patrick.net

I prefer a small project that I do away with a deadline. Tests no I wont do it at all. For the same reason we all came with different result for the same task in your thread about Node.js.

I avoid interviews where I'm not talking to a technical person who can gauge my skill level just by interviewing me.
Or the owner of the company directly. I'm not the guy that works in a modern dev shop. I am the modern dev shop.
If they want it done in less than three months, and done right they call me. If they want a team of 5 coders and 2 sql halfass dbas and non technical project manager then they obviously got the 2 year budget for them to fail to deliver on time. Maybe it's South Florida but shops like that in South Florida are all revolving doors of failures. Because of the non Tech SJW Democratic meetings software developing process.

I work with another guy at work but he owns the part of one side of the Enterprise and I own the other. When they talk about hiring other people, we let them know they'll need to get all 7 of them because we'll probably leave. Their public facing website managed by the Design Team and done in open source and PHP, is a revolving door of Drama we want no part of that.

somecrappynumber says

Out of curiosity, why do you deem this a requirement of the company?

Respect and good will.

Before I hire a plumber to fix my leaky faucet, I insist that he demonstrates his technical skills by building an aqueduct.

On a serious note, how should companies evaluate prospective software developers? How should expertise be assessed?

I think pair programming is pretty good for evaluation. Just sit with one of the company developers and work on something for an hour or two with them. Back and forth, get used to each other and the levels of skill. Even make something useful to the company! Why waste time on frivolous interview questions when you could really get something done?

rando says

I think pair programming is pretty good for evaluation. Just sit with one of the company developers and work on something for an hour or two with them.

How is that not an hour or two long test? And it's one in which the applicant is thrown into code he's completely unfamiliar with to solve an overly specific problem. Sounds like the worse kind of test, or a way to get free labor.

I agree that tests are a bad way to evaluate skills, but currently they are the only way. Having been on both sides of the interview, more the interviewer lately, I have looked for better ways to evaluate and compare candidates. So far, all ways I've found are tests in one way or another.

You can ask them to write a function in pseudocode on the whiteboard, but that's a test. You can tell them a problem and ask them how they would solve it, but that's still a test. The better tests let you see how the person thinks, but even then you still have to test their technical knowledge on a platform because most people lie embellish their resumes.

rando says

Even make something useful to the company! Why waste time on frivolous interview questions when you could really get something done?

When I interviewed for a job 3 years ago for a Oceanic Satellite manufacture company. The guy gave me a url to a service and make anything that can consume that service. They normally mean a Console app or just a simple post or get to consume the methods.
The data came back with a guid and a bunch of comma separated numbers. I figured Ah Points! Dug around more in the wsdl I got from the service and got the schema for other relational objects. I used the points by enumerating through the data in pairs to extract the x,y .
Put the httprequests to the service endpoints in the model, created controller and in the view, I plotted all the points onto Google Maps. In less than 12 hours. Having the schema I got feature rich and creative with it. I can script 90% of my code and focus on business logic and Javascript.

Then they didn't call me back after 3 days I went on a rush interview for the company I'm with now. And of course I liked the industry, company and people and I get to Usecase my technology which has been a smashing success btw.

A week after I submitted my project to that Sat company, they called me wanting me to start right away. They said the guy that makes the final decisions was away and couldn't give the go ahead. I said I'm sorry but I got the other offer.
They were blown away by what I had done, said my model and design was better than what they had been using and asked me if I mind if they use it and build on it.
I said I hope they do.

Dan8267 says

I agree that tests are a bad way to evaluate skills

I think tests are fine! I just want the company to put in as much time and effort as they ask of the applicant.

Pair programming is probably best though. "Let's solve this problem together. How would you approach it?"

Tenpoundbass says

And of course I liked the industry, company and people and I get to Usecase my technology which has been a smashing success btw.

That's great if they ask you to create something useful. But if it's a typical interview project, it's not really re-usable at all.

few years ago, truecar.com's recruiter wanted me to do a homework before the interview. it was a small programming project would take a few days and i had a week to do it. beach view office and good pay. i turned down the job not due to the homework but because:
- very low work life balance
- very high turn over rate
- i sensed great hesitation when the recruiter said they could beat my current pay after he asked me how much i currently make

i accepted an offer with a much bigger and more profitable company which paid about the same. they didn't waste my time. they asked some very easy programming questions on the spot and i had to meet 6-7 different people. but all on the same day.

i sent an email to truecar's recruiter saying i changed my mind and won't continue with the homework but i thanked him for contacting me. he didn't even bother to respond. that said something about the company.

a bit of irony. turns out the new company gives out free stocks every year which i didn't know at the time. and our stock price has almost doubled since i joined. so i actually THANK truecar for their stupid homework requirement. without it, i might have interviewed, worked for them and missed out on this GREAT opportunity.

RealEstateIsBetterThanStocks says

they didn't waste my time.

Yes, that is one mark of a good company.

Dan8267 says

You can ask them to write a function in pseudocode on the whiteboard, but that's a test. You can tell them a problem and ask them how they would solve it, but that's still a test. The better tests let you see how the person thinks, but even then you still have to test their technical knowledge on a platform because most people lie embellish their resumes.

I think we have the right answer here.

I once had to take a test during an interview. It was a joke. It did not even cover the skills that were needed for the job.

RealEstateIsBetterThanStocks says

a bit of irony. turns out the new company gives out free stocks every year which i didn't know at the time. and our stock price has almost doubled since i joined. so i actually THANK truecar for their stupid homework requirement. without it, i might have interviewed, worked for them and missed out on this GREAT opportunity.

Username does not check out!

A tried and true technique is to drop trou and take a fresh steamer on the table, pull up, and while admiring your work with pride, ask the interviewer what they think.

rando says

Tenpoundbass says

And of course I liked the industry, company and people and I get to Usecase my technology which has been a smashing success btw.

That's great if they ask you to create something useful. But if it's a typical interview project, it's not really re-usable at all.

I was saying the second company after I interviewed at that Sat Com company. That's the one where I'm putting my Enterprise integration technology to work.
The interview project, I was just curious to get all of those comma delimited numbers into a set of points, so I could post it to Google Maps and see what those Ships courses were looking like. It was just fun trying to comport that over to an MVC model since I was interviewing for more than besides that job. It was keeping my skills fresh and gave me a chance to create something in the latest .Net version.

When I'm unemployed I like to take a Month or two off and just Develop some of the wild Ideas I had while I was too busy working to make them.
While using the newest features of the latest language releases. It then gives me a chance to go on an Interview and convey that have used those techniques or frameworks. Rather than explaining why I think I can that it's not that much different, it's just new features. It's a ten minute learning curb at most.

Patrick I suggest you check out manufacturing. They love talent, and are usually owned by families full of piss and vinegar and love independent self starters.
I'll never work for a Tech or Financial industry ever again. Also in Manufacturing, it's easy to find niches for Software that industry has a boxed solution need for.
Leave the Facebood and Twitters to the hell they have made for themselves. They are all on their way out anyway. Especially after Steve Banon regulates them like a Utility.

As 3D printing becomes more common place they will need more Delta Automation that can roll custom applications to batch print jobs into XML and create a work flow for the finished goods. As everything gets automated and robotic more people will be needed to make sense of all of the digital property these robots will create and consume.

It's the one place that has no room for SJW's our 5 top competitors have been crippled for the last 5 years, I single handedly surpassed all of them and made my company the industry leader in less than three.

I think coding tests are very reasonable, and represent time well spent. Coding is an excellent, and time efficient way to prove that you know how to code - particularly great at the initial screening phase.

I don't feel there's good reason why time needs to be equally invested by both parties. In fact, I bet in some cases, the number of applicants is much greater in numbers than the team doing the hiring, which at some point makes equal time investment infeasible.

I will agree to a point with you, in that the initial test needs to be brief. I think coding tests, (and the entire interview process) should be done in phases, so that the applicant can fail fast if they lack the skills, respecting their time.

I think as the applicant pool gets narrowed down, if there's some candidates who look promising, it's reasonable to ask them to do a longer coding exercises - if there's a good chance that they will advance in the hiring process should they perform well.

I will agree to a point with you, in that the initial test needs to be brief. I think coding tests, (and the entire interview process) should be done in phases, so that the applicant can fail fast if they lack the skills, respecting their time.

Tangentially related: I hire mechanics and have found that interviews alone and longer work tests can become unreliable predictors of job performance. The best interview technique I've found is to give the candidate one of these

http://3.bp.blogspot.com/-LD3iDIgySCM/T2iqLkm2RII/AAAAAAAABx4/sjCmNmO4g6o/s1600/Heart+Breaker.jpggoat says

and see how they go about taking it apart under the pressure of me watching like a hawk. To a one, people who figured it out quickly and enjoyed it turned out to be quality, productive mechanics. Those who struggled for a long time or never figured it out sometimes even with help did NOT make good mechanics.

I imagine a relatively brief test coding problem would be the equivalent touchstone for candidate quality in the software industry. If the INITIAL coding test takes 4 hours, it sounds like they're doing it way wrong.

Tenpoundbass says

Patrick I suggest you check out manufacturing

Probably very few, if any in his area.

Automan Empire says

and see how they go about taking it apart under the pressure of me watching like a hawk. To a one, people who figured it out quickly and enjoyed it turned out to be quality, productive mechanics. Those who struggled for a long time or never figured it out sometimes even with help did NOT make good mechanics.

Mechanics is different there's only one correct bolt head size for any screw.
Any tool but one single intended tool would be the wrong choice.
When I did Flooring I loved performing for people standing over my shoulder. It's how I got $200 tips.
This is different.

What part of "corporate slave" didn't you understand?