OK I was wrong about VSCode
For the past 3-4 years, students in my undergraduate Natural Language Processing (NLP) course were asking if they could use VSCode for their projects. I warned them that it may not be compatible with some of the libraries we were going to be using but if they wanted to try that’s fine.
I’m not a dogmatic person. I’m not going to tell somebody that they can’t use a tool they like. However, I had some outdated ideas about VSCode in particular and Microsoft in general.
Let me back up and say that years ago, those of us in the technical community turned our noses up at all things Microsoft. We found their products to be bloated, slow, and just not cool. We preferred our Linux or Mac machines with open source software. For example, during my PhD work in the mid-2010s we were a Linux lab with a few Macs. After my PhD, when I started working at UTD, they issued me a Dell laptop piece-of-sh** that came installed with Visual Studio, as it was called at the time. That software took up so much room on my computer that I had to delete it just to make room for the files I needed to teach my classes.
VSCode rises
Fast forward a few years and students were telling me they liked what was now called VSCode. Around this time we went online for the first covid shutdown. I tried teaching online with Microsoft Teams, but it just did not play nicely with my Mac. Another tick in the anti-Microsoft column for me. However, I have to say that Microsoft stepped up their game quickly and got it running quite well for Mac. So +1 Microsoft.
Microsoft Research
Speaking of Microsoft upping their game, while I was working on my PhD with an emphasis on natural language processing, I had much respect for Microsoft Research. I would see their people at conferences where I was presenting my work. They released some great NLP tools. So I had a strong positive feeling about the research arm of Microsoft, even as I dreaded having to use their products.
Microsoft Research must have taken some steroids. Their investment in OpenAI was brilliant. The integration of large language models (LLMs) with search apparently was enough for a code red to be issued at Google. Microsoft deserves to get a bigger market share for being there first. In June 2023, I got access to Google Lab, which is their version of incorporating LLMs with search and I’ve been impressed.
As I said, I heard students say now and again that they like VSCode. I listen to Lex Fridman’s podcasts and he started using VSCode, so I think, well, maybe it’s time to open my mind and give it a try.
After finally giving it a go, my general impressions are that it is sleek, fast, and helpful.
So, to my former students who asked me about using VSCode and I had an inner (hopefully not outer) eye roll. I apologize. You were on to something.
