So, what are you measuring? I don't get it. You made offers mostly to people with few typos in their resume. Is this evidence that people with few typos are good, or evidence that you have some kind of bias against resumes with typos?
I find the action words particularly disturbing. Every resume that crosses my desk has people that 'drove' a 'process', and they 'innovated', and so on. Buzzword crap. I'd bet my company on somebody that "implement an image algorithm" (all bad words apparently) over somebody that "enabled client connectivity" any day (all good words).
And what is the actual relevance of worked for a big company? Again, I suspect bias on your part. Explain how working for Microsoft is a good predictor for how a person will perform in a 10 person startup?
There's a difference between action buzzwords and clear explanations of what you did. "Drove a process to innovate that led to massive KPI improvements" is very buzzwordy but not at all clear. :-)
If anything I think bias would lead to lower-than-expected scores, as more resumes would end up in the "to interview" pool. But it would be good to validate this against the rejected resumes.
I doubt most of the y combinator startups count as "big" companies, and he said he included all of them. He was trying to model "successfully held down a high-expectations technical job" and it seems like he chose a reasonably good heuristic.
1. So, once someone was "in" (as in, they got to interview stage), we didn't consider any of these factors. Everything after getting into the interview process was purely about interview performance. As it happened, people we made offers to tended to have less typos than people we didn't.
2. Keep in mind that these action words were words that showed up a lot on resumes of people we made offers to and not a lot on resumes of people we didn't make offers to. We didn't look for those words, and again, as with the previous point, once someone was in the interview process, how they worded things on their resume wasn't a factor. This is something I observed when counting word frequencies after the fact.
3. It's not about company size. I looked at where people who were applying to TrialPay were coming from, and of those companies, I chose the ones that are known for having a high hiring bar independent of company size. In addition to some of the big guys, Y Combinator companies were also on the list.
This is awesome! The biggest threat to validity I see is that your process of filtering from 10 resumes to 1 interview could introduce a lot of bias; if people ignore other warning signs that would weed out the resume because an applicant went to a top school, then you're going to end up with more stinkers with that attribute in your interview pool and undervaluing it in your final analysis.
I would not suggest that you go through and rate all the unaccepted resumes on the same scale and then regress against both "got an interview" and "got an offer" since that seems like a ton of work, but I think it would be the best way to resolve the threat. :-)
Also, the stuff about MS degrees totally tracks with my subjective experience as a hiring manager - it's not a red flag exactly but I don't think I've ever been wowed by a person with an MS on their resume.
(I'm sure they do exist, don't hate me people with them!)
I was lead and I screen people with masters. They all failed. I swear they went back to school for master because they couldn't get a job.
One of them was an Indian dude.
So what's your focus?
"Database"
Neat what's your favorite relational database?
"All of them."
.... Ok, what do you think about NoSQL?
"They're neat."
What's your favorite?
"All of them."
In my head, I'm like what the fuck. Did this guy tried ALL of the NoSQL and all of the relational database?
Anyway, the dude was pretty bad, no real skills just a piece of paper. Couldn't even solve simple program puzzles (they're just Project Euler problems).
I met another guy like this (who happened to be Indian too), when I was interning at a big 3-letter named company.
So, what are you focusing on for your grad studies?
"Quantum computing"
Oh neat, what exactly are you doing?
"Some quantum stuff, blah blah blah"
I don't remember the exact answer, it was a while ago. But it was clearly just some buzz words that did not contain any substance; and that he had no idea what was going on.
There is also another Indian guy at the company I presently work for. Fresh grad with an MS degree. We were writing a small internal web service, and he was all over the place about how "we must use a NoSQL data store", because "scalability", and "flexible and dynamic, no schema" etc.
Thankfully, the final call was not his, and we went with a plain RDBMS -- which naturally suited our purpose pretty well. We had no need for NoSQL just for the sake of using it.
Disclaimer, I hold an MS degree. I was privileged enough to get admitted to a top Canadian school with an excellent CS program. Almost all the people I met during my stay there were brilliant, and definitely did not go back to grad school because they were not able to get a job. Many of them did internships at top companies; and a couple of them even came back from the industry 20+ years later, because they wanted to work on something different. So definitely do not clump all MS/PhD degree holders together. I think the author here is making the classic correlation vs causation mistake.
Sounds like we aren't all crazy. :-) Also, the other posts on the blog (at least that I have read so far) are definitely worth your time if you liked this one.
This phrase is the Godwin's law of discussing data. :-) As a hiring manager, I don't care about causation; I just want to know what aspects of a resume actually correlate with someone being competent (or not) so I can waste less time on pointless interviews.
I find the action words particularly disturbing. Every resume that crosses my desk has people that 'drove' a 'process', and they 'innovated', and so on. Buzzword crap. I'd bet my company on somebody that "implement an image algorithm" (all bad words apparently) over somebody that "enabled client connectivity" any day (all good words).
And what is the actual relevance of worked for a big company? Again, I suspect bias on your part. Explain how working for Microsoft is a good predictor for how a person will perform in a 10 person startup?