in 2016 are 7.83x as many dollars as the benefit in 1960.
The number of actual dollars of income per American over the age of 65 in 2016 is 11.77x as many dollars as the benefit in 1960.
In other words: for every $7.83 that Social Security needs to pay out in 2016, there is $11.77 available at the same FICA tax rate--IF applied to the same portion of all income. Put another way, we have $1.50 available taxable dollars for every $1 of taxable dollars we need to cover COLA-adjusted Social Security benefits starting in 1960, assuming we're taxing the same portion of all income,
It turns out that, yes, there's more available every year. What we do: in 1980, take 8% of 80% of the money. In 1990: take 10% of 60% of the money. In 2000: take 12% of 40% of the money. So in 1980 we had 6.4% of the money, in 1990 we had 6% of the money, in 2000 we have 4.8% of the money, etc.
What part of this don't you understand?
The PCB is actually pretty cheap, likely around 10 cents to a quarter. The MicroUSB port is a 26-cent part; the MicroHDMI connector is around 93 cents (yeah). Looks like you can get a MicroSD slot for around 96 cents in 1,000 quantity, around 74 cents in 2,000 quantity. $2.19. Not bad...prices have improved. If pressed, you might be able to get the MicroHDMI down around 40 cents; the MicroSD has springs and other mechanical bits. Call it maybe $1.50?
The Pi has 62 components, including SOC, plus 5 connectors, at an average 7.4 cents each. Given the above, estimate 5.6 cents each. The SOC seems to be around $1.50 in volume, leaving 3.2 cents per small component.
A TV tuner is an immensely-complex device with multiple large circuits, and has to decode color channels separately. There are a lot of parts in there. That's ... actually changed since I last bothered to look: not only is TV now digital, but people came up with CMOS tuner ICs--expensive to set up, but cheap to manufacture in quantity, eliminating much of the cost.
Wow okay.
The weather patterns weren't unusual and unpredicted; the probe's performance given those patterns was.
You don't seem to understand anything about estimating. Pathfinder gave NASA ground data to correlate with telescope data, enabling them to model the Martian atmosphere--that was the point. The biggest miss on Opportunity was predicting a continuous coating of dust on the solar panels, which didn't happen at all.
A month is enough to get a read on insolation at ground versus in space versus weather, by the way, so NASA has a pretty good estimate of year-long insolation data, and can work out charging rates from that to relatively-high accuracy because they have a lot of data about how atmosphere interacts with insolation. They also knew a lot more than you do about batteries, and estimated a shorter battery life and higher wear.
As to extrapolation versus interpolation, most estimating isn't exactly that. A lot of estimating is "A is like B, and X is like Y except smaller/bigger and made with B. We've been using a new technique to do C, which is completely-unlike B, but should apply to parts of the process of doing B, so B should be improved in budget and time by these factors. Doing X should thus come out in this manner. Here's the time, budget, and resources necessary." We've never done anything like this, but we know about how much it should cost and how long it should take--and what the variation is in our estimates (i.e. risk).
They didn't just design for the worst-case scenario. They overbuilt.
What's your excuse for NASA not being able to figure out how much fuel long-term space probes consume?
If you expect anybody to approve increased spending and cuts in the future, you are crazy.
This actually reduces tax burdens without cutting programs. It's revenue- and deficit-neutral. Yes, it sounds like black magic; it's an accounting trick. There are further reductions in the future because of downstream effects; I don't account for them first-pass.
Also fund the SS trust with good assets _first_.
The Trust has good assets. It holds Federal Treasuries and they perform their function quite well. The problem is its FICA source is broken. The amount of income available per retiree has increased year after year much faster than inflation, yet the proportion of income exposed to FICA taxation has decreased. In other words: 2x as much income is being made, and 40% of that is exposed to FICA taxation (=80% as much being taxed, thus falling revenue).
Increase taxes now for spending cuts in the future should NEVER be believed.
The actual proposal in 2016 was a 33.5% CIT (vs 35% at the time) and a blunt 3.2% reduction in the top personal tax rate. I don't have the data to tweak personal income brackets further or I'd have set it back up to 39.6%.
Under the 2016 rough model as such, a 2-adult household making $400,000 would have taken home an extra $1,674/year. At $50,000 of income, that household is taking home $8,070/year additional after taxes and Dividend. Meanwhile a household with $25,000,000 of income is taking home an additional $841,117 after taxes.
The proposal is actually to restructure programs now to pay out cash to a lot of people, offsetting tax burdens and acting as an effective tax cut in order to reduce costs, increase revenues at current tax rates, and lead to further tax cuts later while we roll out even bigger government services and pay off our debts.
So let's divide the Gross National Income for the United States by the number of Americans age 65 and older. Then, let's compare that to the same value in 1960 plus inflation since 1960.
Have a look..
Without a cap on the FICA, a 6% FICA rate in 1960 would have needed to come down to a 4% FICA rate in 2016 to avoid simply loading the Trust up with money it was never going to spend.
The treasuries are as good as cash--better, really, because they avoid the economic drain of simply taxing money out of circulation and totally avoid interest. The Treasury could tax $10 billion in 1995 and spend it; but the Social Security surplus of $10 billion is already here. If we spend that $10 billion in 1995 and then put it back in 2005, then we're theoretically only spending $5 billion of 1995 money.
Moreover, if you raise taxes to pay it back, you only need to raise them 40% as much as you would have to fund the $10 billion originally. It's not just inflation: the GNI/C is higher by more than inflation, and the number of people is also higher. The longer the US holds debt, the easier it is to recover that debt with fractionally-small taxes.
Think of it this way: Inflation at 2% over 15 years is 35% (yes, I used a rough 50% number up above). Productivity growth of 1% means the number of inflated dollars per person, 15 years later, grows by 56.25% (1.02% * 1.01% each year). Population growth of 3% per year means you have an increase of 55.8% population. Take the 156.25% GNI/C and multiply that by 155.8% C and you get 243.45% GNI increase.
That's 2.4x as much money out there to tax, and 1.35x as much to recover with inflation. You're way ahead. If you raise taxes now to pay for your debt, you only have to raise them about 56% as much as if you'd gone the revenue route initially, assuming your interest rates are in-line with inflation.
So your nation needs to not bury itself in debt; it can leverage debt pretty damned well, though, and should absolutely favor being indebted to itself over holding piles of cash.
A working UBI doesn't exist. A universal dividend provides a basis for social security and welfare, while keeping your economy functional. You need your social insurances and welfares because the standing income necessary to stabilize temporarily-displaced workers is huge, and the standing income necessary to eliminate all recessions and end the existence of poor inner cities as a phenomena is small.
There's no reason Social Security should have ever been insolvent. In practice, it should have been able to reduce the FICA rate over time while increasing benefits.
The funding structure is just broken.
Most UBI proposals are for about $500/month, and even that requires dismantling Social Security, which would deprive tens of millions of people of their retirement income, generating a firestorm of political opposition.
I drew up a Universal Dividend proposal that pays $500/month to each adult while making social security permanently-solvent. I dismantled nothing, raised no taxes, and created no additional deficit.
There's a $900 billion increase in Federal outlay (I restructured $1.1 billion into $2 billion). When you remove the amount of the benefit offsetting its own funding from this, you find a $1,200 billion reduction. That's net $300 billion reduced taxes.
For example, if you pay $4,000 into the FICA and receive $6,000, that's $4,000 offset. If you're wealthy and you pay $100,000 into FICA and receive $6,000, that's only $6,000 offset.
Because that offset exceeds the additional outlay, we can adjust the general tax brackets to draw more income at various levels and create tax cuts across the board. The base model even includes a 1.5% corporate income tax cut (from 35% to 33.5%).
You provide welfare and social insurances on top of this. The Dividend keeps economies healthy (yeah, huge job creator, breaks recessions before they begin) while your aid programs keep families fed, clothed, and housed. I didn't dismantle the welfare system when I ran the numbers.
There are two things countering that argument.
First, marginal opportunity cost.
Think about everything you would buy if you had unlimited money. You'd buy food, clothing, rent, car insurance, games, fast cars, rocket ships.
If we cut your money back, you start to prioritize. You must eat. Rent is pretty important. You need your car to get to work. The tighter it goes, the more you refine your priorities.
One of your priorities is savings.
Extremely-rich people will buy financial securities such as stocks, bonds, and commodities on the secondary securities market. When we issue stocks and bonds, money goes from the buyer to the business (or government); when we trade them on the secondary securities market, money moves back and forth between traders. These people are essentially sucking money out of your 401(k).
Poor people can't afford all that, and the middle-class don't make much money trading stocks. Middle-class may have a modest emergency fund; the poor don't.
Those things don't produce economic activity. They're just idle cash. When you move some of the income down (please no asset taxes), it moves into the hands of people whose marginal opportunity is in spending, creating revenue and jobs.
The second factor is distribution.
Distribution is simple: a poverty-stricken area is full of people who can't work because people are too poor to buy, thus there isn't revenue to pay wages. Those people are unproductive (and they cost welfare).
Moving spending power into high-unemployment, low-income areas creates jobs. That means people work, produce, and draw taxable income. That income is linked to production, so your nation's overall wealth and wealth per person increases (these people are working and producing).
You know how salt bridges form in water softeners and you have to open them up and break them apart? It's sort of like that with capitalism. The system figures out how to operate itself, but it does become stuck. It develops defects and needs some maintenance.
Bernie is more of a Democratic Socialist posing as a Social Democrat. I dislike him.
but only because it -- like Qatar, Brunei, Kuwait and UAE -- exports massive amounts of fossil fuels
Norway also has a massive fund and planning for decommissioning its petroleum economic machine.
I haven't seen any rankings of after-tax per-capita GDP, but Norway would have a much poorer showing in such a ranking
They tend to measure their median incomes in after-tax income, and it tends to be competitive with the US. The Lorenz Curve doesn't account for taxes and benefits, when it really should. This is a sticking point, and one you're right to question; your prediction is incorrect, though.
The other Scandinavian countries have significantly lower before-tax per-capita GDP than the U.S
Of course. They're at about 85% of US. They also have lower crime rates, lower poverty rates, and better resilience to economic disruption. Sweden, as you point out, looks like the US, nearing 10% (year average around 8.8%) after the 2008 crash; Norway moved from 2.5% to 5%; and Denmark got up as high as 6%. These nations also had strong balance sheets during the recession--unlike Greece, Spain, Italy, and Portugal.
there's absolutely no reason the U.S. should want to emulate the Scandinavian Social Democracy model.
The most interesting component of these nations's economic model is their egalitarian social insurances. Their aid programs, unemployment insurances, and other such things pay large, long, and well above the poverty line. This drives their higher taxes, and offsets with some government assistance even for middle-income families. Their taxes are flatter because the effective tax rate for lower-income households, when deducting assistance from tax burdens, is lower.
That model is quite confused. We can do better than that.
Every gain in productivity--every per-labor-hour GDP increase, every per-capita income increase--comes from structural change. Assume you can make 30 donuts each hour. I get you a machine that stamps out the dough, which costs 500 hours of work to produce and maintain and which operates for 5,000,000 hours, or 0.36 seconds per hour. With the machine, you can make 300 donuts per hour--that is, 300 donuts per 3,600.36 working seconds. That's 9.999 times the production per labor-hour.
Great, but we all understand technology.
To make 300 donuts, I needed ten workers. Now I only need one. There's a single extra machine worker per 1,000 donut makers, so let's ignore them because they're statistical noise.
Now, donuts become cheaper. People aren't stupid: they economize. I'm enwrapped in price wars with the other bakers and we end up lowering the cost of donuts to a comfortable 10% profit margin. People are buying three times as many donuts, so we have about 7 of 10 workers displaced, and we're all making great profits.
People aren't buying ten times as many donuts. That's structural change. Folks have the money, but they're not buying more of your product. They're over there buying TVs or Netflix. Boston experiences an unemployment uptick and Silicon Valley becomes super rich as the money that was supporting Boston businesses floods into the tech market instead.
Real-life example: Baltimore. Detroit. Flint. Blackwater. Kansas City, Montana, soon, since a Harley factory--getting money from everywhere--is moving to Pennsylvania. Those 800 factory jobs draw income from everywhere, spend it locally, and support thousands of other local jobs. When that factory moves, your economy collapses, and it's not getting back up on its own: if your local economy had the entrepreneurial capacity to do something else, it would be doing both in the first place.
So what do we do?
Welfare pays until you get a job. The inflow of welfare money creates local spending--effective deman
The whole problem with your premise is that it requires omnipotence and premonition.
Dude I have certification in a field centered around this, and I don't have certification in the specific KA for people who took extended study in estimating. Yes, there's actually certification specifically for estimating, and for risk management, and a number of things.
In the specific case of the rovers they didn't know the severity of dust storms or how badly it would degrade battery charging.
The Rover should have had a more-detailed analysis. NASA was actually pretty accurate with their predictions of its lifetime; they had anticipated dust buildup on the solar panels, and the dust storms were actually cleaning the solar panels, which caused the discrepancy.
Space vessels are a different matter: NASA has plenty of experience getting them into space and operating them in space. They know the power drain and the conditions of operation. It'd be nice if there were some fuel left over after launch, except the launch stages drop off the probe and so that extra fuel is lost. This leaves NASA with a pretty good idea of exactly how much fuel they have once they get into space--and, besides, once they're in space they know how much fuel they have and can re-estimate from there.
That means NASA should be able to predict space vessel lifetimes pretty well. They're built to exhaust their fuel source, and NASA knows how long fuel lasts. NASA knows the operating conditions and the fuel demands. At some point, NASA should be able to determine the probe's lifetime, rather than calling it at 5 years and having half the fuel left at 10.
The same is, again, not true of things in orbit which we repair. We end up extending their lifetimes by intervention. This is also why nobody can predict how long a car will last.
Over-engineering is when you pay $350Bn but you needed $200Bn to finish the mission. That's waste.
Efficiency is when you need to pay $200Bn to finish the mission and you pay $210Bn to get twice as much out of it.
NASA has a record of claiming that things like the Mars Rover are performing amazingly, and that they don't understand how it's been able to keep going so long. They should have a risk profile that shows e.g. 50% likelihood of meeting mission objectives at $200Bn spend, 95% at $250Bn spend, 98% at $252Bn spend; and also, a graph of likely mission opportunities (e.g. extended operating lifetime) at each spend level.
In other words: it's perfectly okay to say you built this to have a 98% chance of completing the mission and that there's a 50% chance of going an extra 2 years and a 20% chance of going an extra 5 years. It's deceptive to say you built this with an expected lifetime of 5 years and holy shit it's still going strong at 15 years! An expected lifetime of 5 years could be that you had a 6-month variation, and could be 50% likely to make it to 5 years, 90% likely to make it 4.5 years, and 10% likely to make it to 5.5 years. It could also mean that you had a 50% chance of failure by 5 years and a 10% chance of failure by 6 months, so you built it to have a 5% chance of failure by 5 years--and it has about 50% likelihood to make it 8 years and a 10% likelihood to make it 12 years.
That all sounds fancy, but it's standard. More than that, it misses the elephant in the room.
NASA didn't build Kepler or the Rover to last a few years before breaking down from component wear.
NASA built them to outlast their mission, and provided fuel expected to exhaust in the mission time.
NASA claimed this truck got 12mpg, fueled it up with a 35gal fuel tank so it could go 420 miles on a tank, and got 1,050 miles out of the tank because it actually goes 30mpg. WTF?
It turns out the tank only needed to be 14gal.
Type of Demogrant. UBI tries to pay out enough to live; a Citizen's Dividend pays a dividend.
I designed one that would have been an effective $300Bn tax cut in 2016, paying $6,000 per adult in 24 payments throughout the year, based on 1/8 of personal income and corporate profits as the FICA source. You build Social Security retirement and disability on top of this, along with a welfare system.
The Dividend doesn't supply enough for anyone to survive; rather it tends to strongly increase buying power and effective demand in impoverished local economies. That creates jobs around those who need jobs where unemployment is high. You supply welfare to ensure access to needs of individuals, and a dividend to couple the economy to itself and ensure that any wounds heal. Quickly.
A Dividend as such grows faster than inflation. It grows with the GNI/C trend, which is always faster than inflation in medium terms (1-3 years).
People propose UBI as an alternative to work and welfare, absolving the government of responsibility. A Citizen's Dividend isn't an alternative to work and welfare, but rather a system to ensure people get a fair share and that the economy stays healthy and provides employment where employment is needed.