School Project
 

From 'Prober', one time P1 on VC9, HS25, B75/76 TRI.
So long retired I now no longer even get staff travel!
A grandson has a school project on the future of travel and he proposes examining the likelihood of pilotless aircraft.
What I am seeking is a reasonably reputable source for statistics of probable human causes versus mechanical causes of airframes being bent or destroyed over the past maybe fifty years. Don't worry, I am not a traitor to the cause - I will ensure that the last comment will concern the length of the queue to get on to such an aircraft.
Any help would be much appreciated.
Prober

I would try have a look at US NTSB website or your national equivalent. Perhaps ICAO website has some info.

What exactly does he expect those statistics to show? All aircraft have been piloted by humans for the last 100+ years, so obviously there will be a lot of crashes due to human error. That doesn't mean that machines would have done it better.

School Project
 

Thanks, T54A and I will make a start there. The angle which I feel should be highlighted is not so much the disasters - we all hear about those, but it's the ones we do not hear about - the disasters averted because there actually WAS a pilot there. How the boy will do this, time will tell.
Prober

Prober, the links below might be a bit heavy for a school project. The debate is open, continuous; and without solution until it happens!

There could be greater value in debating the issues, man or machine, or more likely man and machine where the machine has greater emphasis on computational capability.

Statistics have little meaning without context.

There may be value in finding the very old ICAO diagram (unscaled) of how the primary causes of accidents have change from technology to the human; but again beware context, subjectivity. The need of a systems view.
Then consider that diagram squashed into recent the safety statistics, and compared with the last 20 years such that whatever the relative cause, there are very few of accidents with cause, man or machine.

Airbus publishes a useful review, https://accidentstats.airbus.com/sit...-1958-2020.pdf page 17 shows the proportional factor reduction of 10 for all accidents, thus technical or human components (ICAO digram), are relatively very very small. Beware the law of small numbers (fallacy) we tend to make more of small values even though not justified. The analysis also considers technological advances, new generation aircraft.

Why is the low safety value flattening out, can this ever be reduced, is this due to the human or the wider operational system? i.e. there will always be accidents even without the human, but will the rate be any better than with the human who contributes so many hidden ‘saves’ in the much larger area above the safety line.

A good subject to debate, but not to conclude.

Re the issues we don't hear about, because there are no reports, then a better guide is to search for articles by Hollnagel on looking for things that go right (many refs, ppt simpler ones)

A recent HF view is at https://aviation.vision/wp-content/u...hite-Paper.pdf
Note than the majority of views focus on AI and consider today's problems, not really forward thinking. The one item on IA (Intelligent Assistnce) is more likely, but it still involves the human.

Also;
https://static1.squarespace.com/stat...ech-Report.pdf

https://safety177496371.wordpress.co...ritime-safety/

Thank you all, especially alf/h for the detailed response. I agree, no conclusion can really be reached - the size of the queue rushing to get on will settle the argument and we can leave the thread there.
A grateful Prober

I overlooked two important reports:-

https://web.archive.org/web/20200717...Automatica.pdf

and 30 yrs later

http://johnrooksby.org/papers/ECCE20...er_ironies.pdf

and 30 yrs hence

…

About 12 or 13 years ago a Lufthansa led survey by a German university looked at in house data from the company pilots regarding the last time they suffered confusion/error/incident. Lufthansa made it err, rewarding to take part. Approx. 3000 responses came in.

University came up with a pretty standard pie chart of incident analysis but they also added a new chunk of pie - Technical/Human which I suspect ties in with Alf's observations. It proved a sizeable chunk but with a single airline group looked at not huge numbers.

Rob