I’m with Paranoid Randroid on the idea that this is of thing that should be researched. I’m not sure i it’s something that could work now or we need our coming utopia/dystopia for it to be useful, but that is the point of testing. We need to see how people act under such a system to see where it’s useful.
I’m not sure where you are going with that. I was thinking of any legacy factory that needs to be converted to full automation. It’s incredibly hard, and usually you wind up leaving people in place all over because they do things that just can’t be automated. For instance, you need operators to clear jammed machines, to grab extra input materials when waste goes up and leaves you short, etc. A million little things that have to be done each day that we can’t automate.
Then you run into problems in the backoffice with processes that are poorly defined and require humans to pick up the slack, or legacy systems that can’t talk to control systems, inventory and other processes that have too much manual activity to automate, etc. We worked on one factory for probably ten years trying to solve issues like this.
Fun fact: the push to automate doesn’t really involve saving labor costs in most places. Automation is bleeming expensive. Rather, factories are doing it to improve quality, and also because there is a serious labor shortage in many of them and it’s getting worse. The average factory worker is now in his or her mid to late 50’s, and it’s a hard sell getting millenials to work in factories. One factory manager told me they have had a position open for months, and interviewed a number of people and offered the job to many, and they just never came back after the tour. And this is a modern factory, clean and pleasant to work in, with a good salary and benefits.
What they are really afraid of is what happens in five or ten years when these workers retire en masse, taking decades of domain and process knowledge with them. So some factories are frantically trying to collect and digitize this knowledge while they can.
Sure. Especially true for new factories that can build in support for automation right from the start. But that’s not easy either - ask Tesla.
Farm employment has been going down steadily for over a hundred years, while farm output has increased dramatically. Yet there are no armies of unemployed farm workers.
Don’t forget online resources like StackExchange and other internet code repositories and information sharing sites. Problems that used to be solved from scratch repeatedly all over the country are now shared, and this has saved a huge amount of work. I would guess your average programmer today is at least 20x more productive than we were in the good old days, and maybe much more than that. Hell, I can remember spending days building simple drivers for hardware, implementing quicksorts, spending huge amounts of effort to build simple databases, etc.
And yet… I notice there is more employment in the software industry than ever. All that efficiency wasn’t used to get rid of labor, it was used to build better and more sophisticated software. And as software got better, the market for it grew.
Think hard about that filing job, and what other little tasks went along with it. Think of the people you had to communicate with, the judgement you had to use, and ask how hard that would be to replace with a computer.
When most people talk about automating jobs, they often reduce the job down to a label or caricature, and they don’t think about the real complexity. So a self-driving truck will obviously replace truck drivers, because they drive trucks and a computer does that now. But if you deep-dive into what truck drivers actually do all day long, you find out that the physical act of manoevering the truck down the highway is only a part of what they do.
This is the kind of analysis I’ve been doing for a long, long time. Analyzing factory work to figure out how to automate it, gathering requirements, observing people doing their jobs to figure what hidden processes there may be, etc And it’s not nearly as easy as some people assume it would be. Very few factories even have their processes perfectly defined and still require human judgement and paper processes to keep the place running.
And we are a long, long way from black boxes that can just be plugged together. These machines interact with each other. They have different tolerances. They respond differently to heat and humidity. They vibrate and transfer vibration to the floor and other machines. And so it goes.
As a result, individual machines may be operating at spec, but put a few together and complexity rears its ugly head and things can get weird. There is an entire field of Statistical Process Control that is used to analyze such effects to lower waste and improve operational efficiency. Nothing about standing up a new factory, or automating an existing one, is easy. And factories are the low hanging fruit of automation.
One other thing - when most people think about factories, they think of assembly lines. They think of a car factory with robots doing welding and other tasks. But most factories are not like that. Take a factory that makes jet engines. The engine is built from subassemblies. It may sit at a station for a week while people come by and install parts, plug in test equipment to check things, test for tolerances, remove parts that don’t make it, etc. Then the thing moves to the next station where the next level of assembly starts. Composite parts are still layed up by hand, and so forth.
Some such factories simply cannot be easily automated. Sure, you’ll find maybe an automated spray booth, or an automated Kiln or something, but ultimately 90% of the assembly is done by people turning wrenches. Automating a factory like this is so far the future that these companies don’t even have such projects in their long term project plans (and they plan out decades in advance sometimes).
Then there all the small bespoke shops that make custom parts, and endless supply chains that communicate amongst themselves and organize activity. This is done by electronic data interchange, but also by humans talking to each other. Many today are not even capable of being automated.
One day we might become a post-scarcity society due to AI and automation. But if it happens, it’s because we will slowly evolve into it. And the society that exists after having slowly adapted to increasing automation might not look much like ours today. So who knows how they will react to this? Maybe our economy will become based on non-renewable products, real estate, personal attention, or some other commodity we don’t value today.
This isn’t going to happen in a decade or two. Even if the ai’s were available now, just integrating them into industry in a way that can replace humans would take decades, and probably more like 50+ years, with the plus being maybe an order of magnitude greater. It’s even possible that we’ll go only so far before unintended consequences or rising difficulty as we pick the low hanging fruit just makes it uneconomical, and we have an economy with a significant increase in automation while still maintaining a significant enough work force that the economic growth caused by the automation more than makes up for it, and all rhose workers are employed in jobs created bybthe growth.
Which is what has always happened.
I think automation will happen regardless because companies will always look for ways to be more efficient. If UBI happens, it will happen because the sheer number of unemployed and underemployed people and the high levels of income disparity will demand it.
What makes you think they won’t still be living in poverty? I don’t believe UBI will provide everyone with middle class standards of living. It will likely provide enough to not starve to death on the streets.
Just below middle class is not poor. Where the payment gets set depends on a lot of things, including, I’d guess, how possible it is to find an unskilled job.
I’m still not seeing how UBI can financially work, if it’s meant to literally be some kind of universal income that is adequate to support someone who for whatever reason is not employed. I’m assuming this would include children, retirees, etc…
Let’s do some math here- we have roughly 300 million people. If the plan is to give each of them something like a minimum wage income, that’s roughly 4.5 trillion dollars that would need to be paid out.
The federal budget is only 4.1 billion in total- including military, social security, etc… Even if you entirely fold the social-type programs (SS, health, housing, labor) money into offsetting the UBI total (roughly 2.3 trillion), you still have something like 2.2 trillion to cover, and that’s outside of all the other stuff the Federal government does- military, science funding, arts, veterans’ benefits, the FDA, the USDA, education funding, department of energy, transportation funding, etc…
So by my reckoning, that amounts to raising the federal budget from 4.1 trillion to about 6.4 trillion, or if you prefer, by 153% And there’s also the risk that people would be deincentivized from working and the burden on the working people would only increase from there. Who in their right mind is going to vote for a proposal that raises their own taxes by half again what they currently are, and enables people to not pull their part of the load? Few I suspect.
Now you could make it some sort of stipend that would be helpful, but not a livable income in order to keep the costs in check, but then what’s the point?
A lights out factory is one where there are no people (or very, very few) and it is true that there are some jobs where people are a lot cheaper than machines. My point with the fab is that there can be high output factories with relatively few people.
Big issue in the early days of automation, where no one wrote down what they did and when outside systems analysts came in to automate stuff they wound up missing important steps in the process. But there are lots of examples where automation works. And plenty of incompetents to be sure.
I’ve mentioned before that I know of a case where a factory in SE Asia was not automated (though how to do it was well understood) because the workers were cheaper than depreciation on the machines. Sometimes you have to automate - some chip resistors and capacitors are so tiny that you need a machine to put them on a circuit board. But I don’t buy the labor shortage. There are plenty of MW workers who would love to get a higher paying factory job. (Make those jobs MW and maybe not.) I’ve been in plenty of big factories in the US, and they are all clean and well lit. The cannery where my wife worked and which was quite old was about the worst and even that was not bad.
Tesla is down the road from me. Plenty of car makers already do what they are trying to do. Tesla seems not to know how to manage their supply chain, how to control vendor quality, and the differences between large scale manufacturing and what they’ve been doing up to this point. More a case of Silicon Valley overconfidence than a fundamental problem with manufacturing. I’ve also heard some people who used to work in the plant before the current crisis, and they were not impressed with the management. Volume manufacturing is not a core competency of Tesla the way it is of GM.
Software got better? Could have fooled me. I think it is more the case that there are more places where software runs now. No market for apps before you had smartphones and tablets. No embedded software market before cheap microcontrollers. And IoT is going to blow it up more. Who knew that we might be writing code for lightbulbs? Even crap insecure code.
Almost no one, none, and I’m sure my job has already been replaced by a computer. It was medical records related, and all that stuff is online in my current medical office. Delivering mail was more intellectually challenging than this job was.
I’m dubious about automating truck driving at least partially because they’d be easy pickings for hijackers. Just read an sf story about this very subject.
I suspect you spend most of your time on problems, and so don’t really process the stuff that does work. I worked in IC quality and monitored field failures, and deep down I think that half the chips out there are going to fail any second. Not true but I never see the stuff that does work - even though I had the numbers.
Back in the mid-80s we implemented ways of using SPC to check to see if various testers were getting out of alignment so they could be fixed before they started to give bad results. Yeah, interfacing different boxes can be tough. Did you know that there is a lot of sophisticated technology behind plugging the pieces of a component stereo system together?
But the growth of manufacturing while manufacturing jobs are shrinking shows we know how to do this, though I accept that the people you deal with might not. (Or else you wouldn’t be called in.)
This seems like a relevant quote:
Some factories won’t get automated, they’ll get shut down as production migrates to more efficient factories.
You mentioning wrenches made me think that a lot of factory automation involves product design for manufacturing, not just changing the factory. For instance in mainframe assembly eliminating screws was a big deal. If you can design components so that they can be plugged into each other securely, as opposed to being screwed together, you save a ton of money as well as increase throughput.
Factory automation is a lot more than just bringing in a robot.
Yup. Silicon Valley overconfidence. I see it all the time.
The Facebook disaster is yet another example.
Here is a list of countries with tax revenue as a percent of GDP. The US is 26%. Sweden is 49.8%. South Korea is 33.6%. So there are plenty of countries which tax at levels which might make this kind of thing feasible.
I didn’t say that it couldn’t ever work, but that giving everyone a 2018-era minimum wage income would cost more than 1.5 times the federal budget, and that’s if you entirely scrap the social security and healthcare programs (i.e. Medicaid) to do so. And then expect the populace to be willing to absorb a 150% tax increase to do so.
And on top of that, I kind of doubt you can just shit-can all the social programs like Social Security and Medicare and assume that the UBI stipend will cover everything. Nope… they’ll still have to be around, and cost more money.
Finally, there’s the problem that if you redistribute the money like that, you’re only going to drive prices up on the lower end. If everyone with a minimum wage job now makes roughly 30k instead of 15k, prices are going to adjust upward as a result. Which is just going to leave the people living on the basic UBI stipend badly off anyway. Actually I suspect the market changes due to something like this would be interesting- they probably wouldn’t drive prices on luxury goods up too much- it isn’t going to line the pockets of anyone with money to buy many luxury goods- the tax increase will either offset the UBI stipend, or be more than the UBI stipend (probably the latter). And on the lower end of the income scale, it will drive prices up. Probably good for manufacturers and retailers, but the real question is whether that infusion of money into those segments is actually going to drive any job growth or just be extra profit for the companies serving those segments.
Actually, it would only $2 trillion a year, if it were gradually phased out at 50K and did not apply as much to children, before taking out Social Security and other non-medical programs (I assume medicare and medicaid would stay):
300,000,000 people in America
150,000,000 with income < 50K, since that is the median.
no clue what the numbers are for children and those who have some income are but let’s round down to 100,000,000.
$20K a year is $2 trillion.
$2 trillion is 1/2 the federal budget rather than 1 1/2, but it is still very large. You could completely eliminate unemployment, non-medical welfare, and much of social security. $20,000 a year is more than most recipients get, so we can assume that we can eliminate all of it and balance that off with ignoring the other aid programs that are eliminated.
So in actuality a guaranteed income would cost more like $1.3 trillion a year. Which is still pretty darn expensive for something with an unknown benefit.
Why are you assigning this only to people who make < 50K? If you are only going to give this to the poor, then it’s just welfare. And if you are going to take it away when people hit an income threshold, you just destroyed the purpose for going to a UBI in the first place, which was to eliminate the negative incentive effects caused when people lose their benefit if they take a job.
The U in UBI means universal. That means everyone gets it - or at least everyone of working age. That includes Bill Gates and Elon Musk. And it includes people like students, retirees, etc. Anyone who could conceivably work if they wanted to.
Anything else is just expanded welfare.
I’m not even for UBI, but given that the US has very low taxes compared to much of the developed world, raising them to support it is not an economy breaking matter, especially in that it helps reduce income inequality.
Political feasibility is another matter entirely.
Call it “just welfare” if you like; the objective is to ensure some non-impoverished standard of living, not necessarily giving everyone the same amount of money. The negative incentive effects will remain an issue, but they can be ameliorated by grading the payouts such that the marginal loss of UBI/welfare income per dollar of increased labor income is sufficiently low.
You’d think so, but ironically the SIME/DIME link on one of the posts that I saw quoted would imply that the marginal percentage of taxation/phaseout doesn’t have a measureable effect on the amount of work the recipients do. I say “ironic” because the people who would oppose this program are the same people who somehow think that marginal rates are very important to the motivation of rich people.
But the term “just welfare” is indeed just a word. If the phased cutoff ended at $100K instead of $50K, it would be very similar to social security which cuts off at around $140K, and there are a bunch of hardcore conservative voters who would deny that SS is welfare, since “welfare” is what those other, shiftless, undeserving people get.
Furthermore, there are several different ways it could be partly offset while still leaving a $1-2 trillion debt. There could be a 40% tax on incomes up to $40K, or instead you could try to soak the rich with a (say) 15% tax on all income. As long as everyone gets the basic money, Sam Stone’s post implies that the former is welfare and the latter is not, which is crazy talk.
This is true. I’d note that the tax rates in that experiment began at 50% – there’s room for more investigation with respect to magnitude. I’d also like to see the impact of increasing rates (as IIUC the experiment only looked at rates that decreased with income), since intuitively that would less discourage initial labor force participation.
Speaking on behalf of the DFH community, I don’t much like the “consumer economy”. But there it is, if we want change, we change what is, not what isn’t. Here’s the thing: if the consumer has no money, you have no economy. Zero, zilch, zip, nada damn thing.
That’s it, short and sour. Deal. Perhaps an investment portfolio that centers on for-profit prisons and gated communities?