Okay, that one went well, at least for me.

I have eight jobs noted in my notebook, though I'm pretty sure I missed recording one or two that I can't remember off the top of my head. Even at this, the last five I had to put down later as it seemed I didn't have enough time between clients to take notes...

But out of that, I only had one item I couldn't repair. It was a #Sony bookshelf stereo-type system, the kind where it's basically a cube 30-40cm on a side, containing the amp, tape decks, tuner, CD player (or in this case, multiple-disc changer).

The reported problem was that the CD tray would not eject. Pressing the button did nothing, although the display would indicate that it was either opening or already open (not sure which).

Taking it apart as much as I could, I got to the point where the disc mechanism could not be removed because it was held in place by the front fascia (the fancy "door" on the front of the CD drawer mechanism), and fascia appeared to be attached to the drawer by screws or clips only accessible when the drawer is open. #Catch22 Sony over-complicate everything.

A highlight was fixing a talking stuffed #elephant that had stopped making sound. Small child very happy.

The high success rate, I think, is down to very few of the items I worked on being computer (#microcontroller)-based. Those black-box boards dying with no easy way to diagnose or repair the faults are the bane of our existence.

#MCU #BlackBox #board

2/

The commentary on the first point:

"most of the tools available were developed by #private companies. This is problematic for #reviews that evaluate drugs and medical devices, because these need to be #independent of industry. What’s more, few AI models are open source, with most relying on opaque, proprietary ‘#BlackBox’ processes. This means there’s no way to examine whether a tool might disproportionately include trials with results favourable to one drug company"