Hacker News Reader

Comment by Chris2048 | Hacker News Reader

Chris2048

If someone puts heart and soul into something, I'd sooner they keep it for themselves. I really just want a good coffee, nothing more.

View on HN

The issue is, for now, a machine can make good coffee, but it cannot make great coffee. There is a lot of science behind the perfect cup that we still don't understand, thus making a machine understand it is impossible. It is still a craft.

If I go down to one of my local roasters, I know I get servered the near perfect cup. A machine would have to be calibrated multiple times a day to achieve the same level of consistency. And the barista would need to hone his/her skill to keep doing this.

You're most likely right but you seem also much more enthusiastic and passionate about coffee than most. Most people are fine to just get a good (not great) coffee in the morning on their way to work if only it goes fast enough and friction to use this system is low enough.

I certainly am. I'm a coffee nerd to the core. You're completely correct that most people just want a good cup, and they can already get that from McD - it is seriously a good cup. It's not great, but it's fast and the price is reasonable, albeit somewhat expensive for that it is (though, it is cheaper than starbucks and the like).

> There is a lot of science behind the perfect cup that we still don't understand

Are you sure?

> making a machine understand it is impossible. It is still a craft.

Why? What is a 'craft'?

> A machine would have to be calibrated multiple times a day to achieve the same level of consistency

Don't you mean a simple machine?

There's a lot of science suggesting the perfect cup isn't so objective.

Things like humidity, water chemistry, and others affect coffee. We don't understand this process in a scientific manner yet, and the art of calibrating rely on our tastebuds to fine tune every method from an espresso machine to a simple turkish coffee. It is still a long way from doable with a machine if you want that perfect cup.

That is not even touching the different parameters such as preinfusion, variable pressure, grind, and changes in temperature during the brewing. To my knowledge we do not know how to process coffee in a way such that a machine can model and understand the human taste buds. And it certainly haven't been applied successfully to coffee brewing yet.

All of these variables can be determined by a human and then programmed into a machine. We're talking about easily programmable things, not brain surgery.

Yes, we can. But every bean is different, even a different roasting profile means it needs to be calibrated and adjusted differently. It's not a simple matrix of input parameters and bean (yet).

It seems to me a machine would be better at keeping track of the difference. How many Baristas lookup a bean/roasting-profile/table to see what temp to extract at?

They don't have to. It's something you taste. Pull a shot, calibrate, rinse and repeat until it is as it should be.

We don't really need to have a machine which models human tastebuds. Coffee brewing is ultimately a chemical process, and so it's naive to suggest that a machine couldn't do it in a better and more repeatable manner than a human can.

What I think you're suggesting is that the settings of that chemical process are configured by the barista based on taste, and that they vary based on a number of environmental conditions.

Surely then all we need to do is to correlate environmental conditions with brewing settings. Assuming there is a relationship to be found, it sounds like the perfect application of either statistics or machine learning.

> Surely then all we need to do is to correlate environmental conditions with brewing settings. Assuming there is a relationship to be found, it sounds like the perfect application of either statistics or machine learning.

Big companies have been trying for many years, and after millions of dollars spend in R&D they still don't have it figured out yet. Imagine providing a full automatic machine that can produce a cup better that WBC contenders - it could cost thousands of dollars in service fee every year, and still be cheaper than your run of the mill barista. They really want to do this, but it is far from trivial. Either the blend and roasting profile produce a bland coffee that is less susceptible to variations in brewing parameters, or you need to fine tune it multiple times a day, each being different for every roast you have. Things such as when it was harvested, how much rain the plant got, how it was roasted, how old it is, ect. all influence it, along side things such as current external parameters such as humidity.

But there are coffee machines that produce amazing coffee automatically, they are expensive though. It's possible there is no market for it. It's also possible this is an emerging market.

They produce amazing coffee, once they have been dialed in, and will continue to do so, as long as they are continuesly calibrated. That is my entire point - we still need baristas to do the calibration. Maybe some day we don't, but until then we cannot say we have it figured out at all.

> We don't understand this process in a scientific manner yet

Again, are you sure? There are industry scientists that study this.

I'm sure. Either it's a well guarded secret, or they havn't been able to full map the relationships or it's infeasible to make a truly great cup purely with a machine (yet).

What about "Espresso Coffee: The Science of Quality" book?

If you can translate it directly to a computer that can adjust an espresso machine, then yes. But you cannot. It is an amazing book (I own it myself), it is a encyclopedia learn the craft of growing, roasting, and brewing some of the best coffee in the world.

You would be shocked by the amount of high end coffee bars moving to automatic and PRE dosing for their service. Even barista champions are speaking up the benefits of automated drinks in the favor of good, high quality service.

A machine will make a more consistent cup, day in and day out then any barista.

A good barista is a machine operator...

A machine calibrated multiple times a day by an experienced barista... I have never questioned the value of this. A machine is likely to produce a more consistent result, but it cannot calibrate itself.

Fully agreeing with your points here. The perspective I think that hasn't been mentioned is present here in Switzerland - where they make 'Jura' brand superautomatic coffee machines, and culture is such that in any establishment that is not specialty-coffee focused, you will be served your coffee by a robot. You will still pay top dollar for the middle-aged woman to walk over and press the button for Cafe or Cappuccino, but it will probably taste like burnt rubbish.

To me in particular, any hope of a quality standard of the milk aspect of the coffee is lost completely.

But returning to your points - The 'Jura' standard of coffee is 'good enough', and only not good enough recently to hip-swiss or expats living here. The standard of coffee expected from a barista is significantly higher - I imagine the theoretical Jura coffee machine that can do this would cost an insane amount to build, and to maintain and calibrate.

That is, until we get a real AI powered robot that can optimize and adjust its technique to what is happening in the cup and in the milk jug.

Caffeine grows a like towards the experience of the medium. It's possible some will prefer Jura? Point is, the 'perfect' coffee depends on brain chemistry, which is influenced by coffee...

Reply on news.ycombinator.com