The Beer Game (1992) | Hacker News Reader

The Beer Game (1992)

web.mit.edu

58 pointsdunefox29 comments

I love the beer game as a management simulation. You get to see how difficult communication and future uncertainties can create positive feedback loops in supply chains. Those positive feedback loops cause the supply chain to break completely, and, in most cases, never recover.

My mother and I actually "solved" the beer game a while ago based on the simplifying assumptions that the beer game makes. It turns out that the actors other than the factory actually have no power over inventory in the system, and since inventory costs the same everywhere, you should be holding 100% of inventory in the retailer and just passing inventory along the supply chain. The factory is the only player that should be making decisions about how much to produce (everyone else should just order what is coming along the line).

That reduces a complicated system into a fairly simple control system with a long delay, which would be hard to solve except that the beer game is based on having almost no dynamics! The relative lack of dynamics lets you easily create a stable control algorithm at the factory to keep system-level inventory and backlog at 0.

That said, the facilitators of the game will often taunt you to try to create dynamics, and you are not supposed to communicate with other members of your team (except when setting initial strategy and meeting each other) so chaos generally ensues.

> It turns out that the actors other than the factory actually have no power over inventory in the system, and since inventory costs the same everywhere, you should be holding 100% of inventory in the retailer and just passing inventory along the supply chain

I was expecting this result despite never actually playing this game and only knowing things from the article alone.

If this game is what business-managers are playing to learn dynamic systems... then it is almost certainly the source of "just in time" inventory management of today's supply chains, then something like that is natural.

So the question is: is the model wrong, or is just-in-time inventory management (for middle groups, like distributors and wholesalers) really the best way to manage inventory?

I think there are two simplifying assumptions that are problems:

1. The cost to hold inventory is equal at all parts of the supply chain - this is demonstrably not true for most supply chains (in fact, it's usually cheapest to hold supply somewhere in the middle of the chain), but retailers like Amazon and Walmart have figured out how to make this almost true. That allows them to operate with less slack.

2. There are almost no demand dynamics. This seems to usually be a good first-order approximation, but we saw during the pandemic that there were some demand shifts that the system was unprepared for. You can handle dynamics in demand if you build slack into the system (in the case of the beer game, this means controlling inventory to some positive number).

I think we were caught in a situation where 1 and 2 were both considered to be true, and we just violently exited that regime. It will be back - we will forget the demand shock and JIT will return since it has the lowest unit costs - but for now we are facing the full brunt of the bullwhip event while we build up stockpiles.

JIT never went away, why would it? It was never applied for parts sourced from overseas anyway. Where exactly inventory is best located depends on the particular supply chain and part, sometimes it's the middle sometimes it's somewhere else.

And the bear game is all about demand dynamics, primary demand is almost stable. All it takes to trigger the bullwhip effect is one stock out, and those are intentionally built into it.

The beer game was purposely designed to isolate human mistakes as a cause of instability, separated from weather and stuff like that. What weather, technological changes, etc. do is it makes it impossible to design a simple control system.

The beer game is a quick, easy and fast way to show the behaviour of supply chains, and the dreaded bullwhip effect. It is not meant to simulate whole supply chains or teach inventory strategies.

The answer to your question is, no the beer game isn't wrong, and it depends. Because there is no "best" way for anything in supply chain management.

So, so glad to see this shared here. I've had many friends ask if they should get an MBA and I almost always say no, and give them a reading/activities list instead, which ALWAYS includes the Beer Game. For better or worse it was one of the most illuminating parts of b-school and it's still woefully under-studied and under-practiced today. ESPECIALLY today, with all of the supply bullwhipping going on.

Would you kindly share said list ?

I would also appreciate the list please

Here's a short list:

Strategy/Ops: • The Innovator's Dilemma - Clay Christensen • The Toyota Way - Jeffrey Liker • The Beer Game - duh :-)

Leadership: • High Output Management - Andy Grove • The Hard Thing About Hard Things - Ben Horowitz • Love is the Killer App - Tim Sanders

GTM: • Start With Why (Tedx Talk) • Benefit ladders (Human Centered Design exercise good for product AND positioning) • Everything on Kellblog • Everything from Andy Raskin (especially "the Greatest Sales Deck I've Ever Seen")

A couple years ago Llamasoft/Coupa released a beer game optimized with reinforcement learning.

https://beergame.opexanalytics.com/#/

I don't know where they took it, but my startup at the time saw huge gains in performance applying deep RL to industrial ops and supply chain problems like that. Shifted the whole Pareto frontier.

Any writeups on either?

DM'd you on Twitter.

The Beer Game -or- Why Apple Can’t Build iPads in the US - https://news.ycombinator.com/item?id=3802515 - April 2012 (246 comments)

Played this at university, can confirm it ends in utter chaos.

Does this game exist in a easily accessible board-game format? If not, is there some PDF or something that can help a "board game enthusiast" like myself set it up and play it?

I see that this article discusses a "deck of cards", as well as various records that must be managed, and a "Stack of pennies" representing something. So there's some game that really exists out there.

There is a board game, in the form of a roll-out playmat and some decks of cards. I don't know if you need to email John Sterman to get it, but it does exist. You also need poker chips or a lot of pennies to represent inventories, and a pen and paper to log how you do.

This is a co-op game made for business school classes, and you won't be able to get the full experience of it unless you enforce the "no talking" rule between each player, and have some other people to compare against.

As a game with a punchline, it's 1) not worth playing twice and 2) the owner/proctor of the game already knows the punchline.

There's probably some good way to make a bullwhip effect game, maybe with each player playing all stages, but each stage in a different supply chain (and maybe with the retail availability of the product of one supply chain generating the demand on another.)

For games in the same ballpark, Schoko & Co., Container, and Energie Poker are good ones.

Something similar can be found in the book “The Goal” but with a six sided die and matchsticks. The premise is explained here: https://math.stackexchange.com/questions/2149283/math-behind...

Forio has a digital version https://forio.com/app/showcase/near-beer-game/

I was on a team that wrote a beer game app for HBS, and the rules are really simple. It's not really worth playing more than once, though.

It's kind of interesting to recognize that, at least with beer, the whole thing can be circumvented/disrupted with a 'radical new technology' like widespread homebrewing. I once had an enterprising neighbor who dove into this project, using large (new, clean) garbage bins to prepare grains, and a repurposed gas-burning hot water heater as a brewpot. Each batch was ~50gal/200L of quite decent beer. (This incidentally is somewhat above the 'home use' limit and bordered on a commercial operation).

It's not likely to see anything like this in the world of computer delivery, although the widespread interest in repairing things like iPads and iPhones is becoming a factor. As Moore's Law starts to flatline, it's becoming clear that the next gen of phones/tablets etc. will not be stupendously superior to the older versions, and consumer demand for repairable computers will only rise in the future, and expected usable lifetimes will become comparable to those for automobiles.

If you're feeling extra froggy, here's an implementation of the game [1]

[1] https://forio.com/app/showcase/near-beer-game/

I like this idea, but what is the ultimate lesson, other than "chaos ensues, and it's nobody's fault in particular" ?

I mean, it doesn't seem to provide any guidance on how to actually solve these problems, only make you aware that they exist. Usually a "flight simulator" experience would address that side of things, too.

I suppose a worthwhile takeaway is "don't be so quick to blame others," which is good for keeping humility, but that doesn't address the mechanics side of things.

Systems have properties beyond individual actor choices or actions. If it causes harm (bullwhip effect here), identify the systemic factors at play, and try to change the system.

Here, lack of communication was one of the primary causes of the chaotic effect. So improve communication across the supply chain. Slow feedback loops and limited communication are likely to produce this bullwhip effect regardless of the intelligence or discipline of any of the individual actors (retailers, wholesalers, factory). They have no way to know, at the factory, that the surge is only temporary.

More broadly, examine whatever systems are at play in your work and industry. Are they producing positive or negative results? If positive, try to reinforce them, if negative try to alter them.

Well, the lesson is that communication and collaboration are the key to mitigate the bullwhip effect. That, and preventing local optimisation.

I guess it's mostly just about putting people in a stressful management situation to get them used to it. Not much of an ultimate lesson there, just some experience so they can keep calm once they encounter it in the real world.

The Beer Game is a simulation of something that takes place over weeks and months and years in the real-world (you don't normally double or triple your manufacturing capacity in hours, unless you went from one employee to two). It's not about experiencing a "stressful management situation." It's not the same as pilots in flight simulators experiencing a simulated real-time disaster in the making and having to respond correctly (or at least more correctly) in real-time.

The objective of The Beer Game and the other materials related to it and the research by Forrester et al. is to develop an understanding of systems dynamics and systems thinking. In this specific game the systems effect they are exploring is the bullwhip effect and how imperfect, incomplete information can cause it. Other games in this genre of games may have specific other systems effects they aim to explore, but in total their objective is to get the participants to start thinking about systems. Not to help them avoid panicking (though thinking clearly does help with that).

Played this as a Boeing intern. It was wild and so much fun.

I am sure there exists a beer game that is designed for software "production" not beer.

if not I am pretty sure it can be built - and I think it would be instructive

- delays in getting code out to production mean the fix that allows for beer to be sent to different pubs once a day is not available ?

- Lack of specifications means ...

Reply on news.ycombinator.com