Ever tasted popping candy? It's this powder-like candy that *pops* when you put it in your mouth. I've never craved it or anything, but whenever I see it in a store it reminds me of just how sensational it felt for the first time.

And somehow, graphs do the same for me. I imagine a bunch of my neurons sparkling when I read an article about some new development in graph-related technology ✨ It's very much like what happens to Remy in Ratatouille here.

[a graph?]

No, not those kinds of graphs. I meant these.

[cooler graph]

Looking back, I realize I even wrote my Extended Essay on graphs. It was about seeing how well NLP could be used to create knowledge graphs from reading Wikipedia abstracts. This was my cover page.

My interest in graphs got serious when I discovered Graph Neural Networks – something about applying deep learning to graphs just fascinated me. I read a lot about it, but quite frankly, I still don't understand how it works haha.

But that was until yesterday, when I found Fireship's video on my feed.

It caught my attention when he mentioned "graph-relational database" (4:46). Apparently there was this new database called EdgeDB? I had to check it out.

*sParkLE moMeNt*

I had to give up on my English Paper 2 prep (which I had a practice exam on the next day) and stay up until 1 AM to get EdgeDB finally working on my laptop. If you search up EdgeDB right now you'll see why it took me so long – there's only 1 tutorial page on how to set it up, not even videos! I had to try disabling my firewall and switching to my phone's 4G to get it to work.


Well, today is that next day, and welp, I had to say goodbye to my Paper 2 haha! But on the flip side, I felt like EdgeDB has so much potential to be the next big thing for databases, and I'm honestly surprised I'm one of the few to realize that in its early stages.

If you look at these three people who tried out EdgeDB, all of their reviews/impressions are incredibly positive and honestly quite promising.

EdgeDB | Everything I know
My DB of choice for all projects. Sits on top of Postgres but has insanely well made query language & declarative schema modeling that is a breeze to edit. Nearly all queries are easy to construct & are super fast without doing any optimizations.
My experience with EdgeDB · divan’s blog

I ran through their quickstart guide and read their textbook, and so far I'm loving it. The real deal for me was their link properties feature, which I thought was perfect for the recommendation system I wanted to build with GNN (I'll write another blog post on that).

Right now, I'm just tinkering with EdgeDB to see how far I can go with it. If you ever get interested in trying it out too, do let me know!


UPDATE (April 16)
It seems like EdgeDB was backed by OpenAI... the company that made ChatGPT👀

OpenAI’s Bets on Tech Startups Hint At the Future of Artificial Intelligence
OpenAI’s $100 million Startup Fund and Converge accelerator program have backed about a dozen startups that give a hint at what its next product might be.

EdgeDB