Hacker News Reader

Comment by Iv | Hacker News Reader

For more than 20 years I have advocated and pursued the idea that society should abolish work. I am astonished that in 2020 most of our production is not automated.

We have had the ability to automate most work for two decades. Food, clothes, furniture, electronic devices, all of these could be produced with close to zero human labor. Why don't we?

After exploring a market (agriculture) in order to launch my own automation company, I have come to the conclusion that it is not the tech that we lack, it is the will. A culture change is needed in order to make people realize that a work-less society is:

- possible

- desirable

- is not going to make them poorer

View on HN

Many things we consider "automatic" require complex maintenance and upkeep, and inappropriate automation can actually increase the net amount of work needed to be done.

It often coincides with a loss of ownership. If you can't make your own value, and you're letting something else make it for you, you might save yourself labor, but you're giving up your ability to create that value. Yes, the system as a whole will produce more, which is good, but the loss of leverage is an important factor to consider.

There's also a wasting and dependency effect that occurs when too much of a system is automated. If people aren't needing to work on or maintain a system, they don't need to know how it works to use it, pretty much by definition. It's doing the work for them. That creates a dangerous situation where essential systems aren't really understood, and fewer and fewer people end up knowing how to fix things because there isn't the same need to distribute the knowledge of upkeep/understand the work it's doing personally by doing it yourself.

Automation is extremely beneficial, and I'm often frustrated by what seem to be clear cases of not taking advantage of it, but I think what you're saying here over simplifies things.

I think the solution is for more people to learn how to set up their own automation and to automate things without making them too centralized.

I am trying to not delve into a 20 pages presentation, but I have gone though and through these themes for most of my professional life. The question whether we can really go to 100% of automation is moot if we can go to 99%. Either way it means that full employment is unnecessary and leads to the creation of bullshit jobs.

> I think the solution is for more people to learn how to set up their own automation and to automate things without making them too centralized.

That's my sad conclusion as well. We could get to an automated society with far less pain and much faster if it was decided collectively though.

Look at car automation: if a city wanted to make automated cars a reality in their streets, there are tons of accommodations they could do: from radio beacons to official maps, standards on how to signal construction work, purposefully designed roads...

Instead, we are trying to design automated cars with the assumption that zero efforts will be made to promote them. Worse: we assume they are going to be so criticized that they have to perform better by a magnitude on day 1. That's making us waste 40 years.

> if a city wanted to make automated cars a reality in their streets, there are tons of accommodations they could do: from radio beacons to official maps, standards on how to signal construction work, purposefully designed roads...

Sorry, but that sounds hilarious. If "a city wanted", it's still people who would need to ensure "to signal construction work". And people don't care. And for other stuff, people would need to pay for it with their taxes. I'm sorry, but as an outsider, I would say the roads (usual roads!) in the US are in "perfect" condition only in California. In other states, it's the usual asphalt-with-cracks, which will turn into a hole when a heavy truck rides it thru the rainy/snow season.

Heck, majority of the world has problems with trash on the streets, and cities can't neither teach their people to not litter, nor clean up timely after them.

> And for other stuff, people would need to pay for it with their taxes.

Automation benefits cities as well, you know. For instance computers probably drastically reduces the number of manual processing of paperwork. That's tax money you can use for something else.

> That's my sad conclusion as well. We could get to an automated society with far less pain and much faster if it was decided collectively though.

I disagree. I think we have decided collectively to progress towards automation as fast as possible without unduly impacting people's quality of life.

> Look at car automation: if a city wanted to make automated cars a reality in their streets, there are tons of accommodations they could do: from radio beacons to official maps, standards on how to signal construction work, purposefully designed roads...

I believe there already are official maps and standardized signs in the developed world. I agree that incremental improvement is possible and desirable, I also think people are working on these things already. It is possible that signs could be redesigned to make them easier for machines to read but I'm not sure that's much of a bottleneck.

> Instead, we are trying to design automated cars with the assumption that zero efforts will be made to promote them.

I see lots of effort to promote them, they just aren't technologically ready to perform at scale yet.

> I disagree. I think we have decided collectively to progress towards automation as fast as possible without unduly impacting people's quality of life.

We have decided to move all our factories to China instead of automating them. We still have subway drivers despite having the tech to automate subway since the 1960s.

> Many things we consider "automatic" require complex maintenance and upkeep, and inappropriate automation can actually increase the net amount of work needed to be done.

As seen in some test suites and CI/CD pipelines...

> Food, clothes, furniture, electronic devices, all of these could be produced with close to zero human labor. Why don't we?

For a large subsets of those categories, they already are. That's why, if you mostly buy food that's handled by automation (combines etc.) you can sustain yourself on a $50/month food budget. Similarly for furniture, check out IKEA factory videos to see the degree of automation employed in making cheap furniture. Electronics are mostly also produced by machines, humans do the last stage assembly only (and it's mostly because labor in Asia is just cheaper than sophisticated robots required to perform assembly).

Clothes are more difficult from a robotics standpoint (mostly because, unlike wood, cloth is not rigid/does not retain shape, which makes it insanely tricky to manipulate), so we're not there yet. But, on the other hand, making of the cloth itself, which previously required an insane amount of labor, has been fully automated for a long time.

Basically, once you own a place to live in, you can easily sustain yourself with a very part-time minimum wage job. Make the job pay more and you'd need to work maybe a month in year. Most people don't do that because they want the comforts and pleasures brought by market enough to work extra hours for them (usually up to a full-time job).

Sustain yourself just means not die in your text. But there's more to it than that.

Most people won't do the $50/month food unless forced. The Soylent-junkies maybe, so if you're some robotic Western software dev whose sole purpose in life is placed in its optimization to be "productive" in a corporatist society.

Food is a central cultural part of billions of people's lives, provides joy and is a critical component in socializing. Food is more than just plain sustenance unless forced by natural or artificial circumstances.

Fair enough, but even including all those extras that stil puts the cost at maybe $150/month? That assumes cooking for oneself, not buying a lot of meat or other expensive products etc. On that budget, you will probably not eat worse that a median human.

That's probably true, yes.

> We have had the ability to automate most work for two decades. Food, clothes, furniture, electronic devices, all of these could be produced with close to zero human labor. Why don't we?

Because we actually can’t do those things?

We do a lot of things that we couldn't before. A lot of it produced loud laughter 100 years before.

If I understand it properly the author suggests we should find the time to play with these things. Who knows what might happen? It will no doubt be interesting.

Our indevidual performance is certainly mind blowing in a play settings. I see "players" do what seems impossible to me every day.

The fact of the matter is that automation is far from perfect. There are many tasks that are simple for a human, but fiendishly hard for a machine due to their unpredictable nature. Take driving a car or cooking in a restaurant. How is a machine supposed to know if the meat is a bit too old to be palatable? Would it really be worth it to implement an elaborate apparatus to determine meat freshness, and deploy it in every restaurant? I believe such a thing is firmly in the realm is science fiction, where effort and energy are free, unlike the real physical world we inhabit.

Now consider this, can joy not be had in cooking? I can certainly tell you that I find cooking good food to be a very fulfilling task. Unlike engineering projects I can go from start to finish in under an hour. And it fulfills an immediate and visceral need.

I think that the argument for play is a good one. In engineering I encounter many people who engineer, not for the joy of it (which is spread very thin in many jobs anyway), but for the money. Now imagine if there was no quantitative social status (money) associated with engineering. You would see engineers self select purely on a basis of authentic interest rather than financial status seeking. Would that not bring more promising talent to the table?

Let me now attack the idea of automation from another angle; to claim that we should be able to automate all product is as outrageous as claiming that Atlas holds the earth up. How could we automate everything? Who makes the machine, and perhaps more importantly, who fixes them when they break. You cannot possibly convince me that you could make a machine to fix machines. Again, such a device would be in the realm of science fiction. In reality, even in the relatively controlled environments of factories things inevitably break in unpredictable ways, and there you are, back to needing humans to clean up the mess.

Now, automation does have a place; doing repetitive tasks. But I’d rather order a meal from an expert chef than a robot.

First, the disagreements: no, driving and cooking are not hard tasks to automate. They are stupidly easy if you decide to create the infrastructure for that. You have industrially cooked food, mostly through automated process, in supermarkets, sold as "prepared food". and actually, many cheap restaurants will have several microwaved prepared food in their menu items.

Automatic driving is a very easy task in roads and cities that are designed for it. Instrument all bus lines with ground wires, put a radio beacon on each crossing and sign, a front radar/lidar to detect obstacles and 90s tech is enough to automate a whole city. If it was desired, we would already be there.

Now, like you point out, there are many tasks that bring joy and people would do voluntarily. Said otherwise, there is a non-null amount of volunteer productive work.

Ergo, the goal of automating a society is not to automate 100% of the work, but to automate (100-V)% where V is the amount of voluntary work available.

"But I’d rather order a meal from an expert chef than a robot."

It would be interesting to make a Turing test for cooking. I am sure that there are many robot-made meals that are undistinguishable from chef's preparation.

I’m not interested in or know much about self driving cars so I’ll leave it. I think your points about cooking are really off base though. It seems like you’re basically describing mass production which we’ve already had for a long time, a wasteful process that produces inferior food.

As for a turing test, we’re talking about a professional chef here. Someone you can take an enormous variety of ingredients to in various forms of freshness or preservation, who can produce something that tastes really good. Theres often a large amount of creativity, improvisation as well as fine motor skills and expert timing involved in the process. It would be a monumental, hugely expensive task to create a machine that could even produce something edible given the same constraints. This just seems so far fetched to me.

The fine motor skill and expert timing are a given in automation, so let’s start by assuming the chef is not doing the physical work. And if we are trying to recreate the experience of eating at this chefs restaurant, then we aren’t taking them any food to improvise with, we will just eat one or two of the dishes they have already designed - so you may be picturing something more complicated than what would actually do the job.

> The fine motor skill and expert timing are a given in automation

I'm not convinced. For example, robots are absolutely terrible at handling cloth tasks (for example, folding laundry). I suspect that handling food, which has complex and unique (depending on ripeness etc.) physical responses to manipulation isn't much easier. Just try to give some in theory simple slicing tasks (such as - slicing an onion or a chicken fillet) to a 7 year old and see how well he does. The robot is no better.

I'm not talking about the robot you bought off Amazon. Robots do surgery and manufacturing these days, they are a little more capable than a 7 year old.

So if a machine does the physical work and if the chef has designed the dishes and likely needs to adjust them to account for differences in ingredients as well as tasting during the process to get feedback, is this actually fully automated?

>Ergo, the goal of automating a society is not to automate 100% of the work, but to automate (100-V)% where V is the amount of voluntary work available.

Isn't that the economic reality rather than "the goal"? And doesn't this also put an shade on your original argument that it is "the will. A culture change is needed"?

I haven't worked for 20 years but I work in certain automation, and from what I've deduced it is rather that the V in (100-V)% is quite high, and then there's W = a x V (a is an positive constant) that specify required labour, specialization & investment for automation; which is almost always lower in supply than "voluntary work available".

> They are stupidly easy if you decide to create the infrastructure for that. You have industrially cooked food, mostly through automated process, in supermarkets, sold as "prepared food". and actually, many cheap restaurants will have several microwaved prepared food in their menu items.

Everyone agrees these products are inferior in quality to the comparable manually produced offerings.

> Automatic driving is a very easy task in roads and cities that are designed for it. Instrument all bus lines with ground wires, put a radio beacon on each crossing and sign, a front radar/lidar to detect obstacles and 90s tech is enough to automate a whole city. If it was desired, we would already be there.

It's not that easy to redesign an entire city to accommodate a second transportation network. A lot of buildings would have to be modified. Its probably better to approach this task incrementally using the existing road system.

> It would be interesting to make a Turing test for cooking. I am sure that there are many robot-made meals that are undistinguishable from chef's preparation.

I'd like to see a cooking Turing test as well because I disagree with you and this is in fact testable, one of us is mistaken.

In rich societies with strong social safety nets, it is possible to live as if work had been abolished. One only needs a fraction of the ressources consumed by a normal Westerner to live a materially comfortable life. But what you realize when you try reaping the benefits of our already largely automated economy, is that work is about ideology. You'll be materially comfortable, but a social outcast. People don't like to work, but they hate non-working people even more.

> People don't like to work, but they hate non-working people even more.

Probably because they wonder how those people are figuring out how to eat without contributing to the economy.

I agree with you in the sense that's how they rationalize their reaction, although I think they don't actually wonder about it and you made a figure of speech, and instead they don't feel it's fair that some stop working until everything is completely automated.

But I see an inconsistency in their feeling of injustice, because some people have to work a lot to get comparatively very little benefits from our largely automated economy. Anyone who works should feel like they live a luxurious life. Instead we see a lot of addiction and despair, despite lifestyles that consume huge amounts of energy.

Returns on investment in automation increase with the scale of the system being automated. Therefore, automation promotes centralization of control.

Be careful what you wish for.

The agriculture market already spends billions on automation. For example row crop tractors are mostly self driving. Machines milk cattle with no humans involved. Etc. If you failed in that market then it's not due to a lack of will by customers. It's because you're not a competent businessperson or engineer and failed to achieve product/market fit.

Is your automation company still active?

Reply on news.ycombinator.com