Flight incident: Passengers titled ‘Miss’ counted as children [pdf]
assets.publishing.service.gov.uk
assets.publishing.service.gov.uk
Surely the solution is to fix the software? Miss is the 'appropriate' title for an unmarried woman or girl. Why not have a person select 'adult' or 'child' when booking (I'm pretty sure most airlines I fly with do this already).
Pure speculation on my part, but it may be a data-hiding issue - if they realized late in design that the passenger manifest system had to feed into the weight estimator, but the passenger manifest (for privacy reasons, or simple "Don't access more data than you need" coding best-practices reasons) only had access to passenger names, surnames, prefixes, and IDs.
It shouldn't have happened (clearly, that's the point an engineer should be pulling the red lever on schedule and doing the work to get the data that's needed to the service that needs it, including clearing legal hurdles for privacy, if necessary), but I can easily imagine it happening if quality took a back-seat to deadlines.
"The incident occurred due to a simple flaw in the programming of the IT system, which was due to the meaning of the title ‘Miss’ being interpreted by the system as a child and not an adult female. This was because in the country where the system was programmed, Miss is a child and Ms is an adult female."
Not sure what the country in question is - the report doesn't name it.
> The system programming was not carried out in the UK, and in the country where it was performed the title Miss was used for a child, and Ms for an adult female, hence the error.
"Cultural differences" is an easy scapegoat.
So if you’re presented with Ms. and Miss as options in a form, and neither has to do with marriage status, it’s natural that Miss will refer to the young.
She used "Miss" to say unmarried female. Her ticket was for unaccompanied minor.
Miss is the one specific to unmarried women.
I also never said we should or shouldn't use the terms. I don't have an opinion on that and it's not a debate or argument I'm going to waste my time on. I was simply explaining how they are conventionally used. This kind of thing was taught in schools as recently as 20 years ago as part of explanations on how to write and format letters.
Edit:
It seems using it until the age of 12 is a convention in the US. Using it until 18 is the convention in the UK. Another argument for why software shouldn't make decisions using terms that can mean different things to different people :)
Not anymore, that is. It used to be an abbreviation for two appellations, one of which was associated with an unmarried gentleman. It's rarely used outside of the British Commonwealth, anymore. Except on Fresh Prince, of course.
I got this impression from my father. When I was little I would help stamp and address sales mailers he would send out, and he always used Ms. When I asked him, he said start with Ms. and let them correct you if they want something different. But, some divorced women get angry at being referred to as Mrs.
Not since decades in the US at least. Women are not titled according to their marital status.
https://www.theguardian.com/world/2007/jun/29/gender.uk
> Miss and Mrs are marks of the old world, reminders of women's second-class status as wives-to-be (Miss) or simply wives (Mrs). If you are a woman who doesn't use Ms - particularly a woman under 30 who has never even thought of it - then ponder this: how do you want to present yourself to the world? Are you an appendage or an appendage-in-waiting? Don't be branded and marked by old-world convention. Let's kick against those fools at companies such as Atlantic Data. Let's put two fingers up to employers and bureaucrats who want to define us by our marital status. Choose Miss and you are condemned to childish immaturity. Choose Mrs and be condemned as some guy's chattel. Choose Ms and you become an adult woman in charge of your whole life.
However, HN is overwhelmingly male (and threads like these illustrate why it is likely to stay that way), so what is obvious elsewhere may not be obvious here.
It seems to me that the passengers themselves were selecting their own title.
I still say sir, mam, mister, and miss (Mrs is generally considered unneeded and overly formal) because I come from a culture that finds them not only to be appropriate but pretty standard, especially when dealing with strangers. You find that declaring them universally acceptable is wrong, but if you came to my state you'd also be equally wrong, and maybe rude.
I do understand that peoples attitudes over generations will change about words. That seems to be the natural evolution of languages, but it also seems we've never overtly adopted a system that deprecates language outside of outrage, which seems inappropriate in itself if you're getting angry with someone who is going out of their way, in their mind, to treat you with respect.
My family has an interesting habit of addressing mail to my children by these titles. I don't know why or when they started it, but I myself, was never addressed as Master, yet my son is.
> Definition of miss (Entry 3 of 4) 1 capitalized a —used as a title prefixed to the name of an unmarried woman or girl
> The system programming was not carried out in the UK, and in the country where it was performed the title Miss was used for a child, and Ms for an adult female, hence the error. A manual solution for correcting the problem was quickly identified that involved a team identifying upcoming flights, checking each booking, and changing all adult females with the title ‘Miss’ to ‘Ms’, which overcame the problem. Subsequently, this work was shared between two teams, and the process was completed every afternoon and evening for the next day’s flights. It was checked again every morning, where possible, before flights departed...
> The upgrade programmers adapted a piece of software, which changed the title of any adult female from Miss to Ms automatically, and this was implemented on 17 July 2020. This adaptation was only capable of changing bookings before check-in. Any passenger bookings with the title Miss already checked in, including online up to 24 hours before departure, would not be amended. On 20 July 2020, the programmer was making enhancements to the program to improve its performance. This should not have stopped the program from working, but as this was a ‘fix’, it could not be known for sure.
This serious incident is a great example of that process in action. The initial issue they were investigating was that the flight gets computer-analyzed to compare the actual dynamics of the plane to the theoretical dynamics. That analysis revealed a 1kt difference between what the takeoff speed should have been and what it was measured to be (about a 0.76% deviation from expectation).
That <1% deviation led to an investigation and a 7-page report.
Such a level of scrutiny is expensive, but it's how commercial air travel has gotten to the "safer than cars" level it currently supports.
Either way, a bug that results in an aircraft weighing 1,244 kg more than expected is indeed deadly serious. I write plenty of bugs, but luckily worst case is usually a web page failing to load rather than hundreds of people falling out of the sky. Our development process is clearly not adequate to this domain.
An adequate process for this would necessarily slow down development greatly and make it far more expensive. It's so much easier and cheaper to cut corners.
This situation, though, where input data to those systems is bad before it even gets there, is a loophole of sorts.
Somewhat related, this makes me worry a bit about all the data privacy regulation that keeps you from testing (easily, at least) with real production data. Good testing from the W&B software might have unearthed the lower overall weights per aircraft.
Seen this a couple times in articles about Boeing.
Another article had a title about how an airline cancelled all their Boeing orders.
Reality, they cancelled all their Airbus orders too (clearly stated in the article) ... and the airline reiterated in the article would continue just flying Boeing planes and in the future planned to buy a different model of Boeing instead of the one they canceled.
But the title implied something different, and the comments were filled about how this was a big deal for Airbus... and the typical tribal riffing off the title.
Which means the weight is a ballpark anyhow. They aren't weighing anyone until totalled when on the plane.
Would they blame overweight people, if a bunch of over average people got on, and skewed that??
edit: just re-skimmed the article, I still don't get how a statistically averaged weight, which can be wildly variable for specific flights, causes issue because it was wrong.
Am I missing something? Is the average supposed to mostly be OK, thus not requiring fuel changes and delays for almost all flights, sav the outliers?
Just to pull some numbers out of, ah, thin air for an example, if you were using 110lbs as the average weight of a child and 160lbs as the average weight of an adult woman, your airplane seats about 200, and roughly one-third of your passengers are adult women, half of whom use "miss" as their title, that works out to 50 lbs difference * ~30 passengers = 1500lbs.
That's gonna make a big difference.
This seems to be in the UK, so what if an all female travel group, from the US, for the portly, showed up? When I visited the US, there seemed to be an inordinate amount of people in scooters that were quite.. large.
At least, this is what was in my head.
Then again, I just realised... do they have to buy two seats? Maybe I'm over analyzing all of this.
I would also point out, though, that it's just as likely that there would be something like a gymnastics or other sports team on a plane, which I think should be somewhat under the average weight (despite the extra muscle mass).
The root cause was that the system assumed "miss" = "child". So it's a case of a mislabelled category.
The real root cause was a mistranslation: "in the country [India?] where [system programming] was performed the title Miss was used for a child, and Ms for an adult female".
a) the assumption that country in question was likely English-speaking (which increases the risk of translation mistakes, due to the "false friend" effect)
b) the fact that India is a powerhouse when it comes to offshoring software development.
If it's not the case, well, I wonder if someone is able to identify the country? Out of sheer linguistic curiosity
The whole thing is just bizarre - I guess it was outsourced to some chopshop.
The reason that the mislabeled category was a problem was that the system used those cultural honorifics to estimate takeoff weight.
The reason takeoff weight was estimated using those categories, rather than from more accurate birthdate data, or calculated from actual weight data, was...
(The actual details of their architecture I'm not privy to).
If you have a flight which is performance-limited, they will limit baggage hold (checked) bags which are weighed and force you to carry that same bag into the cabin (where it is not weighed and added to the performance calculations).
I seem to recall that there are specific weights used for aircraft where the passengers are overwhelmingly different from typical adults (think of an NFL football team flying to/from a game), but that general variances among passengers do not need adjustments to performance.
To my knowledge no airline accounts for general variances among passengers leaving Mississippi as opposed to, say, Colorado. The standard passenger weights do change in the winter vs the summer to account for heavier clothing.
"As a result, the aircraft departed with a takeoff weight 1,244 kg more than stated on the load sheet."
If 50 women weigh 94 kg instead of the assumed 69 kg, that difference is also reached.
I'm surprised that the 737 has such a low error tolerance.
In a sense, it's not too terribly low; the end result here was that the plane took off outside of the correct flight envelope but well within a safe and controllable envelope (given a loading difference that - back of the napkin calculation - is 2% of its dry weight).
But AAIB pursues this sort of thing aggressively, under the principle of "Don't assume everything is safe because the O-rings didn't fail if the O-rings are showing not-understood ablation."
The performance calcs are done to ensure safety in a one-engine inoperative case.
The performance calculations for all multi-engine turbojets require at least a balanced field (simplified meaning: that you can experience an engine failure at the most adverse moment in the takeoff sequence and either bring the aircraft to a halt on the runway surface [including EMAS or other stopway surface] or you can continue the takeoff and meet initial obstacle clearance and second stage climb requirements with the most adverse engine inoperative and maximum thrust from the working engine).
That's why this is a serious deviation, even though turbofan engines rarely fail. 737s do reduced power takeoffs as a matter of economic optimization. The error here caused too little power to be commanded during the takeoff.
A few overweight passengers can't do that, but a combo of too many passengers and etc can have an impact.
TLDR: 5481 crashed because the figure used as the average weight of a person for the purposes of calculating the planes centre of gravity was from before America got fat.
Edit: It usually means a disaster when people try to read "between the lines". One favourite example is trying to infer purpose/role of the server based on its hostname. When hostname starts with "X" it means only one thing - the first character of the hostname is "X", no more, no less.
That approach is a lesson...and then the software guys turn up to fix their bug report:
upgrade programmers adapted a piece of software, which changed the title of any adult female from Miss to Ms automatically
Note this didn’t help the flight in question, because their data had already been added in advance of the fix (due to online check in).
It’s interesting that the issue was known about in advance of the incident flight, but perhaps not the significance if it happened en masse. On the face of it this just doesn’t sound like a fix that addresses the root cause. There are clearly internationalisation issues at play, along with entity design issues, and UI issues. So the issue lurks in the system, ready to form part of some future accident chain; the report is a page too short.
> The upgrade programmers adapted a piece of software, which changed the title of any adult female from Miss to Ms automatically, and this was implemented on 17 July 2020. This adaptation was only capable of changing bookings before check-in. Any passenger bookings with the title Miss already checked in, including online up to 24 hours before departure, would not be amended. On 20 July 2020, the programmer was making enhancements to the program to improve its performance. This should not have stopped the program from working, but as this was a ‘fix’, it could not be known for sure.
I thought that given the complexity of trim in modern airliners with massive fuel-tanks on long hauls, etc. that aircraft would include integrated weight / balance sensors.
I'm disappointed that the AAIB did not identify this.
If you need to identify the passenger as child/adult (especially for something as safety critical as this) then it should be its own field.
It feels like something is missing from this story.
It sounds terribly old fashioned though.
Based on the article, it seems like they only use that to determine how much engine thrust is needed for takeoff, so it doesn't seem they need to know the weight in advance.
We have the technology to weigh aircraft. But they'd either be onboard, and thus safety critical and subject to a lot of abuse (being inline with the landing gear presumably), or they could be on the ground/at the airport, which means every airport in the world needs them, and they need to be compatible with every aircraft's footprint (i.e. huge or highly adjustable).
Plus the time cost/queuing for getting weighed. Then adding fuel after (since fuel is based somewhat on weight).
But that's probably the issue here as well, compatibility with software systems going back 30, probably more years.
Basically you apply some thrust to the engine and look at the resulting acceleration with an accelerometer.
Knowing the remaining length of the runway one can compute the acceleration that produces takeoff velocity at the end of the runway minus a safety margin, and then change engine thrust by the ratio between that desired acceleration and the measured acceleration.
The remaining length of the runway can be computed by knowing the total length, initial position and integrating the accelerometer data, or by using GPS knowing the coordinates of the runway ends.
This also automatically handles factors like wind, variable friction on the runway and miscalibrated engine thrust.
In this case we seem to be talking about a heuristic for estimating weight, and heuristics are kind of a weird middle ground on that score. If all you need is an approximate value and can compensate for errors, it's reasonable to use a properly-designed heuristic rather than an explicit value.
That said, a lot about this story smells weird. They already had an explicit value in the age field, and this particular heuristic is completely useless for male passengers. So it's less about proper programming practice than the fact that somebody is just plain doing something stupid. So stupid that I'm not sure we're actually getting the story correct.
Which make you wonder why they still use the estimated weight when they can have the actual one.
To use actual weight when individual passengers weights are unavailable (as in this case) would require first boarding all passengers, then waiting at the gate while the fuel tanks are filled and the cargo in the hold is loaded and distributed based on the actual passenger weight. The passengers would not want to wait for this loading order, nor would the airlines as it would directly increase their turn around time for the aircraft.
Surprised this classifies as a serious incident.
> As part of the prestart procedure, the flight crew reviewed the flight plan, which gave an expected takeoff weight (TOW) of 66,495 kg (Figure 1), and the load sheet, which gave a TOW of 64,889 kg (Figure 2). They noticed that there was a discrepancy, with the load sheet showing 1,606 kg less than the flight plan. They noted that the number of children shown on the load sheet was higher than expected, at 65, compared to the 29 which were expected on the flight plan. The commander recalled thinking that the number was high but plausible; he had experienced changing loads on the run-up to the temporary grounding as passengers cancelled and altered trips at short notice.
If the weight estimate is too low, then it would be possible for the ground crew to load too little fuel, leading to a situation at the destination of nearly or actually running out of fuel.
So it may have been labeled "serious" due to the possibility of not loading enough fuel if the weight estimates are too low by too much.
For example if the aeroplane had been on a relatively short runway, and its passengers consisted entirely of overweight women who were miss-classified as much lighter children, perhaps it would have struggled to clear the runway at calculated power.
Flight crews are not trained to attempt to overrule the computer by explicitly commanding maximum thrust because humans are bad at figuring out if they're accelerating as much as expected - and so in the absence of a check that the plane is performing as anticipated † they are going to have a bad time if the calculations are badly enough wrong.
Why not just always use maximum thrust? Well if you're an aircraft carrier, or maybe even a military base you could definitely do that, but if you're a civilian airfield then the tremendous noise from jet engines at maximum performance is a constant problem and so you want jets to calculate a lower take off power that's still safe in order to reduce noise.
† [in some newer Airbus configurations the plane can work out its real acceleration and which runway you're on and conclude whether it will not successfully take off before it runs out of runway and so caution the pilots to abort before V1 when it's still practical - it doesn't matter why it isn't accelerating what matters is that you won't fly if you don't go fast enough, they can spend as long as they like to troubleshoot when they have stopped]
And...
"Whilst an incorrect takeoff weight was used for aircraft performance planning, the thrust required for the actual TOW and environmental conditions (88.2% N1) was marginally less than the thrust used for the takeoff (88.3% N1). This meant the safe operation of the aircraft was not compromised."
So it was a "something bad could have happened" situation that called for a change in procedure to avoid a future, real problem.
While my armchair opinion is that ~1500kg mis-calculation of payload on a Boeing 737 is minor - this is one of those problems you should treat pretty seriously as it might have repercussions on other flights.
> safe operation of the aircraft was not compromised.
"There was a fault in the system which, when a female passenger checked in for the flight and used or was given the title ‘Miss’, caused the system checked her in as a child. The system allocated them a child’s standard weight of 35 kg as opposed to the correct female standard weight of 69 kg."
So titling it "Root cause: adult females checked in as 'Miss'" is wrong. The linked article does not even have the word "root".
The mistake was how their software reasoned poorly about passengers with "miss" as their honorific.
So it's more appropriate to say that checking in as "miss" was what exposed the problem, not that it was the root cause.
> UTC AAIB Bulletin: 4/2021 - G-TAWG - AAIB-26814
So creating a title to describe it would have always been required.
I also assumed that the the title had been editorialized/was click-bait until I read the pdf and the incident was filed and issued by the AAIB as a 'Serious Incident'.
Which is understandable given that the flight was 1,244 kg overweight at take off due to a programming error that assumed a 'Miss' = 35 kg / passenger instead of the correct average weight of an adult female of 69 kg.
True that the actual incident was a lot different than the title suggested (and I assumed) but taking off a ton heavier than what the manifest stated is a ‘bit of a problem’ to use the English art of understatement.
If load weight is such an issue, why wouldn't they just have scales at the gates or somewhere alone the taxi route that weighs the entire loaded plane. In the trucking industry there are even certain trailers with scales built-in and portable enforcement units that the highway enforcement use to catch and fine overweight loads.
I have heard stories of helicopter passengers needed to be seated according to their weight to balance a load but would hope the margin of error in a 737 covers atleast 2X passenger weight, they made everything else double redundant on them.
It is an issue, passenger weight is included in calculations, but you don't have to weight everyone.
Huh? "Miss" is not a pronoun.
This is terrible system design regardless, as "Miss" can refer to an adult woman. Per Wikipedia:
> Miss is an honorific for addressing a woman who is not married
Lol wat?
> If load weight is such an issue, why wouldn't they just have scales at the gates or somewhere alone the taxi route that weighs the entire loaded plane. In the trucking industry there are even certain trailers with scales built-in and portable enforcement units that the highway enforcement use to catch and fine overweight loads.
Did you even read the report? They show the readout with the estimated vs actual weight. The captain was aware of the discrepancy, they knew the load sheet was off, but went with it anyway.
> I have heard stories of helicopter passengers needed to be seated according to their weight to balance a load but would hope the margin of error in a 737 covers atleast 2X passenger weight, they made everything else double redundant on them.
The plane didn't crash.. it was within safety factors. The captain decided it was a small enough error. Some random error is expected, but still, it's an issue if the estimates are systematically wrong.
Have you never flown on a passenger flight, or are you somehow unable to imagine how catastrophically inefficient it would be to have a flight delayed/canceled when an airplane has been boarded and is taxiing down the runway?
And if fat-shaming is a concern, how do you think forcing passengers to line up and have their weight recorded? Having an accurate-enough algorithm for estimating weight is obviously a desirable and achievable solution to satisfy safety and engineering requirements.