The creepy capital efficiency of Goldman's cafeteria | Hacker News Reader

The creepy capital efficiency of Goldman's cafeteria

cnbc.com

34 pointspearkes49 comments

It doesn't sound capital efficient at all - instead they've got people deliberately milling about doing nothing while waiting for the "penalty window" to lapse.

If anything, this is a brilliant example of how applying measurable incentives can distort motivations and make people do stupid things to please whatever metrics are being measured.

Left to their own devices, these very smart and ambitious people would no doubt make up their own mind about the value of their time and ensure they don't waste time milling about when they're busy, and so go to lunch early or late or in the middle if they're not too busy anyway or want to chat with someone in the queue. Instead, they're now forcing themselves to fit a stupid "penalty window" to save a few bucks, because that's what the incentive system in place dictates.

Measurements are a very, very dangerous beast. Apply with caution.

(Great book on the topic: http://www.amazon.co.uk/Measuring-Managing-Performance-Organ... )

>"they've got people deliberately milling about doing nothing" //

But that also means that those who need to get a quick lunch have a window towards the end of the cost penalty period in which they can breeze through the line and more easily get a table.

I like the idea in general, but presumably it's the employees own time they're wasting.

Surprised it's not a graduated system, like a normal curve with a plateau over the traditional dinner hour.

Do Goldman also allow staff to have flexibility in their work hours - that would tend towards reducing choke points like this I imagine.

> But that also means that those who need to get a quick lunch have a window towards the end of the cost penalty period in which they can breeze through the line and more easily get a table

Doesn't seem that way. From the article.

> "If you find yourself in the cafeteria sometime around 1:20 pm, ... the cafeteria area between where the food is collected and where you pay is quite crowded. The Goldman lunchers are chatting with each other, waiting for the final minutes to tick down until they can save a dollar or two."

In other words, if you want to breeze through the queue, you have to come after the 1:30pm queue for the cashiers has subsided. If you want to get a table, you may have to come even later.

In general, I find this to be a bizarre and overly complicated solution to the basic problem of 'time wasted in queues'. There are may other ways to tackle this but it doesn't surprise me that an investment bank thinks manipulating prices is the 'right' way to do it.

> Do Goldman also allow staff to have flexibility in their work hours - that would tend towards reducing choke points like this I imagine.

Well, the trading hours of the markets determine, more-or-less, the required work hours of many employees. However, successful traders will be at work both before and after the market trading day to preview/review the events of the day.

Is it a great book with advice even small companies can use, or just an interesting read? The only people I manage are contractors via oDesk.

My company is small (11 people, £1m turnover) and yet this book provided some solid data and insights to convince me that measurement-based management is very dangerous (which I was thinking about already, but I was lacking a solid explanation and some data about).

The point of the book is relatively straightforward, but the explanation is pretty solid, and it provides some data to substantiate that point, and a model about which approaches work in which situations. Imho it's a very valuable read if you're in a position where you're designing an incentives system (e.g. if you're trying to create a company with employees).

For contractors via oDesk - it will still help. The model is valid in general. You won't necessarily be able to do all the things that you would with a team that's physically there, but the general ideas still apply.

instead they've got people deliberately milling about doing nothing while waiting for the

You're equating the time wasted due to the probably large lines during the rush hours to the time wasted by a few people hanging around for a few minutes before 1:30.

I object to your automatic assumption that those are equivalent and that the people who put the policy in place have no clue as to whether or not their incentives were a benefit. Without actual data, you're just giving us your preconceptions and labeling them as "a brilliant example".

So instead of a hump at 12:30, they've created two humps at 11:30 and 1:30?

Why not offer a continuous discount curve? If you show up at 1:25, you get 23% off your meal, so there is only a 2% discount for dawdling 5 extra minutes in the food line before paying.

Sorry to be offtopic, but is your nick a typo? The german "saalwächter" would translate to "saalwaechter".

Yes, but an old typo. Name got misspelled about a hundred years back or so.

I was thinking the same thing. Continuous incentives are better, to the extent that they're not too complicated. But it's Goldman Sachs - I'm sure they can figure it out.

This was my thought too, hand out a set of discount cards (or passbook barcodes) with a moving window from 11:30 to 1:30 on them. If someone doesn't like their time frame, they could swap it with a different employee.

...there's almost an app in there somewhere.

I don't like pre-planned organization, it adds cruft. Maybe just do a "current discount" depending on the people in the line, and put it on a mobile website ?

I love how some companies are "smart" but get every details wrong anyways. When I was in Argentina in 2003, there were giant internet cafés (up to 250 computers I believe) were the price of the minute of internet was floating depending on how many people were online: you had a max price, a min price and the current price was just the proportion of used computers (so it was not a bidding market). You had a few screens on the wall telling you the current price (I think they billed at the beginning of each minute). In the middle of the night the price could get quite low.

Even that relies on a model of the utilization of the cafeteria over time. If Goldman wanted to incentivize spending a minimal amount of time in the cafeteria, they should just measure that. For example, offer a discount of (25 - X)% where X is the number of minutes between entering the cafeteria and paying for your food. People would learn to anticipate events such as backed-up food or cash register lines and adjust their behaviour appropriately.

A continuous discount curve would be hard to implement and hard to explain. The two humps will be smaller than the single big hump. The 11:30 hump would be a race to beat the cutoff. The 13:20 hump is the delaying described in the article. And there's probably still a hump at 12:30, it just doesn't last an hour anymore.

Could do, but why tie to time? What you actually trying to minimise is time spent queueing. So record when people arrive and apply a discount/surcharge depending on how long they queued (or just based on the number of people ahead of them in the queue).

Seemingly perversely, a long wait would imply a higher cost for the meal.

It's basically congestion charging based on queue length/time rather than time of day.

pub/sub the queue time discount automatically from register transaction rate; client side compile local vendors and approximate arrival rate by gps.

The main advantage of tying it to time is that you can let people plan for it. If you have dynamic pricing based on unpredictable conditions (how many people are in the cafeteria right now?) you can't plan for it, and the best case scenario is that employees get up, go to the cafeteria, see that the demand/price is too high at the moment, go back to their desk, come back 10 minutes later, repeat until the price/demand is low enough.

Unless someone has observed otherwise, I'm betting continuous wouldn't work. The difference between "now" and "later" won't be as big, and it will take effort to calculate, which makes it feel smaller too. To effect people's day-to-day habits, you need to present a motivator to the affective decision centers.

I have completely forgotten about the discount outside of rush hour at Goldman. It just didn't figure much in my lunch time planning, and I loved lunch.

The one remarkable thing about their cafeteria that stuck with me was the cutlery. All disposable utensils outside of chopsticks were made out of biodegradable corn and they were the best disposable utensils I have ever used. The utensils were full sized and the right stiffness. There were no sharp edges and the texture was slightly rough, much nicer than plastic.

Since leaving GS, the only thoughts I had about their cafeteria was why haven't I encountered as nice disposable utensils as I had there.

I can't quite find the brand and google is drawing a blank, but I have recently seen some disposable metal silverware (not coated plastic, real metal). I guess some french luxury brand, like Starck or Degrenne.

Just because you choose to throw something out and have it buried in the ground does not make it disposable.

It was made and distributed as disposable.

I got the explanation by someone who decided to keep some instead of throwing them.

I had to look this up because it sounded so crazy. It Is: http://www.wna.biz/UserFiles/File/Metallized_Disposable_Cutl...

They coat plastic utensils with stainless steel, and then you throw them away. Unreal.

I precisely said that I was not talking of those.

Actually when I saw them at first I assumed it was one of those, but upon further inspection, it was a full metal thing and then I got the explanation that they come from a very pricey disposable thingy (I think it's Fauchon takeaway food).

>>nice disposable utensils as I had there

Why does this not surprise me about Goldman Sachs.

sounds like the cutlery at Biogen Idec. Similarly biodegradable and very lovely.

Do you mean these?

http://inhabitat.com/spudware-cutlery-made-from-potatoes/

Potato, not corn, but they're good.

They look right. Exactly the right color. Spoons look a bit different from what I remember, but I think this is it. Thanks!

Good job figuring it with my Dan Quayle moment (corn vs potatoe).

Meanwhile, very far from all of this, with no press, no hacker news front page, no praise for their cleverness, street restaurants in Brazil located on areas with high-density of corporate offices have been doing this for years. No capital efficience fancy wording, they just realized that people were going elsewhere because they had a very long line and did something about it.

Funny how hype and self-importance reinforce themselves. This is big news about what looks like a fantastic idea from the incredible minds of insanely clever bankers. If you are Goldman Sachs, whatever you do must be clever. If you are CNBC whatever you talk about must be important.

Applying a lunch hour surcharge is probably the simplest way to even out the load a bit if that's your only problem.

But I know food vendors in various places around the world with near-constant long lines, and surely the answer in that case is to increase capacity, take orders in advance, maybe do delivery, etc. High demand is a good problem to have, but it is a problem, and it usually doesn't get addressed.

It's a fluff story about a powerful company (local to CNBC and active in the same circles) that does well with capitalism and efficiency using market principles in everyday ways.

I don't think it was intended to be a "first ever in the world" slight against Brazilian street restaurants.

Meh, slow news day overall what with the resolution of the End of the World(tm) non-crisis over the debt ceiling.

They did this because they're obsessed with avoiding queues and doing everything efficiently, but have a culture where everyone will queue up for 10+ minutes waiting to avoid paying a couple dollars more? Seems silly the way the author explains the motivations.

Perhaps the implementer of the scheme likes to get lunch at 1:20pm? Jump the queue, get what you want while the poor saps wait in line - now that sounds like capitalism to me.

"Everyone" doesn't queue-up for 10 minutes.

They're not taking this far enough, imposing a blanket 25% discount nothing but autocratic perversion of what could be a functioning lunch time credit market. They could maximize revenues and employee throughput in real time! Sure, there'd be some overhead, but financialization axiomatically improves efficiency. Perhaps there could even be a lunch time credit futures market for people who occasionally work from home or on business trips?

How can you be trusted to manage bond traders if you can't even properly manage a cafeteria?

Gattaca, not Gattica. A, G, C, and T.

As others have noted, this isn't efficient at all. They've incentivized some of the highest-paid (ie. most expensive) employees in all of business to waste their most valuable resource - time - in pursuit of a token savings.

The folks at GS are extremely smart people, which leads me to believe that this isn't unintentional at all. Instead of achieving actual savings, they have:

* An effective PR stunt. * A low-cost signaling mechanism for employees to show their dedication to Goldman's "cultural values"

Make no mistake, this has nothing to do with "capital efficiency" and everything to do with "image"

It seems like this discussion has forgotten that there was an enormous line at 12:30 that was wasting more people's time.

One concern I have here is I know my productivity is impacted when I'm hungry. It's quite possible this is optimizing for time-at-desk at expense of worker productivity.

A lot of people are mentioning the idea that this should actually be a continuous curve... As someone who has used the cafeteria for a couple months, I would say this is most likely an over optimization. You are creating a complication for every transaction, and making it so now there is now set points where you can save the most money, namely at the beginning/end of the window. Which actually creates more of the two 'two hump system' swombat, saalweachter, smickie and others have mentioned. Just like there is a natural gradient to lunch from 12 to 1, you still get a gradient in the discounted hours. 11:30 isn't a better time to eat lunch then 11:55, It's simply an attempt to split up lunch times so that the cafeteria can effectively have 3 rush hour instead of 1. And since traffic effects aren't linear, you get a much quicker overall experience as a result. "The people standing around" is overplayed, Those are the people's elevator rides who were are little shorter/longer than expected.

They should use entry tickets, valid at a specific time and date. Sold at auction. With a second hand market for when you change your mind.

"Creepy" would be if they had people test blood sugar levels and only admit them into the cafeteria when their blood sugar starts dropping too much.

This is just mildly interesting. It's not like differentiated pricing for peak periods if particularly unusual, though admittedly I haven't seen it at a cafeteria.

> Goldman beat estimates but many analysts say a driver of this was Goldman's ability to cut costs

If they ever cut costs somewhere, it's from the pockets of their clients.

http://www.theguardian.com/business/2013/aug/01/fabulous-fab...

Am I a commie if I think Goldman Sachs is a bit creepy in itself?

So is there a derivatives market for the food purchased at a discount? Seems like an opportunity for arbitrage. Corner the market on meatloaf!!!

I refuse to read an article that cannot even get its Sci-Fi references straight.

As usual a technological solution would have saved more money. All Goldman would have to do is let people reserve their place in the lunch line through their cellphone or at their desk. When their number comes up, they can go right to the front of the line and order. There would be a specific physical way to access the front of the line when your number is up.

The people who don't use this policy would quickly learn to use it since the others would be constantly skipping them. So that would push everyone to use this system and avoid the physical line. The system could even show them estimated time remaining via an app or webpage.

For that matter, Goldman could have its own app for its employees. So can many other organizations. Lots of government inefficiency comes from this. The NY DMV finally put forms up online!! Imagine how much more efficient it is to fill out a form online than waste half a day standing on line (not online) to get the right form, then fill it out and wait for your # to come up to finally do something.

By the way, if you liked the above idea of organizations making apps for their employees, reach out to me through http://qbix.com/about . We spent about two years building the technology for making such apps for organizations. We could really use someone with good PR or Marketing experience to partner with us though. And developers are always welcome (if you use Node or PHP.)

Reply on news.ycombinator.com