Hacker News Reader

I think people without any real background in the difficulties of automating purely human tasks (particularly manual ones), consistently underestimate the cost, difficulty and expense of maintenance for said automation.

I used to work in a kitchen, so maintaining food service grade of cleanliness on robots, sounds frighteningly expensive to me.

On the other hand, you can walk into a Krispy Kreme and watch the machine work. There's no reason that machine couldn't make french fries with a few changes.

Maybe everything won't be automated all at once, but if the economic incentive is there, it will happen. Pizza making robots are already a thing. There's no reason the shake machine can't measure its output to the cup instead of pulling a lever; drink machines in many places already do this. Dispensing the cup too is not a big step.

McDonalds already has those.

French fry hoppers automatically filling baskets from a frozen mini-closet. Out comes fresh fries.

Same for the cups that they can use in drive-throughs.

I think it’s up to the franchises to buy/install them. Depends on volume/space whether its worthwhile/possible.

Or they just push it to the customer to do it.

About those automated drinks at McDonalds...when I order mine with 'no ice,' the machine is apparently incapable of filling the cup with the additional amount of coke that would normally be displaced by ice. It will simply fill the cup just 2/3 of the way. They seem to have realized this though, but instead of increasing the volume of coke dispensed, the machine drops ANOTHER cup which the employee is supposed to use to fill the actual cup to the top. The employees only do this half the time, however, which makes the entire process even more maddening. Somehow neither man, nor machine, can get a no-ice drink right at McDonalds.

I hope one day they integrate the ice together with the drink and scoop out a single solid cube of ice to keep it from watering down right away.

That is because those machines are made in a way to mix the 'perfect' ratio of water + syrup. They take in account the ice as well. If the cup is 0.5l, it plans to be 0.4l of liquid + 0.1l of ice (random values), so if you chose no ice, you get 0.4l since that is the 'perfect' ratio. (And probably what they advertise ie '0.4l coke = 1 euro' or whatever, they just give a 0.5l cup to account for ice)

> There's no reason that machine couldn't make french fries with a few changes

That machine is, like, 5 machines including an oil moat. With all of them together it makes up something like half the entire store's footprint. It's only remotely viable because that shop mostly only makes one thing. No breakfast sandwiches in the morning, no burgers at lunch time, just donuts.

Plenty of employees are still required just to make the different types of donuts and donut-adjacent items (and also presumably to be there in case a monumental grease fire starts, or the dough machine blows a gasket starts spewing dough all over).

> Pizza making robots are already a thing

Presumably making the frozen ones sold at grocery stores, which might be a meaningful percentage of all pizza sold, but effectively 0% of any other pizzas?

As you note, Krispy Kreme is a small scale donut factory with a store front ;-)

This is an entirely different thing than a McDonalds.

Yes, I could trivially make an automation machine to cook meat, we have them, they're called a charbroiler, similarly, toasting buns is pretty easy. The hard an expensive part is assembling all the parts into a package.

I would argue it's substantially easier to produce a fully-automatic McDonalds than it is to produce an autonomous car that can actually deal with any reasonable situation in the real world.

With a McDonalds restaurant, you can control your environment almost perfectly. If you throw away the kitchen that was designed to be used by humans, and then develop a kitchen never intended to be used by humans, you can drop a lot of environmental constraints in terms of the automation concerns. Trying to fully-automate the existing kitchen layout will certainly end in disaster.

As you reach 100% automation, it actually starts to get easier because the system is now all oriented in this direction, and you have standardized specifications/interfaces across the board. Also, once you fully-automate a single McD restaurant, the next one becomes trivial because it's simply another copy of the exact same thing.

100% agreed. A friend of mine is doing a food automation startup and in 6 months they’ve basically got it working. They’re planning on six more months of nailing it all down and launching. Based on what they’ve done I think they could do a fully automated McDonald’s in about two years. I would imagine it would then take another 5 years for such a system to spread to nearly all fast food restaurants... which implies to me within 10 years it’s pretty likely we won’t see humans in the kitchen at quick service restaurants..

And I think it’s more than that before we have no humans behind the wheel of cars!

So then what happens when someone makes a mess of the dining area or the bathrooms? I've seen roombas encounter dog excrement—not pretty. Or if a fight breaks out among customers? Most Mcdonalds operate pretty lean, sometimes two people. I can't imagine a complex automation setup and hiring a pricey/hr guy to service it every week is going to be cheaper than two guys on minimum wage and no benefits.

Perhaps you simply disallow that customers be able to enter the actual restaurant at all, and operate it more like a gigantic drive-thru/walk-up vending machine.

> Ordering kiosks are the low-hanging fruit, and then comes a long, long stretch of nothing,

DOM Pizza Checker (Domino's camera 'AI' that checks the pizza toppings are correct) could be used for other things, other than it's 'other' current use of store tracking. https://www.itnews.com.au/news/dominos-turns-its-pizza-check... Stores now get updates ;)

Designing food so it's optimised for cooking has been going on for a while but I suspect there gains there. Even if you can design more food variety for fast food that can be cooked or created efficiency, it can cut out labour at other food outlets as they close and society moves to highly efficient food outlets like McDonald's.

Automation in processes is where it's at. Designing food and processes for mostly automated machines and schedulers.

But you are correct, so far the threat of automation to keep wages low seems to have had more impact than automation.

The cooking machines are nearly automatic already. They are all designed to accept input and produce output without human action in between. I want to create an automated pizza shop, and have made some mock ups. I don't think it's as hard as you mention.

But you concede that some jobs would be automated if the labor prices were higher, and yet you think that's fine? Do you acknowledge a trade-off whereby fewer workers can get higher wages OR more workers can get some wages and prefer the former?

For argument's sake, let's assume that no jobs can be automated. The cost to the franchise owner to operate goes up, and in response they raise prices -- but at some prices they're going to really start losing business, so they also have to lower costs elsewhere or simply get lower profits. How does it all shake out? The fact is that it has to be paid by someone, and it'll likely mean fewer McDonalds are opened because it'll just be harder to make them profitable.

Maybe you're OK with that trade-off, but it IS a tradeoff that will happen, and it means fewer jobs.

Your analysis presumes that the business owners can and will absorb anything without reacting, as if more value can be extracted but there won't be a new equilibrium. I assume that the franchise owners aren't making the kinds of profits where that is true, I assume they're mostly bumping along on fairly low margins.

This is an entirely different question that has nothing to do with the original article, and nothing with the thread starter's comment.

But for the sake of it: yes, in a time of practically full employment, I do indeed prefer less workers to make more if this trade-off you assume is even a realistic model of reality (which is a big if, in my opinion).