My favourite interview question
Some years back, I was interviewed by an interesting start-up in Palo Alto. I was really tempted to take their offer, mostly because:
- One of the interviewers was an avid Ultimate player;
- Stanford. In walking distance. (And the regular Stanford Friz-ball variant game);
- They asked me a question which has become my favourite interview question to ask and to answer.
The question they asked was:
How might you design a program that lets people play Monopoly with each other over the internet?
Of course there’s a lot to be learned on both sides when an interviewee is asked to design something and explain their design. Both parties learn a lot about each other’s communication styles and approach.
It works both ways: for example, if I were sketching out a design and the interviewers repeatedly interrupted me to discuss my UML notation, I could infer certain things about their culture.
Those kinds of issues apply to any reasonably sized design problem. Anything larger than, say, “write a procedure that reverses a string in place.” But let’s look at three of the characteristics of the game of Monopoly that I find attractive for this exercise:
- Monopoly is poorly defined;
- Monopoly requires more than simplistic object design, and;
- Monopoly is too big to be designed in one session.
First, Monopoly is poorly defined
. Chess, for example, is rigorous. The official rules of Monopoly
are silent on some critical questions and vague on others. A tournament-playing aficionado will realize this immediately, but it’s easy to guide a candidate to this realization by asking some of the well-known FAQ questions
This ‘problem’ makes it a great interview question. It drives a lot of valuable interaction between the candidate and the interviewers. You have to ask questions, make assumptions, and know when to stop gathering requirements and start driving the design.
There have been some blog posts pointing out that even trivial requirements can have many hidden implications
. A determined and finicky developer can drive questions for the length of the interview. I think Monopoly’s missing requirements actually improve its suitability as an interview question.
If the candidate uses up all of the interview time trying to obtain perfect requirements, we have a problem. In the software development I do, the requirements are never perfect. I don’t demand that a candidate try to create an agile, iterative process on the spot, but I look for someone who knows when to say “close enough, let’s move forward.”
Another good way to move forward for both interviewer and candidate is to say, “ok, we’ve covered the most important requirements. Let’s make a bunch of assumptions and document them. In a real-world situation we could obtain feedback on the assumptions after presenting an initial approach.”
After all, who’s to say that a programmer, designer, or architect is always 100% beholden to others for requirements?
The second reason I like this problem is that Monopoly requires more than just a simplistic object design
. What I’m about to say will be blindingly obvious to the Enterprise crowd (sorry, not you over there with the Greasemonkey
script that translates your web email to Klingon). The rules must be considered as carefully as the entities
. Enterprise developers have known this for years: that’s why you see rules engines, table-driven designs, and visual workflow editors in many Enterprise applications.
Now let’s talk about ‘object-oriented programming’ for a second. 99% of the stuff you read discusses modeling real-world physical objects. Things. Nouns. It is a Kingdom of Nouns
. Most candidates start every design by dutifully listing all of the nouns they can think of and then they spend the rest of the time available thinking about piling them into phyla, hierarchies of “IS-A” and “A-KIND-OF.”
This approach, which I will call “noun and verb,” is so limited I’ll dare to call it brain damaged
. In fun. But seriously, competent software developers spend much more time on relationships, responsibilities, and constraints than they do on trying to prematurely optimize
reuse through inheritance.
Now let’s ask a question about Monopoly (and Enterprise software). Where do the rules live?
In a noun-oriented design, the rules are smooshed and smeared across the design, because every single object is responsible for knowing everything about everything that it can ‘do’. All the verbs are glued to the nouns as methods.
Let’s take a look at a simple rules question. If a player owns Baltic Avenue, can she add a house to it?
Well, there’s a bunch of stuff about whether she can afford it and whether there is a house available in the bank. Where does that live? In the bank object? And there is a bunch of stuff about whether it is either the player’s turn or between turns. Where does that live?
And there is a bunch of stuff about whether the property already has four houses. Where does that live? Somewhere in the property hierarchy? Sounds reasonable. Now what about mortgaged property? If Baltic is mortgaged, the answer is no
. That’s easy. But what if Mediterranean Avenue is mortgaged? And what if, for example, Baltic has one house but Mediterranean has none? Where does that logic live? Both of these last two questions involve knowing something about the other properties of a colour group.
Now you can debate which verbs belong to which nouns, but here is an opportunity to step back a bit and consider the larger implications of maintaining such a ‘classical’ OO design.
Consider a ‘noun and verb’ design. First, an easy question. How well does the design document the actual game of Monopoly?
If someone were to read the source code, do you think they could learn how to play the actual game?
Is this a stupid question
? I think it strikes at the root of sustainable software development. In real world applications, software lives for a long time and teams change over that time. Programmers are often asked to maintain software with insufficient training in its intended use.
OO programs can be brilliant communicators in some designs (“here’s everything I need to know about a Money Transfer”) and terrible in others. I think this is part of the appeal and effectiveness) of the Domain Specific Language
approach. It does a better job of communicating its intentions than a simple OO design.
I’m not saying that a DSL is right for Monopoly. Or wrong. Just that it’s valuable to consider this issue when discussing an approach. There is a lot of opportunity for a candidate and the interviewers to learn about each other’s thinking that goes way beyond coding if you spend a few moments discussing the importance (or irrelevance) of a design that communicates its intentions. And I think that makes Monopoly a good design subject.
Another issue about the rules. In software development, requirements often change. What kinds of requirements changes are easy with the design? What kinds are difficult?
I suspect it’s easy to change the prices and names of properties, but very difficult to change the fundamental rules. Well, it’s probably easy to introduce the popular ‘lottery’ variation where fines are paid into a pool and anyone landing on Free Parking’ scoops it up. But some of the more esoteric variations would require making lots of changes to many different classes.
The ‘noun and verb’ design is tightly coupled, and distributes the rules across the design. If you really want a mess, consider that popular computer versions of Monopoly allow you to pick and choose rules variations for each game.
The key to making this easy is to find another way to represent the rules. I generally provide hints along these lines by asking the candidate how they plan to implement the Chance and Community Chest cards. What is the relationship between the cards, special squares like “Luxury Tax,” and introducing variant rules?
(You can probably save yourself a lot of interview time. Instead of all this hoopla, ask the candidate to describe when they have actually used the Strategy, Visitor, and Command patterns outside of a framework.)
The bottom line is that designing a Monopoly game requires giving a lot of thought to representing the rules. There are a lot of ways to do that. I’ve hinted at two, and you I’ve heard of a lot of other interesting approaches. It’s a great chance to really go beyond OOP 101.
Now to close as rapidly as possible with what I consider an essential characteristic of a good design problem, and one that applies here. Monopoly is too large to solve in one interview
. This is related to the vague requirements characteristic I mentioned above.
The reason this is important is that it forces the designer to pick and choose what elements of the design to solve in limited time. Some candidates will fail outright because their problem solving style is to consider all of the implications before starting. It’s easy to spend hours on a problem like Monopoly. But will you get up and apply the marker to the white board?
>Does a candidate start with the trivia? Who cares how to represent the colour of a property? CSS? A getter in each object? Yawn. I’ll prod him to move along.
Perhaps a candidate spends all of her time working in an area of the problem she knows well. Some candidates can spend forever engineering a Monopoly Online site up to Web 1.0
scale. Or building FTD.com
I’m not saying that’s bad. Or good. But I will say that given a problem too large to solve in the time allotted, there’s fantastic value in evaluating what’s important. And that cuts both ways. If you think that some of the OO design choices are critical but the interviewers hand wave it and want to see how you plan to handle 10,000 requests per minute, that’s revealing.Summary
I like asking (and being asked) about Monopoly because it provides fertile ground for discussing issues that are critical to judging both competence and cultural fit, like:
- Communication style and skills;
- Deep OO design philosophy;
- Prioritization and time management, and;
- Handling ambiguity.
And now, if you have a moment, I’d like to ask for your feedback. Is there a better design question to ask? What is your perspective as an interviewee? What other issues about a design should be discussed during an interview?
I’m looking forward to hearing from you. Thanks in advance!Update: A cautionary note
Some people have objected to this question because it may be a trivia challenge: how well does the candidate know the game of Monopoly? In the comments I mentioned that the question really only applies to candidates that already know the game.
Others have suggested it may be a form of “guess the design I’m thinking of” where there is exactly one right answer, the design the interviewer has already come up with. I think this is a legitimate concern.
It really applies to any
interview question. For example: “name your three strengths” or “tell me about a challenge you faced in your last position and how you overcame the obstacles.” If an interviewer is looking for something specific, they can and will twist any interview question into a variation of “guess the answer I’m thinking of.”
I won’t say that Monopoly is somehow better than other questions in this regard, nor is it worse. If you have an interview based on reviewing a candidate’s past accomplishments, don’t you think there are some interviewers that will discount those that don’t fit their personal criteria for merit?
I once had an interview where the interviewer spent forty minutes talking to me about my education, and something like fifteen talking about my fifteen years (at that time) of actual career experience. Bias in an interviewer is orthogonal to the choice of interviewing technique.
What do these two cautions have in common? Only that the most important thing for a successful interview is that the interviewer be genuinely seeking out a view of the candidate’s strengths and weaknesses. All I can really say about this question is that if (and it’s a big if) the interviewer is sincerely attempting to evaluate a candidate’s technical ability, and if the candidate has some familiarity with the game, then the session can be very productive.Update: Some relevant links
It would be ludicrous to think of hiring a juggler without first seeing him perform. That’s just common sense. Yet when you set out to hire an engineer or a designer or a programmer or a group manager, the rules of common sense are often suspended. You don’t ask to see a design or a program or anything. In fact, the interview is just talk.
The bottom line in my interviewing technique is that smart people can generally tell if they’re talking to other smart people by having a conversation with them on a difficult or highly technical subject, and the interview question is really just a pretext to have a conversation on a difficult subject so that the interviewer’s judgement can form an opinion on whether this is a smart person or not.
Think of three things you want the interviewer to know about you that you think they are unlikely to find out if they ask all the questions. The important ideas are that (a) you want the interviewer to know about each of the three things, and that (b) the interviewer is unlikely to ask about all three if you don’t exercise some control over the interview.
In case you didn’t know this, about 1/3 of all technical interview questions worldwide are variants of “How do you fit 10 pounds of crap in a five-pound bag?” Examples: how do you reverse a string in place, how do you sort a billion numbers, how do you write a decent compiler back end for an architecture with only four frigging registers, how do you handle fair scheduling when someone just forked off 1000 copies of some Towers of Hanoi simulation, etc.
I’m exaggerating a lot, but the point is, when you select 1 out of 200 applicants, the other 199 don’t give up and go into plumbing (although I wish they would… plumbers are impossible to find). They apply again somewhere else, and contribute to some other employer’s self-delusions about how selective they are.
And then there is the issue of why would someone care if I knew how to write a file copy function, ahead of what the difference between an interface and an abstract class was. Or when would I use encapsulation? Or in my design, how would I choose between polymorphism or inheritance? Or what pattern would I recommend for dealing with a general or specific problem?
I know lots and lots of interviewers, at many companies, who’ve decided that they can fully evaluate you based on whether you can solve some particular convex optimization problem (or graph-search problem, or logic problem, or whatever their pet Elephant Question is), and they ask every candidate this question regardless of their background or experience. In fact, I’d estimate that some 10% of all technical interviewers ask the same questions, year after year, and they could care less about your experience.
Labels: jobs, popular