I get that building a _new_ automated metro is easy. Converting an _older_ line to automatic and fully driverless is what's hard. AFAIK, there's only one conversion so far to full auto: Line 1 in Paris. It took 10 years and billions of euro to deliver (it's a total success though) https://en.wikipedia.org/wiki/Paris_Métro_Line_1#Automation

Interestingly, it seems the main obstacle is software - contractors are few, code is highly specific, ISO certification and warranty issues are very unattractive.

Very interested how TfL will approach this. They're really good at what they do, even at communicating (responsive website!). Don't think anybody does it better.

> That said I wonder how much of it is due to the decks door.

I'm sure they help (if only psychologically), but the Lyon line D doesn't have quay doors and it's been running fully automatically since 1992. The only incident I know of is a drunk person falling on top of the metro (trying to jump from a mezzanine to the quay and missing) and hitting a camera in a tunnel. No death.