Hacker News Reader

Comment by chrisseaton | Hacker News Reader

I don't see how 'hire people who aren't proven' can be compatible with 'can this candidate do the job'. You want to know if they can do the job, but you aren't interested in any proof for that?

View on HN

30-50%(used to be 90%) of any programming job is doing crap you didn't know how to do before.

If they have a proven track record of figuring things out experience is minimally valuable.

> If they have a proven track record of figuring things out experience is minimally valuable.

I can't understand this point of view at all.

When Google wanted to build the world's best JS JIT they didn't hire anyone who was generally able to figure things out - they hired the person with the most experience in building dynamic language JITs in the world. Experience is everything! Experience is knowing which rabbit holes to not go down, knowing who to speak to when you need help, what ideas haven't been tried yet, etc.

I think it's totally unrealistic to compare the average coding gig with writing a custom JIT from scratch for Google. In most businesses the difference between 'very good' and 'world class' won't have a large enough impact to spend the extra resources getting that rockstar.

And then there is the issue of actually retaining top talent...

And what about the other 98% of jobs?

That is only true if you see the list of things a person can learn how to do as a closed set. For instance, I really disliked maths in school. I've avoided every software job posting that involved any maths (3D, statistics, etc). However, if I was given a job offer to work on a math-related product I deeply connected with, and that I felt I wanted to contribute, I'd probably be more interested in learning the math stuff I dumped from my brain years ago. I'm convinced this applies to most people in this industry.

I think the issue we have nowadays is that most software companies are creating mediocre to shitty software that solves no real problem, but selling themselves as the next Google or Apple. Most engineers can see through the bullshit and thus don't feel connected to the product. Then it just becomes a contest for who can realistically write the code for the shitty product for the least pay.

You're suggesting companies build teams of amateurs with the hope that they can all pick up the skills amongst themselves as they go.

If you have to for example build a new operating system kernel, or a new compiler, or implement a machine learning system, hiring a bunch of well-intentioned smart people with the vague idea that they can probably figure it out as they go is a bad idea. You sometimes need someone with a track record of being able to do the work you need done.

That someone didn't come fully formed and ready for OS development out of the womb. It's a good argument for looking at track record when hiring for senior positions though.

I've been interested in many fields, but can't afford to take time to establish myself in them unless there is also significant interest from a company looking to hire me.

The market for new OS kernels or compilers is approaching zero.

Check this real-world example out: https://sockpuppet.org/blog/2015/03/06/the-hiring-post/

and CISO of LinkedIn, Cory Scott: https://www.linkedin.com/pulse/evaluating-technical-talent-t...

You could, instead of relying on a resume to determine if someone can do X, ask them to do some X instead of an interview, and then use an objective rubric to decide if they did X to some level of satisfaction.

This is called work sample testing and it works great.

I think that only works for the most trivial things.

You can do a work sample test for 'can this person add a view to a Rails app', but not for example for 'can this person build a team and develop a solution to a complex technical problem in a specific domain'.

Matasano used to bet the farm on it, I did it at my last job, and Latacora's doing it again. I don't think what those companies do is trivial. (Disclaimer: I'm a Principal at Latacora.)

In my experience, when you think you can't measure a thing in a work-sample setting, it means you haven't analyzed it. And, if you have no idea what that thing means, how could you possibly hope to interview for it?

For example: "can this person build a team and develop a solution to a complex technical solution" (that's several things, but OK, let's take a stab at it):

- Someone who can build a solution in a specific domain has to be able to analyze the problem into bite-size pieces. Given a complex problem description, write out tickets that need to get done. Can they identify sensible milestones, objectives and key results? Are the tickets roughly equally sized? Are the tickets self-contained? Do they come with an objective measure of "done"?

- Someone who can build a team to deliver a solution has to be able to estimate appropriate resourcing for a project. Given a complex problem description and a current set of resources, come up with a plan forward. Measurements include: a) Did they identify missing roles? b) Did they write clear job reqs with evaluation and hiring plans? A gazillion hiring managers couldn't write a clear job req to save their lives. c) Did they consider which positions may be effectively contracted out? d) Can they estimate what the P&L impact of this hiring plan is? Knowing how to read P&Ls and cost of development is a skill. e) Can they identify within their own plan which roles are really critical and which ones are nice-to-have? Being able to negotiate in the face of limited resources is a skill.

- Someone who can lead a team to deliver a solution to a complex problem has to be able to analyze when things go awry on a low-level. Put them in front of a PR where someone has subtly misunderstood a poorly-written ticket. Give them the context they need to understand why the PR is wrong. See how they tell people the PR is wrong, and how they react when people persist?

- Put them in a position where they have to talk to an employee who isn't doing well. Can they figure out how to be empathic while remaining professional? Did they say something that will get you sued? Opposite situation: put them in a room with a rockstar developer who's been harassing or badgering one of your employees. Can they speak to that person professional? Can they write an HR file note with a follow-up plan?

On the contrary, it works for substantial jobs. I used it for hiring security engineers with great success. I then used it for project manager and manager of compliance, and it worked with astonishing success. In particular, the compliance manager job is _mostly_ about getting results from people across the organization that do not report to them.

Folks have built highly specialized company teams with this and little or no interviews.

Reply on news.ycombinator.com