Hacker News new | past | comments | ask | show | jobs | submit login
Poka-yoke (wikipedia.org)
160 points by vitoc on Sept 24, 2018 | hide | past | favorite | 60 comments

This seems like an interesting and useful process design concept on one hand, and on the other hand, it also sounds like a Japanese buzzword for foolproofing (in fact, the article notes that the term was originally “baka-yoke”, which is literally “foolproof” or “idiot-proof”). On the one hand, I don’t think Toyota invented the concept of foolproofing; on the other hand, giving it a cool buzzword gives you an excuse to talk about it, which means you’re more likely to actually do it when appropriate.

I’m also kind of amused at how the entire business world basically said, “man, this company made super reliable cars in the 80’s and 90’s[1], let’s learn their secrets” and it turns out that once you strip away the exotic Japanese veneer, it’s mostly just systemized common sense. Which is not a minor thing to do, to be fair, except it’s easy for businesses to take good, well-systemized ideas and utterly misapply them.

[1] As far as I know, they still make reliable cars. But you can still drive a 92 Camry today if you wanted to with little fuss. By definition, we’re 26 years away from seeing if today’s Camry is quite that reliable.

The way in which the term is used in the manufacturing industry is a different from "foolproofing" in practice. Generally, a Poka-Yoke is a step added to a process that requires an operator to stop and think about what is going on and whether everything is within spec before proceeding with the next step.

One example I can think of would be a torque tool used in a critical step of the assembly process. Since modern factories allow different models of roughly similar product types to be assembled on the same production line, it is important that the torque tool is run using the correct settings for each model. In this example, the operator has to scan a barcode on the vehicle and the corresponding barcode on the computerized torque tool before the tool will work. This not only sets the tool to the correct setting, it forces the operator to stop and think for a second when the tool doesn't work. By doing this as the first step, it also ingrains the knowledge of which model number the operator is working on as they consciously have to scan the right barcode on the tool as there's a unique barcode for each variant they'll see coming down the line. That way they are more likely to catch any incorrect/mismatched parts tied to later stages of assembly carried out at their station (i.e. they know they are working on Model X Sport but the parts bin shadowing this chassis has a cosmetic piece that is only for the Model X Limited).

Also, I;ve always found it funny that the phrase sounds like poke-a-yoke (yoke, as in yokel) given how it's used in the industry to describe a way of prodding someone to pay attention.

A NYC-via-Japan subway practice is another nice example:


which led to this delightful friendly prank: https://www.youtube.com/watch?v=i9jIsxQNz0M

More examples of the narrow and broader meaning:


But these aren't great examples because they don't capture the important aspect of an automated system detecting and reacting to an operator violating the process protocol.

People will generally operate with common sense until incentives or pressure is applied.

If you are under extreme pressure to produce a certain count of parts, common sense is suspended and people do things like ship problems on down the line. (See also local optimization)

You have to account for fatigue also. While a repeated process can drop into the 'automatic' mode where you're unfocused but still running through the task, critical validations can be omitted when focus is not retained. These techniques guard effectively against that, as there are clear, focus-inducing markers when something is wrong in the process.

This is the main thing you're guarding against. It's not complacency, or additional pressures... It's just plain old boring fatigue which on a long enough time span will hit every operator.

It's hard not to be dismissive of what appears to be a simple set of concepts, but you're missing on an opportunity to learn a very cool set of skills here. I recommend that anyone who wants to live a better life and do better work read books on the Toyota Production System and Taiichi Ohno. As for Poka-Yoke, It's a word that seems simple but when used in industry contexts communicates a deeper set of requirements than just "make it foolproof". Also, don't undersell how useful systematizing common sense is (and how difficult its application can get).

I don’t mean to undersell it. If anything, I’ve been growing increasingly impressed at how many of life’s secrets also turn out to be things that everyone knows.

Edit: Plus, Wikipedia does a very inconsistent job of explaining conceptual things in non-superficial ways; probably because most extended discussions of “poka-yoke” wouldn’t be “neutral” or “verifiable” or would “favor one particular source”. I haven’t checked it lately, but Wikipedia’s article on method acting used to do almost nothing to explain what the hell it was; it just had a lot of verbiage about who does it and who learned it from whom. Maybe there needs to be a separate concept explanation wiki that has different requirements.

Any specific books you would recommend?

Doing a quick search on Amazon I found: Toyota Production System: An Integrated Approach to Just-In-Time https://www.amazon.com/dp/143982097X/ref=sr_ob_1?ie=UTF8&qid...

Taiichi Ohnos Workplace Management: Special 100th Birthday Edition https://www.amazon.com/Taiichi-Ohnos-Workplace-Management-Bi...

The Toyota Mindset: The Ten Commandments of Taiichi Ohno https://www.amazon.com/Toyota-Mindset-Commandments-Taiichi-O...

There seem to be handful directly from Taiichi Ohno (including a management book), and then many that refer to him and his process. Guidance would be appreciated!

"The Toyota Way" http://a.co/d/eso4HOR might be the best example of a very easy to read book that explores the Toyota system in a practical ways.

It's full of a lot of anecdotes that really show how Toyota thinks in practice and isn't just a textbook you have to apply yourself.

There's also "The Machine That Changed the World":


Generally, anything auto manufacturing related works too.


"The Principles of Product Development Flow: Second Generation Lean Product Development"


We see a lot of words written on "lean" software development but I feel like it's better to learn from broader sources, and from industries that have had a lot more regulation and time to develop good practices.

The best book on Lean for beginners is “2 Second Lean” by Paul Akers. You can buy the book or get it free directly from Paul Akers’ website. https://paulakers.net/books/2-second-lean

Mr. Akers also has several hundred videos about Lean on his YouTube channel. https://paulakers.net/lean-videos

And for an impressive display of what Lean can do, watch this video “Lean Manufacturing - Kaizen Methodology - Lean FastCap Style” https://www.youtube.com/watch?v=su9CulCZTBg

"Common sense" isn't common. Not only that, it's ridiculously unpopular to try to change people's ideas of what common sense is. You can see this even on HN when people suggest that programming language tooling should constrain defects.

It's interesting how much of "Japanese" manufacturing practice is really the work of W Edwards Deming, the prophet without honor in his home country.

Part of that was probably that American manufacturing businesses post-WWII were a lot more complacent, which they had every right to be because they had just out-manufactured the entire rest of the world and won the war. US war manufacturing even started declining in quantity in 1944 just because we had more than enough of everything by then.

Meanwhile, the Japanese, who already know that industrialization is the key to development (which is why after Meiji they started an empire in the first place, to gather the necessary resources) were all too ready to learn.

From my understanding, Deming’s emphasis was on quality, and you still see this as the emphasis for Japanese manufacturing. Perhaps the Japanese were more interested in focusing on quality because they knew they would never match the quantity of a continental power, but quality (albeit not to the standards Japan would later set) was also a distinguishing factor for American war manufacturing. We didn’t take the Soviet approach of building as many tanks as physically possible even when it meant cutting corners, largely because we didn’t have to (and the Soviets surely didn’t complain about their massive numbers of cheap tanks).

Industrial Engineer here,

This is not a buzz word, this is the name of the tool we use. 4/4 companies Ive been at used Poke-Yoke, Poka-Yoke, or Error Proofing interchangeably.

And while I dont own a Japanese car, I can verify the 'common sense' ideas had a resurgence as Toyota applied a process to these principles. It seems nearly every company MUST apply them or they are uncompetitive.

From a plant with 50 employees to a fortune 500 plant, I have not seen these 'misapplied'. Only excellence.

I find outsiders are pessimistic and are unaware that these plants are using SPC and have process engineers looking to optimize daily.

I'm also an engineer building industrial automation systems. And I've sold dozens with poka yokes, done kaizen reviews (improvements), and used scanners/plant-machine communication for kanban (JIT/lean manufacturing).

I think the important thing to note here is not that when you use the Japanese word, it becomes a process, a tool, a requirement you can mandate and sell - not just common sense.

If you just call it idiot-proofing, people too often reject additional cost, process time, and complexity. If that sensor and part test are there because those are the poka yokes so you can't load it backwards, people don't push back as much.

I'm not thinking about actual manufacturers when I think about misapplying cool-sounding Japanese manufacturing concepts. I'm more thinking about other businesses where they basically cargo-cult manufacturing ideas without actually thinking about whether they work (which is ironic because continually thinking about whether your processes work is the key part of the process.)

I'm mostly struck at how error-proofing is a seemingly obvious idea, and yet it's still worth writing and talking about. That makes me think most of the value is in the execution and not in anything that can be written down.

I used to work for a Toyota supplier years ago, and got to visit one of their ports where cars and parts enter the US from Japan for further assembly (like a mix of car garage and factory). It was striking how organized and neat everything is.

It all boiled down to "Kaizen", another simple concept meaning "continuous improvement." From what I remember, one implementation of kaizen was that employees were expected to stop what they were doing two or three times a day and spend 10 minutes organizing their workspace. This seeming waste of 20-30 minutes a day actually increased efficiency by having all tools in the right places at all times.

Working with them was a real eye-opening experience - it showed me how simple concepts can have a huge impact on a business when applying them consistently and with (sometimes maddening) discipline.

As I was reading this Wikipedia entry, I came upon the part where it said it was developed by Toyota and actually uttered out loud "of course it's Toyota!" I have nothing but respect for their work ethic and discipline.

I worked for a similar company previously as well— a Toyota subsidiary.

While I largely agree with you, in North America I can't help but think that some of the principles are sometimes mis-applied.

Related to the principle of Kaizen, I was witness to continuous improvement being used as justification to continually demand more of employees to the point where practices like you've mentioned were actually discouraged. If your machine stopped for that long, management and maintenance would be flagged down automatically. 20 to 30 minutes of downtime in a day would have been flat-out unacceptable no matter the reason.

The cause being that if your workflow increased in efficiency and accuracy to the point where manufacturing goals were being matched and or beat, the goals would increase without investigation. after a few iterations of that, operators would find themselves always below the threshold of required output.

This was often used as justification to let employees go to fill the positions with temp staffers. Keep in mind this was over a decade ago and was local NA management's initiatives.

So I guess like anything it has to be applied properly to see the proclaimed benefits.

> giving it a cool buzzword

They called it the obvious word in their own language, it's only a cool buzzword to non-native speakers reading English Wikipedia.

> systemized common sense

Of course. And still businesses are full of systemic problems, where common sense fails. Just like well written programs are just "systemized instructions", but most programs are buggy and brittle. Clarity in defining and executing on simple concepts is worth a lot.

> They called it the obvious word in their own language, it's only a cool buzzword to non-native speakers reading English Wikipedia.

I'm more making fun of the American business consultants who are going to try and cargo-cult this idea and make it sound more impressive than it actually is by not using the already commonplace American English idiom that describes the same concept.

The term "poka-yoke" seems to date from the 1960's, while "foolproof" is well-attested before WWII, which makes me recklessly speculate that the term is likely a translation from the English in the first place. Since Japan sought out and received a lot of guidance and advice from Americans about rebuilding their manufacturing industry (which is logical, given the fact that we more or less won the war by out-manufacturing them), I wouldn't be surprised if someone casually used the word "foolproof", and merely translating the idiom introduced the underlying concept. (Kind of like how we stole "earworm" and "punchable face" from German idioms). And hey, maybe the extra step of introducing a new foreign idiom, or even reintroducing an idiom that we already have, is valuable because it puts a bright flashing light on it. Buzzwords only become buzzwords because there's a substance behind them, it's just when the bright flashing light outshines the substance that it becomes a problem.

> Clarity in defining and executing on simple concepts is worth a lot.

See, that's the thing, it's execution. You can talk about foolproofing your processes and your tools, and half the time you end up getting in people's way as often as you prevent any actual errors, because no matter how well you foolproof something, there's always some fool out there who's even smarter than you are when it comes to discovering new and creative ways to screw things up. But the other half of the time, not doing it is going to just completely burn you. The judgment of differentiating those two cases from each other is the valuable part, not just the concept of error-proofing (or poka-yoke) in the first place.

In a large part, it's because Western business practices suffer from a lot of ego. I suspect consulting has a lot of cargo-culting because you need to wrap things up in a foreign veneer in order to implement certain optimizations without outright calling someone an idiot.

Maybe that is a good reason why "foolproofing" is not a great way to describe these process improvements. It is pretty condescending to label somebody as a fool for anynsingle erroneous action!

I wonder if that's why American manufacturers didn't listen to Deming in the first place, until after he taught Japanese manufacturers how to kick their asses.

>They called it the obvious word in their own language, it's only a cool buzzword to non-native speakers reading English Wikipedia.

Same place "Kanban" came from

It is surprising how many problems can be solved with common sense. Whta Toyota did was codifying it in a rather elegant way. They gained a considerable advantage by doing so.

Also surprising how hard the implementation of common sense can be. Also true that only a few companies really manage to implement lean / TPS methods correctly in the way Toyota did.

I don't really understand the point of your message, it feels like forced cynicism around the cultural reception of the term.

Perhaps you're saying that the phrase is akin to "big data" or "AI" in that it doesn't really say much. The idea that this concept can be misapplied doesn't serve well to steelman its useful core concept and philosophy which is timeless. My favorite example are electrical plugs [1]. Whatever beef you have with Toyota, Japanese, or buzzwords just seems inappropriate here and lowers the quality of discussion.

[1] https://image.slidesharecdn.com/pokayoke-150112043214-conver...

> Whatever beef you have with Toyota, Japanese, or buzzwords just seems inappropriate here and lowers the quality of discussion.

Hey, I'm sorry it came across that way, because I don't have a beef with anyone. But I've explained and expanded on what I've said in this thread a lot before you came along with that comment, and I think it's uncalled for (and "lowers the quality of discussion") to assume bad faith the way that you're doing right now.

It is more than a buzzword. It is a fundamental tool in process design. And there is a lot of ingenuity in creating good poka-yoke solutions.

> As far as I know, they still make reliable cars. But you can still drive a 92 Camry today if you wanted to with little fuss. By definition, we’re 26 years away from seeing if today’s Camry is quite that reliable.

That has little to do with preventing manufacturing defects and more to do with preventing design defects. A defect-free car can still break down prematurely because it doesn't wear well, used incorrect materials, ect.

>I’m also kind of amused at how the entire business world basically said, “man, this company made super reliable cars in the 80’s and 90’s[1], let’s learn their secrets” and it turns out that once you strip away the exotic Japanese veneer, it’s mostly just systemized common sense. Which is not a minor thing to do, to be fair, except it’s easy for businesses to take good, well-systemized ideas and utterly misapply them.

This was basically all done over UAW's dead body. It's not like process engineers working at the big three in the 80s didn't want to do those things. They were aware of what industry best practices were as any competent professionals would be. They just couldn't get those best practices implemented until the Japanese manufacturers created an existential threat to the status quo.

Cyber security is in a similar situation today. The will to do the known right things is not there and won't be until there's an existential threat to the companies that are being lazy. Whether that threat will be regulatory, market based or something else but I guarantee little will change until it becomes a binary choice between change and going out of business.

> As far as I know, they still make reliable cars.

There's a pretty substantial lag between reality and market perception. All the people paying lightly used 3/4 ton truck and fullsize SUV prices for Tacomas and 4Runners with 100k+ on them had better hope that what they're buying today lives up to the reputation of vehicles sold 10-20yr ago.

>we’re 26 years away from seeing if today’s Camry is quite that reliable.

Unless they make a complete turd it will be more reliable because they hold their value well giving them a longer period of time when they are out of reach of customers who are hard on equipment.

Vehicles move down the economic ladder as they change hands. Rich people have more free cash flow to spend maintaining their car (especially preventative maintenance). This skews age based reliability data in favor of whatever vehicles wealthier people drive.

Economic factors make differences in vehicle reliability from brand to brand look much greater than they are.

If you want to see this in action look for 80s GM sedans on Craigslist in a state that salts the roads. The Buicks, Olds, and Caddys are over-represented because their first and second owners were more likely park them in a garage (or at least off the street), pay for preventative maintenance, etc. allowing them to stick around longer because they were (on average) in better condition than the equivalent average Chevy of the same age, had more useful life left, etc. The upscale cars had to get to their 2nd and 3rd owners before they were treated the way the Chevys were treated by their first and second owners giving them the appearance of greater reliability despite being the exact same platforms. the chevy might have 3-5 owners before going to the junkyard whereas the Caddy may have 4-7 (and therefore stick around for more years) because it basically starts off several rungs higher on the economic ladder.

>This was basically all done over UAW's dead body. It's not like process engineers working at the big three in the 80s didn't want to do those things. They were aware of what industry best practices were as any competent professionals would be. They just couldn't get those best practices implemented until the Japanese manufacturers created an existential threat to the status quo.

Have you come across this podcast / radio programme episode? https://www.thisamericanlife.org/403/nummi

It's about a joint venture between Toyota and GM in the 1980s. It discusses the culture shock the US workers experienced and the labour relations at the plant. The people interviewed give their own views on why GM didn't / couldn't adopt the lessons learnt in their own plants.

tl;dl The previous GM plant on the site had had terrible problems with the workers there. The new venture hired the same people (starting with the shop stewards) and brought in a totally different way of doing things which totally changed the culture. GM struggled to adopt the Toyota way of doing things in it's own plants because the entire manufacturing chain and culture across the industry had to change to make it possible. I really recommend listening to the whole thing.

After NUMMI closed the Fremont site was taken over by another US manufacturer that sought to improve on existing methods with a greater degree of automation.

Why were the UAW guys so resistant?

I drive a 95 Pontiac Grand Prix with little fuss. Treated right, cars are extremely reliable.

Survivorship bias.

Toyota makes extremely reliable cars, it doesn't just come down to maintenance. Poka-Yoke has a lot to do with that reliability.

One of my favorite examples of poka-yoke is the Apple MagSafe charger. The magnetic part itself is foolproof, sure. But the genius part is the small metal knob that meshes with the removable AC prong piece. It is impossible to insert that piece in any other way than it was intended, and the solution is unobtrusive, clean, and simple.

I still don’t get why they removed it. MagSafe was a great thing. Sure some people got things stuck in the connector but that is a much smaller problem than having your laptop pulled along with the cable when someone trips on your charger.

Anecdotally, I just accidentally kicked my USB-C charger loose, from the charger side. Pretty sure that would have stressed the old MagSafe cables.

I expect to have to replace the cable at some point, but not the charging brick or the ports. For those with large dogs, toddlers, or just klutziness (I'm borderline), magnetic breakaway cables are available.

The metal knob actually is PE pin for the AC plug, which is used in some parts of the world with presumably stricter low voltage safety standards which would not allow for the chassis of macbook to float at what is essentially half of the mains voltage (this is why you get slight tingling sensation if you run back of your hand over macbook that is plugged into charger with two-prong plug).

Just want to share some stories:

One of my professors (I am an Industrial engineer) used the example of a SIM card. I imagine people can still get it wrong when putting it in, but much less so (ehem USB) because of that slight cut in the rectangle. It also helps a lot when manufacturing those things to get the orientation right.

Another mistake proofing anecdote I've heard was about one of his students interning at a company that packages soap. Somehow there were empty boxes going through the conveyors after some process change. Some of the mechanical engineers were already thinking about some modification to the conveyors that detect weight, etc. The IE student apparently simply went in one of the back offices and got a spare electric fan and pointed it at the conveyor so the empty boxes fly out. Troubleshooting doesn't have to be expensive.

That’s a story which I have seen a few variants of before this; I doubt this is really one of the professor’s students, it’s more likely an urban legend.

In a similar vein, and also from Japan - 'pointing and calling', used on the railways.


Meanwhile in Australia, Tram drivers just read the newspaper while driving: https://www.instagram.com/p/Bn-DGXYjNOl/?taken-by=browncardi...

This is a very useful concept even beyond physical design. When I evaluate UIs the top 5 things I look into are:

Visual rhyme, Accessibility, Affordances, Mistake-proofing (“Poka-yoke”), and Defensive design.


Take a look at this Audi production video and you start to notice poka-yoke being used when there is a human in the loop. https://www.youtube.com/watch?v=nZkyfZqlptA

In my experience these things are developed over time in response to failure. Always keeping the poka-yoke concept in the back of your mind helps when designing systems, but you don't really know how people will misuse tools until they're in action.

From my understanding, poka-yoke addresses the problem where "[...] the mistakes are allowed to reach the customer".

In software, we know that the later a bug is caught, the more expensive (in terms of time / resources) it is to fix. So we want to catch bugs / mistakes earlier.

The "simple" idea that Poka-yoke introduces is to make it harder to make the mistake in the first place. If a physical component can be mounted more than one way, but there is only one correct way, then design it such that it is impossible to mount it incorrectly. This can be done by making the mounting holes for the part asymmetrical, for example.

IMO, in software a similar concept would be to fail early and fail hard.

I guess that aligns with type systems in programming languages. The type systems make sure that the plain simple mistake of passing an integer in the place of a string is prevented. Of course more powerful type systems like Haskell fool proof it more if the developer is putting the effort of writing more types to restrict things more.

I think that's a clear implementation of poka-yoke on the programming languages level allowing developers to fool proof their APIs

I wrote about putting a line on your shower handle to denote 'good temp'. (used a shower crayon, or anything)

Now I put my handle in the correct location and its warm everyday.

This only works if you have a thermostatic mixer or consistent temperature of your hot water.

The gold standard is a pressure-balanced thermostatic mixer, which always outputs the same temperature at the same setting. Most of them come with two knobs, "temperature" and "flow". I miss the one I had a while ago.

I don't have a fancy mixer, but if your hot water heater is reasonably consistent and you have a working main pressure regulating valve, you should have enough consistency for this to work.

It's funny because you'd think that'd be the case, but in practice the difference is night and day. Thermostatic mixers will often have a temperature dial on the output, and it'll be within ~2 degrees of that temperature as soon as the hot water available exceeds that temperature.

Even without a thermostatic mixer, “good temp” is unlikely to be freezing cold or scalding hot. It’s a good enough first approximation that you can correct after you jump in

Yep, that's a good tip.

Read this, if you haven't: https://en.wikipedia.org/wiki/The_Design_of_Everyday_Things

It's a book about how minor design differences can basically change your life. It will make your life miserable, because you will start noticing all the stupidly bad design decision and blame the designer in most cases.

The term is used quite often in manufacturing. For a part when picking, for example, a part may have two colored squares on the label so the operator looks at the color on the squares to validate as an extra step. We call the colored squares the poka-yoke.

With a similar intention, Toyota also came up with the Andon cord - https://en.wikipedia.org/wiki/Andon_(manufacturing)

This concept is very useful, not only in manufacturing but also at home. As somebody else has explained, MegSafe used this principle. Another great example is our dryer at home: the filter that needs to be cleaned every month or so has to be plugged back in in a certain way. They perforated they filter such that there’s absolutely no way to plug in back in the wrong way. Another great application would be gas vs diesel knobs at gas stations. I’m not sure whether new cars/pump provide that already

I'd never heard of this, but it's an excellent concept that I try to apply to my daily living.

For example I accident-proof myself against spills by keeping a beverage in a place where it's unlikely to be knocked over, but if it is that the consequences are easy to deal with rather than soaking a keyboard or something. In physical terms it mostly boils down to minimizing the potential energy of everyday objects.

I think this aligns with the notion of an Exocortex, extending our brain and our IQ by making it easier to do the correct thing.

Whenever you invent better idiot-proofing, someone else will invent a better idiot.

A poka-yoke in real life is like those games for kids where they have to place squares with squares and triangles with triangles.

Does this qualify as "news" for HN? That's just a wikipedia article.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact