Search engines have been dropping in quality significantly within the past decade, and especially within this past year. The noise to signal ratio has been frankly painful.
Can you please share some resources you use when trying to find answers to technical questions?
For example, STEM, academia, engineering, programming, etc.
most University Libraries have “guides”. Simply google " guide". eg Stanford Library engineering guide. Result like: [(https://guides.library.stanford.edu/aa)] From there use any free local library access you might have to get the details.
That’s not how links work in Markdown; some clients will include the
)]at the end and break the link. Just use the plain URL https://guides.library.stanford.edu/aa or create a hyperlink using[this syntax](https://www.markdownguide.org/cheat-sheet/).Thank you so much!
When doing technical things, I find the best source to always be the provided documentation. For example, when using an external crate in Rust, docs.rs or when coding a Django Webapp the official Django documentation.
When starting out, these often contain examples or guides/tutorials.
When that does not help, it goes back to putting relevant keywords into the search engine and hoping for the best.
Thank you all for your answers!
I wanted to add one resource I found that has helped me find even more relevant search results:
A Lemmy Search Engine https://www.search-lemmy.com/
For all of those topics, I use domain specific sites. So for research I’ll look at arxiv or one of the sites that make research freely available. For programming, I’ll search language mailing lists, documentation, and examples. Searching GitHub also isn’t a bad idea, but watch out for license issues.
Be wary of using tools like got to summarize articles or outright answer questions. There’s no guarantee it will be correct, and if you don’t know the answer you won’t know it’s wrong.
Wikipedia is pretty good for computer sciency stuff. I’ll often use it as a reference for things like protocols or if I need a quick refresher for some algorithm.
deleted by creator
For code I use chagpt for first pass questions. Then I try compiling it and see if gpt is telling the truth
I feel like this is a risky approach. LLMs are designed to spit out text that sounds good, that’s all. If it hallucinates important info away, your compiler will not always tell you.
Yeah, I check stuff with stackoverflow and the documentation
