[kozikow]

I feel like algorithmic interviews are starting to be a red herring. When we are reaching a point that people need months of preparation and special interviewing bootcamps, we start to experience something akin to Goodhart's law. Correlation with real world skills get smaller and smaller, as people stop honing their "real world" skills and focus their energy on gaming the artificial metric.

Many companies are starting to value good github profile more than "how quickly are you going to implement the dijkstra algorithm on whiteboard". I believe this trend is better for candidates and companies, so it will continue.

I am speaking it from the point of someone fairly proficient in algorithmic challenges and algorithmic interviews (e.g. top 100 google code jam, worked/interned at 3 out of big 4).

[conflicted_dev]

Frankly, I agree, but I also don't see another alternative. I have a reasonably decent github portfolio, but most companies I've applied to view it as supplementary evidence, rather than proof of my abilities.

[vorg]

> Many companies are starting to value good github profile

Expect an industry to grow around servicing the github accounts of US-based programmers to make it look like they're doing side projects and learning certain skills. In reality, it'll be India-based college grads doing the github updates to a specified schedule -- the same college grads who presently do course assignments and online coursework for US-based students.

[kozikow]

Maybe with plethora of online info, we could implement reasonable fraud detection?

  - Detect copy-pasta code
  - too inconsistent language style (either code or English)
  - prepare interviews questions about github profile (e.g. What made you choose library X for project Y)