Insistence without sound arguments and relevant evidence is not rational discourse. Defend fiercely with reason and heart, but assent eagerly when proven otherwise.
@AgileJebrim Is Intel's best-case latency better than AMD's? If yes, then I suppose they serve slightly different use cases. It's a trade-off that engineers should definitely be aware of.
@notsky Identify what you are optimizing for:
- Money (popular ones or those with plenty of high-paying jobs)
- GPA (easier-to-score ones)
- Understanding (harder ones that teach low-level concepts from first principles)
- Interest
Make a weighted linear model of those features.
@yacineMTB Those who build any one of those systems instead of only "knowing about" all of them would go further in their careers. Show me someone who builds a toy compiler, DBMS, OS kernel, or DFS, not someone who can recite the definitions of ACID, CDN, LRU, and NP-hardness.
@abhi9u Just as an additional note, these graphs show average lookup performance. There might be applications where the worst-case performance matters. For example, cuckoo hashing has guaranteed constant-time worst-case lookup (used by Tiktok's recommendation system).
@AdamRackis I do recognize that this is a chicken-and-egg problem though. To get competent interviewers, you need to recruit competent engineers. Less competent interviewers might filter out more competent ones. Maintaining the quality of engineers requires constant effort from the start.
@AdamRackis There are also folks from these "good places" who do not fully understand these concepts because they do not use them often and their expertise is on something else. If they become interviewers and misevaluated candidates, is that fair?
@AdamRackis Both roles might eventually want to implement Tarjan's algorithm for some unique use cases. That's where the "continuous learning and being adaptive" part comes into play. However, it should not be an interview question for those roles.
@AdamRackis A technical interview for an embedded SW role should differ from a web front-end one. The former should ask about interrupts, while the latter should ask about DOM manipulation. If both ask about Tarjan's algorithm instead, something is wrong.
@abhi9u I saw the summary tweet in my timeline. Reading the full article might take a couple of sessions across several days. The level of detail requires some time for my brain to process fully. 😅
I also wanted to say that I truly appreciate your work. Please do keep it up.