Azhir Mahmood  🚀 🤖 🧬 🧠 🪐

As someone who's both writing plenty of recommendations and evaluating applications at UCL, I fully get what you're saying, but this strikes me as.. not the worst? A number up to 10 seems a lot better than the "top 2% / top 5%" that seems so common. What would you suggest?

Ask me some of those questions that are relevant to working with the candidate - just not the ones where the transcript is more accurate!

Is the transcript really accurate at conveying e.g. originality, motivation? I'd agree with doing away with these questions entirely, but for a different reason, i.e. they're totally uncalibrated (as a panelists, I get so many 9s and 10s as to be mostly useless).

There is also the everyone gets a high score issue that you've highlighted which makes these forms pointless. I'd really love to see the aggregate scores for these things over a year!

It's certainly not a Gaussian distribution centered around 5 :)