Hacker News Reader

Comment by wink | Hacker News Reader

I wonder if the Joel test is still the gold standard for most people.

Especially "Do new candidates write code during their interview?" seems to run contrary to the much decried whiteboarding (not taking sides right now, also it's not about paper vs whiteboard etc)

I still think the questions are a useful data point, just never agreed with the "10 and less is a failure", because it's just too broad and some of the points the teams I worked fulfilled 100% in spirit, but never to the letter of the test - but then you're in wishy washy gray area again. (One of my main gripes is that if you compare Excel (iirc that was his team) probably has "a single build" with dozens? of developers - so I'd rephrase it as "all your projects have a releasable HEAD/master branch" but I had people disagree on that premise.

View on HN

"write code during interview" was always about real coding, using computer. No paper, no whiteboard, the best way to do it is to give the candidate access to some real code, like a company's internal project, and real task, like a nasty bug to solve, or a small new feature to implement.

I had an on-site coding interview at a place where they asked me to implement a small iOS project on my own computer doing things however I normally would. Had access to documentation, SO, could search google for whatever. They gave me a couple hours to do as much as I could - there was another dev around that could answer questions I had. Afterwards, I walked them through what I had done, answered their questions, responded to feedback, etc. This was code interviewing done right in my mind.

Whiteboards are for design, computers are for code.

Good point, I had totally forgotten we had that at a past company. Give the candidate a computer, a simple task, someone to ask and check results after 2h.

The criteria "Do new candidates write code during their interview?" is something I consider pretty important. I would say it makes more sense if rephrased in a less polite way: will all my peers be able to write a working line of code.

Overall I have to say I'm really impressed with how the list has held up. The questions are pretty specific, and reasonably diagnostic of common issues that create spirit destroying dysfunction.

The joel test seems very applicable to a scenario where you have a large team of people working on a small number (or even just one) of big product-oriented projects.

I find that a good chunk of it doesn't apply in my job, where I'm one guy working on a large number of simple web applications.

Reply on news.ycombinator.com