[spyckie2]

This makes absolutely no sense.

Taking ideal job traits negative and positive, converting it into correlative personality traits, and guessing the personality trait of the candidate is adding unnecessary abstraction layers with no benefit.

Or to put it another way, just interview the candidate for evidence of the positive and negative traits, don’t do the extra step of translating it into the language of personality theory because the translation is more vague, interpretable, and subjectively biased than the original data.

If you want to use personality traits, it should function as a time saver to quickly approximate candidates potential strengths and weaknesses, which you can hone into with more questioning. It is like alpha beta pruning, used to quickly identify the avenues of further questioning so your time is well spent.

[austin-cheney]

> Taking ideal job traits negative and positive, converting it into correlative personality traits, and guessing the personality trait of the candidate is adding unnecessary abstraction layers with no benefit.

Because that is the wrong approach. Instead, an employer should conduct a personality assessment and keep it as data for later comparison against performance to build a baseline of total average employee personality correlation. It would let the employer know what performance traits to select for in order to maximize for performance in a given role. Unlike an interview a standardized assessment is unbiased when comparing the data between different candidates/employees.

Seriously, if you want to know about a candidate you only need two things:

1. Personality/intelligence assessment

2. Make them write an essay, by hand without a computer.

If they cannot write original words they cannot write original software. They can probably still script together some reused framework bullshit, though, so it depends upon the needs of a given employer.

I wrote about this just earlier today in an unrelated thread: https://news.ycombinator.com/item?id=41241838

[bumby]

Is there data on how stable personality assessments are? I've heard they tend to change depending on one's mood or circumstances. I can also imagine some people are really good at displaying the traits they think someone wants only to have that façade fall off once on the job. It seems like intelligence tests would be much harder to fudge, assuming we agree on what constitutes intelligence.

[austin-cheney]

Personality changes over time due to life experience, practice, trauma, and so forth. It’s not permanently crystallized the way intelligence is.

It’s doesn’t change from day to day though. It is durable over time, semi-crystallized.

That said, the same things cause variance in intelligence assessments would cause variations in personality assessments. That does not indicate a defect in the testing, but rather poor preparation on the part of the tester. Just as the goal is to score as high as possible on an intelligence assessment the goal is to score as honest as possible on a personality assessment, which requires participating under the best conditions.

At any rate the best way to determine test honesty for both intelligence and personality assessments is to retest a candidate 3 weeks later.

I can remember the 4.5 hours of intelligence and personality assessments I took for consideration in full time employment with JSOC. It was the greatest single intensity of mental exhaustion I can remember in a long time.

[klysm]

I can’t tell if you are being serious or not

[austin-cheney]

I wish I were. Every single programming job interview during my 15 years of programming experience either wastes time assessing basic literacy or takes less than 30 minutes to achieve selection.

That basic literacy comes in different forms, like can you read code in your desired language, or other various forms. Even still I have frequently worked with people that struggle to read/write whether in code or especially in natural language.

Let’s cut the bullshit and not dance around it. Just assess it directly. If a person cannot write in natural language they will not have the cognitive competence to write original software. The very best you could hope for is some framework junky that can extend/modify boilerplate.