Enjoyed the video? Let me know in the comments also let me know if any suggestions!
@BrettonFerguson
4 жыл бұрын
SUGGESTION: The Cyanide spill in the Tisza river, tributary of the Danube River, aka 2000 Baia Mare cyanide spill.
@YouShisha1393
4 жыл бұрын
Suggestion: 2013 Radiation accident in Elektrostal. Workers of heavy equipment factory melted the CS-137 radioactive source and some radioactivity was released in the atmosphere. Here is the aftermath video kzitem.info/news/bejne/u46XnJephZ6ln4I, we can see the white noise caused by the radiation.
@danielfosin9949
4 жыл бұрын
10000 RAD is fucked up. Suggestion: keep going Thankyou for the content
@cursedcliff7562
4 жыл бұрын
Wtf how did you do that?
@chanvalentine8283
4 жыл бұрын
We discussed this in error codes in computer class. I had seen a picture of the man with the upper body burn. The deep wound was the size of my fist. The patient had commented that Captain Kirk forgot to set his phaser to stun.
@LoserwinS1
4 жыл бұрын
Being locked in a soundproof room with no way out and being zapped with deadly amounts of radiation sounds like a scene from a SAW movie
@MrTruehoustonian
4 жыл бұрын
But unlike the saw movie there was no escaping this death room.
@Afishionado1
4 жыл бұрын
I'm thinking more Final Destination.
@brencrun5068
4 жыл бұрын
The Phantom Empire starring Gene Autry has a similar scene.
@moriaq23
4 жыл бұрын
its typical for any oncology rooms where radiotherapy is used today. also manipulation and storing and disposing chemotherapy bags are another case due to toxicity of it and probably radiation leaks too that's why nurses applying chemo using lead prone... . also same as with RTG or CT screenings.. . people are let alone in room behind thick glass and walls reinforced with lead and getting "healthy cure" ups I mean "safe doses" :DD
@aaronj08ar
4 жыл бұрын
@@brencrun5068 That looks like a good movie! Gonna check it out thanks!
@illogicalGhost
4 жыл бұрын
"either a dose too low or too high had been administered" don't you think it's kind of important to differentiate between the two with, i don't know, separate error codes??
@deadfreightwest5956
4 жыл бұрын
Well, it's -6 of one, half a dozen of another... who can tell? Probably the patient's fault.
@dustinpendergast
4 жыл бұрын
Lmao
@AndrewMerts
4 жыл бұрын
That wouldn't have helped in the slightest, it did indicate what the "actual" dose was in addition to just an error code but because the machine was on the wrong target and in an invalid state the "actual" measurements were lower than the real amount.
@Stoney3K
4 жыл бұрын
@@AndrewMerts Those measurements were in fact, zero, because the machine was trying to measure the dose of a completely different beam. It would have been easily caught and prevented if BOTH beam outputs were monitored at any given time, if the X-ray beam measured active while the electron beam was selected it would have caused a trip before any dangerous situation happened.
@DoubleM55
4 жыл бұрын
Still better than most software now. Todays you will just get: "503 Internal server error".
@MalcolmCooks
4 жыл бұрын
this feels like something you'd read about in terminal entries in a fallout game
@sarowie
4 жыл бұрын
no, that would sound too unrealistic for fall out. The authors would give the machine manufactures a motive for over exposure, because otherwise no body would belive that something like that would happen over the span of years without anyone realizing the problem.
@kbtechandmedia
3 жыл бұрын
I am really glad I am not the only one who thought this. The computers in Fallout kind of resemble a 64k data connected network using an old school BBS like system.
@solo_em1124
3 жыл бұрын
I completely agree lol
@highadmiraljt5853
3 жыл бұрын
Yeah
@jaggerjack36
3 жыл бұрын
True, minus the gong sound.
@sarqf212
4 жыл бұрын
Operator: Stop high radiation Therac-25: I'm sorry Dave, I'm afraid I can't do that
@antisoda
4 жыл бұрын
As with HAL, the problem was in the programming and the inputs given. HAL was told to lie and Therac-25 was given orders when it wasn't listening. :/
@pullt
4 жыл бұрын
It's more like: Therac-25: Stop high radiation Operator: P Therac-25: Stop high radiation Operator: P Therac-25: Stop high radiation Operator: P Therac-25: Stop high radiation Operator: P Therac-25: Stop high radiation Operator: P
@dfsafadsDW
4 жыл бұрын
Why aren't you pinned :(
@philxdev
4 жыл бұрын
computer says NO...
@lisacollins5868
3 жыл бұрын
😆😆😆😆😆😆😆😆😆😉
@sulfreez
4 жыл бұрын
9:25 The phrase "the operator thought it was a usual quirk" regarding an error of the radiation administering device speaks loads about manufacturer QA process.
@painovoimaton
3 жыл бұрын
Right, radiation administering computers probably should not have "usual quirks"... What a mess.
@SimonBauer7
2 жыл бұрын
@@painovoimaton yes my car has quirks with its high beam although i will fix it soon but the high beams not working sometimes isnt deadly. a radiation therapy machine shouldnt have such errors.
@haruhisuzumiya6650
Жыл бұрын
Or lack thereof?
@somebod8703
Жыл бұрын
@@SimonBauer7 A malfunction in your car is just as likely going to kill someone as a malfunction in a medical device. Just as there are unimportant malfunctions in cars, they exist in medical devices, too. When our family car started smoking out of the engine department more or less randomly, we were told by the manufacturer that this was normal and no cause for worry. Problems which are not safety critical seldomly get fixed. Often, laws require medical devices to be safe, not to be operational. It's a customer satisfaction problem, not a safety problem.
@dahliacheung6020
4 ай бұрын
Not to mention having the entire code written by a single developer and never having it looked over independently. I get that they didn't believe software could have errors (which to me makes no sense because it was written by a person and people are definitely capable of making mistakes) but not even checking comes across as overconfident and careless.
@Hydrogenblonde
4 жыл бұрын
I find these medical radiation overdoses more troubling than reactor meltdowns.
@PlainlyDifficult
4 жыл бұрын
Pretty scary
@steve1978ger
4 жыл бұрын
What doesn't make us stronger, kills us
@koghs
4 жыл бұрын
Nah it's nothing, you cannot really harm many people with those things even if you want to. I mean you heard him, there was only 11 of those and they harmed only 6 people in 2 years of usage.
@RCAvhstape
4 жыл бұрын
@@koghs You miss the point. When you go to a hospital you expect that they are taking care of you, that these systems have been thoroughly checked out. Plus, unlike Chernobyl or Fukushima where reactor buildings are exploding and alarms are going off, there's no sense of danger when one of these things kills you. You feel discomfort, then you get sick and die while the operator works another shift.
@willyk2202
4 жыл бұрын
@Helium Road I’m hoping @Koghs was being sarcastic
@Peterscraps
4 жыл бұрын
Patient:: Hay this machine gave me radiation burns Doctor: Seems normal to me *it was not normal*
@fastinradfordable
4 жыл бұрын
I had a nurse pull my iv out and walk out of the room. I felt very cold. Looked down and saw a pool of blood around me. I was bleeding out alone. Fuck hospitals. They kill people.
@LFjuniorful
4 жыл бұрын
Oh no patient died.... OK, NEXT!
@xavier1964
4 жыл бұрын
Peterscraps?????? From the musclek video???
@dion8202
4 жыл бұрын
Love your tf2 vids
@jbh759
4 жыл бұрын
@@fastinradfordable Sounds like a load of bullshit to me
@andrejspecht8217
4 жыл бұрын
I do believe that part of the problem was the arrogance of the doctors as well. If a patient says he's hurting, where he isn't supposed to be, you gotta listen.
@a64738
3 жыл бұрын
Your doctor on average kills one person a year... Doctors arrogance kill a lot of people every year and is one of the largest killers there is in the modern world.
@WitchidWitchid
3 жыл бұрын
I have come across quite a few doctors who seem to routinely ignore their patients complaints. In fact I have found nurses to be much more responsive and concerned regarding patient complaints than doctors. And the doctors sometimes ignore them too.
Doctors do not directly administer the daily treatments. Unless the doctor has a very extensive engineering or computer background, he or she is highly dependent on the manufacturer and others. Physicists and engineers have more responsibility concerning the proper functioning of the unit. The technicians operating the machine should report errors and the situation addressed by a team of the correct people. No patient should ever be treated when the AV system is not working. The instance where the doctors could be faulted is the one where the patient was sent for more treatment depending on the severity of signs and symptoms initially.
@francisharkins
3 жыл бұрын
Why do you think doctors are the worst patients? Because : 1. Either their too arrogant and prideful to admit fault or take another medical opinion. 2. They know how much of f'd up stuff that the medical field practices and don't want to expose themselves to similar bs.
@calvinit21
4 жыл бұрын
"The system did not consider computer software errors" Me, an experienced software developer: No no no no no no no no no no no no
@deusexaethera
4 жыл бұрын
It doesn't help that the manufacturer NEVER INVESTIGATED SPURIOUS ERROR MESSAGES IN PREVIOUS VERSIONS OF THE SOFTWARE. Jesus, I write code that draws points on maps and I STILL have to investigate every error message the users receive.
@anhedonianepiphany5588
4 жыл бұрын
It sounds so ludicrous, doesn't it?!? To be fair though, unless you work exclusively in assembly/ machine code on platforms with extremely limited memory, then it can be hard to grasp how taxing this can be (especially without the aid of modern tools). As far as I know, the bulk (perhaps the entirety) of the Therac-25 software was coded by just _one_ individual (as absurd as that seems) on the old DEC PDP-11 system. Who really knows how well acquainted this programmer was with the operation, and practical use, of the various Therac models the software was designed to control?!?
@deusexaethera
4 жыл бұрын
@@anhedonianepiphany5588: Logic-modeling languages such as Zed or B exist specifically to aid with double-checking the validity of logic operations. (I learned about this in software engineering classes in college.) The programmer should have used one, and told his supervisor to deal with the delays caused by the extra work, because the alternative was possibly nuking patients to death. Granted, I don't use logic-modeling languages to double-check my work, but my work isn't safety-critical.
@anhedonianepiphany5588
4 жыл бұрын
@@deusexaethera I wasn't making excuses for the programmer in question - _at all!_ I completely agree with you. There should have been more than a single individual programmer, and they should have used the best tools available to them.
The part about being trapped in a room where nobody can see or hear you, whilst a big machine zaps of radiation sounds like it must have been genuinely terrifying.
@STARDRIVE
3 жыл бұрын
Just lay down on the bed with the burn marks, and ignore the body bags in the corner.
@warrenSPQRXxl
3 жыл бұрын
If the AV system does not work no treatment should be given. Why a basic safety protocol was not followed is disturbing.
@erikasauer4140
2 жыл бұрын
If the doctor isn't willing to be in the room with you, chances are the treatment is unsafe.
@warrenSPQRXxl
2 жыл бұрын
@@erikasauer4140 No. If a doctor or a technician (who actually administers the treatment) were in the room during a treatment they would be receiving radiation with each treatment. It would be like asking a doctor or nurse to take the same medicine unnecessarily which they prescribed or gave you.
@cats.m.2853
2 жыл бұрын
According to my mom, it’s pretty chill. Nowadays they play you music and give you an eye mask, and you kinda just nap
@korinthian7313
4 жыл бұрын
I think all disasters should be rated with a foam finger and fridge magnets from now on to avoid panic.
@PlainlyDifficult
4 жыл бұрын
I agree!
@deusexaethera
4 жыл бұрын
Someone get Fox News and CNN on the phone.
@bogdangabrielonete3467
3 жыл бұрын
Rating a disaster scale : Panik Using a foam finger and some children magnets : Kalm
@Prizzlesticks
3 жыл бұрын
It would still be nice for him to verbalize the number being pointed at for the visually impaired tho...
@baardkopperud
4 жыл бұрын
Warning: ? Are you sure you want to nuke this patient? Yes No
@Jay-ln1co
4 жыл бұрын
Drinking bird at the keyboard: "Y"
@OffTheBeatenPath_
4 жыл бұрын
More like YES or YES
@dmhendricks
4 жыл бұрын
Reminds me of the old "Abort, Retry, Fail?" error prompt in DOS. Why would I want to choose fail? #Microsoft
@LFjuniorful
4 жыл бұрын
Press "yes" to "no" or press "no" to "yes"
@_tyrannus
4 жыл бұрын
P
@16vSciroccoboi
4 жыл бұрын
It's amazing and frightening how quickly this stuff can have life altering or ending consequences. Especially with little drama involved. No bright lights, no explosion, just a click of a button and by the time you have a reaction you've received a lethal dose.
@fastinradfordable
4 жыл бұрын
I’m sure 5 months to die without use of arms or legs might as well be eternity.
@primodragoneitaliano
4 жыл бұрын
Totally. Makes me think of those nuclear accidents with radioisotopes in liquid form where the guys had no clue of how much there was in the solution/recipient already and what they added it was enough to trigger a prompt criticality/supercriticality, even if this action is by itself something already done a billion times and pretty harmless. And yet.... Here's one example of such an event: en.wikipedia.org/wiki/Cecil_Kelley_criticality_accident
@16vSciroccoboi
4 жыл бұрын
@@primodragoneitaliano yeah, but at least the criticality accidents have people who know there are those risks, and they get a flash of light to let them know that they just became irradiated. Crazy how fast it happens though. Nuclear physics is fascinating, one of the most fascinating parts is how severely quickly things like criticality incidents happen.
@devinwalton408
3 жыл бұрын
Brrrrrrrrrt Omae Wa Mou Shindeiru
@frodo4087
2 жыл бұрын
"The death ray button" ™
@ZRTMWA
4 жыл бұрын
Imagine being the single AECL programmer that designed the Therac software. He probably lived/lives with unimaginable guilt.
@PlainlyDifficult
4 жыл бұрын
I can imagine they would
@ianmoseley9910
4 жыл бұрын
not to mention the operator who managed to kill two patients by, effectively, the same error
@deusexaethera
4 жыл бұрын
Nah man, I was just doing my job. The engineers should've designed the machine better. It's not my fault.
@xsk8rat
4 жыл бұрын
Everyone at the clinic would have been devastated. You see these patients every day for weeks at a time. Everyone at the manufacturing company would have experienced the pain. We get into this business to help patients. (Didn’t work at this manufacturer. But we studied all reports of errors - not just our own issues.)
@ericsingbiel6622
3 жыл бұрын
RICH PEOPLE WITH A CONSCIOUS DON'T SURVIVE VERY LONG IN OUR SOCIETY.
@Acceleronics
4 жыл бұрын
When I started out as a medical device systems engineer back in the late 90s, one of the seasoned firmware engineers gave me the Therac 25 case history to read. 20+ years later, I still think about those patients when I lead a product risk assessment. Charles P from Nellcor, if you watch this video and read this comment - THANK YOU!!! You were a major influence. Jeff S.
@RezaQin
4 жыл бұрын
A doctor hit the P key 5 times, this is what happened to the patient's chest!
@loscheninmotion9920
4 жыл бұрын
Chubbyemu time
@Itried20takennames
4 жыл бұрын
Probably a radiation tech, not a doctor, at least currently. The machines are so specialized, and potentially dangerous if not correct as seen here, they are a whole separate profession.
@freebirthfreddy
4 жыл бұрын
@@Itried20takennames Your response is technically valid, but I think you kinda missed the joke
@Masterkodak
4 жыл бұрын
quirky/funky music plays
@5roundsrapid263
4 жыл бұрын
Plainly Chubbyemu
@sethsims7414
4 жыл бұрын
We had to study the Therac-25 in my Computer Science ethics class.
@codblkops85
4 жыл бұрын
Good
@rolfen
4 жыл бұрын
Where is that? Sounds like a place where I want to study computer science.
@sethsims7414
4 жыл бұрын
@@rolfen Southern Polytechnic State University. It doesn't exist anymore it was absorbed by Kennesaw State University. It's in Marietta Georgia and Kennesaw.
@CMDRSweeper
4 жыл бұрын
Well then you can also study the Airbus Fly by wire system afterwards... Unlike the Therac-25 the requirement were 3 computers, all running the same software in terms of functions, but the software had to be written in 3 different languages and naturally by different groups. And then the computers have to agree in the aircraft... Are they still safe? Well, Quantas flight 72 may be the exception to the norm for that one.
@sethsims7414
4 жыл бұрын
@@CMDRSweeper Therac-25 was because the software used a byte as a go-no-go flag to see if the machine heads were in place. But instead of setting it to 1 to indicate not-ready the programmer incremented it. So if you got more than 255 checks before the machine was ready the byte would overflow back to 0. Which would let the dose begin even though the machine was not configured yet. I remember there be something about only very fast operators could make this happen. They had to have everything ready for the dose quickly enough that the computer would perform that many checks before reaching final configuration.
@compu85
4 жыл бұрын
This is a fascinating incident to read about. At one point AECL instructed the hospitals to pry the up arrow key off the VT100’s keyboards! The Texas hospital was also critical in figuring out what was going on. A doctor and nurse there didn’t believe AECL’s statement that the machine could not screw up, and were able to reproduce the problems.
@mr_waffles_the_dog
4 жыл бұрын
this misses a huge amount of the AECL lies about occurrences of events, the amount of pushback they had for everything, and their repeated incorrect and insufficient "corrections". It also misses that the actual failure was not discovered by AECL, but rather a physicist at a hospital that had had multiple events that AECL claimed weren't a failure in the machine.
@sarahamira5732
3 жыл бұрын
Person: *gets their friggin face microwaved* AECL: what do you mean? That's supposed to happen. The dude doesn't have a problem with cancer anymore 🤷♀️
@MajinOthinus
3 жыл бұрын
@@sarahamira5732 "gets their friggin face microwaved" Getting your face microwaved would be fine, getting your face hit with a super intense X-Ray beam is not.
@sarahamira5732
3 жыл бұрын
@@MajinOthinus chose that term cuz it sounded better for the meme lol. Tbf I'm pretty sure getting you head blasted by a huge amound of microwaves would also probably mess your shit up. Regardless "gets their friggin face x-rayed" doesn't incur enough of a ... Visceral reaction
@MajinOthinus
3 жыл бұрын
@@sarahamira5732 I guess, though as for microwaves, nothing really would happen; at least nothing that's life threatening. Microwaves can't pierce the skin, so the worst you'd get, would be 1st or 2nd degree burns with very high intensity radiation. Ironically, the steam from your boiling skin would likely protect you long enough, to withdraw the exposed part. You might go blind tho if ti hits the eye.
@qdaniele97
Жыл бұрын
@@MajinOthinus I assure you, getting your face microwaved wouldn't be fine
@rolfen
4 жыл бұрын
"The poorly engineered software has led the operators and technicians to become complacent with the error messages displayed to them" A well engineered medical device should not even have a concept of viewing and overriding errors by machine operators and technicians. Either it works safely or it doesn't in which case the engineer should have a look. This had "accident waiting to happen" written over it.
@DrewskisBrews
4 жыл бұрын
Indeed. Making a habit of ignoring errors, alarms, and malfunctions is at the heart of many a calamity.
@Stoney3K
4 жыл бұрын
More surprisingly it didn't even have any substantial logging of errors, after a system reset it was all cleared. I mean, keeping a log of all faults and saying 'Hey, this error 53 has popped up more than 5 times, I'm not going to do anything until you send an engineer to fix it' would have been a good prevention mechanism. But it also means the machine was crippled and hospitals didn't earn any revenue from treatment.
@Audiodump
4 жыл бұрын
I'm a radiation therapist and while I understand your feeling, at the same time its also not really viable. At least not for EVERY error. For example, if I am setting a patient up and bump the treatment head, the computer will read a collision and lock everything down. Obviously there's no issue with just overriding that kind of error because I know the cause and its not a danger to anyone. The errors you don't want to let people override on their own, especially with a single keystroke, are the ones that they don't readily understand. Also, there's some things you have to consider about how RT is done. First we see TONS of patients a day. Treatments take like 10 minutes each and we see patients back to back to back for the entire day. As such they're tightly scheduled and any delays throws everyone behind. Second, most places do not have a dedicated engineer to service the system. They have a company engineer which services dozens of places all over the city or even over several cities. Calling someone in can shut the machine down for hours or even the entire day. And finally, radiation therapy treatments are a daily thing, with each patient getting 20-30 spread out over a month and a half or so (skipping weekends). The more continuous and reliable those doses, the better chance of cure. If we shut the machine down to service it and people don't get treated, it can literally be the difference between life and death for them. All these things combine to make it VERY hard for a radiation therapist to hold up their hands and stop all treatment for an engineer to be called in. Especially when they're not really sure what the problem is, or if its even a problem at all. The expectation is placed on them to be able to keep the machine running and effective. I'm not trying to excuse this, I'm not trying to say this is just the way it is, but I am trying to explain the current reality of treatment so you can understand how things like this end up happening.
@rolfen
4 жыл бұрын
@@Audiodump Thank you. I do have a vague idea about RT, and understand your concerns. Let's address your example about a misaligned head. In this case, the operator should re-align the head manually (just like they misaligned it manually) and the machine would detect that the position of the head is back within parameters and allows them to proceed, either automatically or through a manually trigger. This type of error would not preclude the machine from operating safely, as long as the head is confirmed to be in the right position before treatment begins. On the other hand, if the patient is due for a dose of 20 and the ion chamber measure 10 time that dose (like what happened on the Therac 25), then an error should be raised. An operator cannot cause such an error. Only a serious fault in the machine could. The safe operation of the machine would be compromised and this error should not be overridable by the operator, ever. Modern machines would presumably have a detailed audit / log trail for the engineer to look at. Thank you for bringing this example to my attention, it is indeed more subtle than I imagined. It's nice to see how considerably RT seems to have matured since the 80ies.
@Audiodump
4 жыл бұрын
@@rolfen I will say that when I said I bump the head, I don't mean knock it out of alignment. There's a sensor bar on the head of the machine oncologynews.com.au/wp-content/uploads/External-Beam-Radiation_oncology-news-australia.jpg Its that ring of metal around the head. If pressure is placed on that ring, it will stop the machine. Its a safety sensor to prevent collisions. There are many parts of the machine that have these, including the orthovoltage imaging panels on the sides. I've bumped into them or hit my head on that ring plenty of times. I think the only way to knock the head out of alignment would be with a sledge hammer or perhaps a small car. Its several thousand pounds of solid metal after all.
@keleighshepherd345
4 жыл бұрын
I'm a radiotherapy engineer trainee, and I had to do a presentation on this exact topic! Needless and long lasting consequences, we have *much* more thorough safety interlocks on modern digital linacs nowadays, as a result of this and the Zaragoza incident (and the Essex incident, and more...)
@ChiDraconis
4 жыл бұрын
Please read Dov Bulka; I never really figured this one yet for me is obvious; The Airane was obviously buffer wraparound; X-Ray Zappers or whatever even Glow in the Dark all of this requires that actual hardware be right so that signalling propagates via wire; Instructors will tell you other wise or might yet for let's say MRI they yet get axial wobble-broadening of the spectral lines-incorrect in fancy nMRI texts let alone trying to discuss Quantum Resonance across bandwith → At 1.5 Tera something resonant frequencies of fat and water protons differ by about 220 Hz ... Try suggesting the O-2 resonance and you can plan on using the PhD to wipe floors if there are any jobs
@anhedonianepiphany5588
4 жыл бұрын
@@ChiDraconis What exactly were you hoping to achieve by submitting this jumbled mess of what _you_ consider to be somehow esoteric tidbits, combined with nonsense?!? Don't worry, I don't expect you'll interpret my subtext correctly ... not even with that, uh - _cough cough_ - "PhD" of yours.
@ChiDraconis
4 жыл бұрын
@@anhedonianepiphany5588 I threw my PhD in the trash; Actually it was minister that I got from want-ad in Rolling Stone magazine Major M.L.C. Funkhouser is your hero?
@Dutch3DMaster
4 жыл бұрын
My brother applied for a job as a technician of a linear accelarator once, and if someone were to be able to get off the bed in order to signal something was not right, he or she could enter "the maze" (a U-shaped hallway with walls of a meter thick lined with lead that lead up to the operator of the machine) and trip a sensor, sounding an alarm, and when tripping the second sensor in "the maze" would shut the machine off, if only they had applied that safety mechanism in these. He also learned that the videocamera's meant to be able to check on the patient remotely had to be replaced every 4-5 months due to radiation damage. When my brother got a small presentation in which he was asked to name some of the internal components when the head of the accelerator was opened up he said this upon coming home: "Yea, it's good that that thing is covered up to look sort-of nice, that thing looks freakishly scary when opened up...."
@keleighshepherd345
4 жыл бұрын
@@Dutch3DMaster imgur.com/a/Z2Hxy25 - This was taken by me recently when trying to replace the secondary position sensor (blue/striped silver bit) on the MLC (multiple leaf collimator, shapes the beam to the plan for the patient), Varian make it an absolute nightmare to service, as you have to use a jig to lift the entire MLC bank up to replace something that takes
@neeneko
4 жыл бұрын
heh. I remember this case! It was one of the case studies we went over in software engineering classes. Stupidly enough, this was not even the only example of a medical device where the operator typing at the wrong time could be deadly. I wish I could remember the name of it, but there was another machine that had some kind of buffer overflow problem where typing faster than the system could drain the buffer would result in corruption.
@anhedonianepiphany5588
4 жыл бұрын
It seems as if there is insufficient consultation/coordination between software developers and those who actually administer radiotherapy treatments (in addition to myopic, sloppy, and poorly debugged/reviewed, coding). Have a look at the Wikipedia page "List of civilian radiation accidents" or, failing that, the IAEA archives, if you're seeking specifics of that case.
@AlessandroGenTLe
3 жыл бұрын
Let's speak about the Ariane V rocket and its half *billion* $ sw failure ;)
@neoqwerty
2 жыл бұрын
@@AlessandroGenTLe Holy shiiiiiiiiit. That's got to be the best argument I've ever seen for "watch how you handle your diagnostics data and maybe redesign the software when you get much-improved hardware". Thank you for dropping that in the comments!
@mirandanorman2729
2 жыл бұрын
Nowadays we have a separate remote to control the machine, with green, yellow, red buttons for beam on, beam interrupt, and beam off. While you deliver the treatment you hover your finger over the pause button so you can stop at a moments notice
@phorzer32
4 жыл бұрын
Hardware interlocks.. so important...
@jyralnadreth4442
4 жыл бұрын
Backups for Backups.....on Airliners computer controlled landing gear deployment with a manual landing gear lowering mechanism backing it up
@Stoney3K
4 жыл бұрын
@@jyralnadreth4442 That's called defense in depth and it's a fundamental concept for safety design. The software interlocks should never replace hardware safety mechanisms, only supplement them. This enables the software to catch problems if the hardware safety mechanisms fail, and vice versa.
@kshatriya1414
4 жыл бұрын
Jyral Nadreth like even most video editors have backups on their on their backups backups and they don’t even have other human lives on their hands lol
@deusexaethera
4 жыл бұрын
Hardware interlocks are good, but software should ALWAYS ALWAYS ALWAYS be tested with the hardware interlocks disabled or removed to ensure the hardware interlocks aren't obscuring software errors.
@phorzer32
4 жыл бұрын
@@deusexaethera Yes.
@erenoz2910
4 жыл бұрын
9:13 "either a dose too high or a dose too low had been delivered" genius troubleshooting right there :/
@richarddixon6001
3 жыл бұрын
AECL: "We have a coin in the spare parts kit for you to decide which it was. Good Luck."
@OvAeons
4 жыл бұрын
**people die** AECL: well within expectations!
@ImSquiggs
4 жыл бұрын
One time I messed up the coding in a hotkey at work, and people couldn't answer the phones for a half hour. Doesn't seem like that much of a big deal anymore after hearing this, hahah.
@chrismaverick9828
4 жыл бұрын
I work for a large retail chain. One day a software tester at Corp IT clicked the wrong check-box on the test subject and uploaded it to the main company system. Half the computers/registers dumped their connections to corp, which everyone had learned to simply reset to fix. Upon reset, the entirety of the new software took effect, crashing the affected comps ability to link into the network. This cascaded in general to result in most stores having one functional register, some none at all. An email was sent out ten mins after the fact, but the variability of when people checked the mail and how useful the recipients were played hell on everything. The one check-box click resulted in the company spending the equivalent of buying a property and building and opening a new store to effect emergency operations to the worst affected stores and to get everything fixed and upgraded. On the plus side, it created an immediate requirement to do full upgrades on the older stores' systems that were long past being desperately needed, a complete re-write of the main software which was a paste-and-glue special of barely compatible modules written by different companies, and an fairly important lesson on how to plan bigger if you're looking to make massive expansions. Interestingly enough, the CEO took the whole thing in the "a small mistake has illustrated a litany of massive threats that would have happened eventually, so lets just fix the damned problem and go on with business a bit wiser" way, and my understanding is that the tech didn't even lose their job over it. But no one died from that, that I am aware of, so this story is kind of off-topic except to show that programmers are human too.
@michellejirak9945
4 жыл бұрын
@@chrismaverick9828 Wise reaction from that CEO.
@duckmeat4674
4 жыл бұрын
@@chrismaverick9828 honestly thats great from you CEO. If you can bring down a system just by selecting a wrong checkbox, this was going to go wrong sooner or latter, and you hope that its sooner
@juliogonzo2718
4 жыл бұрын
Years ago in the city I lived in, the entire municipal as400 system went down for a while because somebody plugged a coffee maker into the circuit powering the mainframe, and tripped a breaker or something.
@AngelaMerici12
4 жыл бұрын
Next time you make a similar error: "No one is being blasted with radiation here. Move on!"
@vr6swp
4 жыл бұрын
My mother was treated for cancer at the Kennestone Oncology center. She passed in 1976, which would have been prior to the Therac-25 being in use. My family was always very suspect of the treatment given her by Kennestone, which seemed to consist of heavy doses of radiation treatment and morphine, and little else.
@ArakDBlade
4 жыл бұрын
I worked in medical software for almost a decade out of school (thankfully registration and document management, not stuff like this!). This was one of those incidents that kept cropping up in various presentations and training courses.
@gottabesandi
4 жыл бұрын
I’m glad they learned from it and proceeded to use it as a teaching case study. Way better than sweeping it under the rug and fading into obscurity
@ArakDBlade
4 жыл бұрын
@@gottabesandi Sadly it was more a blibbit saying "don't screw up or we'll get sued" but its the thought that counts. XD
@sarowie
4 жыл бұрын
@@ArakDBlade well, there is also "would you recommend your product to friends and family", to which a faithful R&D engineer would go to after-sale support and ask which treatment facility has good practices and who has bypassed systems in the past.
@mattdouglas2778
4 жыл бұрын
Gotta love how they normalize mistakes to avoid inconsistent programmers from self omitting themselves from situations as these. The programmer, the company and everyone along the road to those deaths should be criminally charged. It's the only was this garbage will stop.
@robozstarrr8930
4 жыл бұрын
very good explanation, was working at a division of Varian during the 80's when this made news. Varian's radiotherapy equipment was all analogue at that time.... couldn't keep up with the orders.
@Vegalyp
3 жыл бұрын
My dad used Varian 21EX's for decades at his workplace.
@_etwas_
3 жыл бұрын
Therac-25: "So anyway I started blasting"
@MrTLsnow
3 жыл бұрын
Patient: "and I took that personally"
@dahliacheung6020
4 ай бұрын
Therac-25: "Born to be wiiiiiiild!" (Immediately followed by life-changing radioactive light show.)
@cmotdibbler4454
4 жыл бұрын
using an ancient at the time 16 bit computer with code written for a different machine, they were lucky that it didn't go Civilization Gandhi on people
@Cutest-Bunny998
4 жыл бұрын
But it did basically go Ghandi as they had a overflow error in a flag variable. So they literally did go Ghandi and Therac-25 became so democratic it activated the nuclear weapons (Really though, as this flag overflow was a direct cause for one of the beam exposures, aka clinical misadventures, during mode selection.)
@ianmoseley9910
4 жыл бұрын
1982? Not that ancient at that time
@lsswappedcessna
3 жыл бұрын
UH OH, INTEGER ERROR! TIME TO NUKE EVERYBODY JUST TO SHOW THEM HOW PEACEFUL I AM!
@mina47879
3 жыл бұрын
Old hardware isn't necessarily a bad thing tho, it can be good for critical systems because it's well understood and proven to be reliable. As an example, there's a reason the new Mars rover is basically equivalent to two iMac G3s (the old ones with the translucent coloured plastic exterior) in terms of computing hardware. A more powerful computer isn't required for the application and the old hardware is well understood and has been demonstrated to be reliable. Newer isn't always better. Copy pasting your old design that's been proved to work reliably really isn't a bad idea. But obviously you have to re-examine the requirements of the old system, how it works and how it interacts with the old hardware and compare that to how the new hardware works, how it will interact with the new hardware, etc, and do lots of rigorous testing to make sure there aren't any problems.
@Bakamoichigei
4 жыл бұрын
Wow, the poor bastard that got his _head_ microwaved... F to pay respects. 😬
@biblemaniswatchingyoumastu1920
4 жыл бұрын
Spoiler alert
@koghs
4 жыл бұрын
Ain't that a kick in the head
@fastinradfordable
4 жыл бұрын
Ffffffffffffffffffffffffffffffff Why won’t it work. Fffffffffffffffffffffffffff
@cleidsonaraujopeixoto163
4 жыл бұрын
F.
@fastinradfordable
4 жыл бұрын
Ffffff just thought I would do due diligence. Fffffffffffffffffffffffffffffffff
@davidp.5598
4 жыл бұрын
WOW!!! Just WOW!!! I am at a lost for words! I can't believe the arrogance of the software company! Thanks PD, for bringing this to us. A great job PD!!!
@PlainlyDifficult
4 жыл бұрын
Thank you
@fastinradfordable
4 жыл бұрын
What about the arrogance of the technician just pushing the button to do away with the error?
@pavelsovicka5292
4 жыл бұрын
@@fastinradfordable If the machine spits out errors on a regular basis and offers a quick solution that allows it to continue, then sooner or later the operators will stop being cautious and accept it as regular operation. Its just how us humans work. Therefore the manufacturer has basically three options: 1) Make the machine work without regularly displaying errors. Therefore when an error pops up it concerns people. 2) Make the errors display what has actually happened. So not just "error 54" but a complete description and do not enable operation without a direct check of the machine or a complete reset of the machine to a default state. This was hinted in the video when mentioning the faulty "P" key continue button solution. 3) Have a totally independent solution that just jams the thing if a dangerous state could be reached - the hardware locks in this case. Of course doing all of it is the best. Making a machine, that annoys operators and through this annoyance a dangerous operating condition can be reached is just asking for a disaster. Its like if your car spits out "check your tires" often. Sooner or later, you will start to ignore it because you have checked the tires many times and always the tires were fine. And a time will come, when the sensor will be right, but the driver will just say "this error again" and drive...
@Kavurcen
4 жыл бұрын
@@fastinradfordable I have never worked on a system as potentially deadly as this... but a failure mode that simple would have been anticipated and mitigated by engineers in the design phase for a kitchen gadget. The level of "asleep at the wheel" here is terrifying. I have worked on a device that was being documented for an FDA CLIA waiver and the idea of something like this making it through an approval process is thankfully very hard to imagine. And even then we had execs saying "we will assume that risk, just send it." Remember that when people talk about "cutting red tape" or removing "unnecessary regulation," the procedures they are talking about are often written in blood.
@compu85
4 жыл бұрын
Nathan Brame This is what AECL instructed them to do! AECL didn’t provide any list of possible faults, and when the operators reported problems they were told everything was OK, just use the proceed function.
@Will-fn7bz
3 жыл бұрын
As a programmer myself I can attest to the fact that absolutely unimaginable combinations of keystrokes/events can result in weird results, and the need for testing by a third party is crucial for any program that could have a significant impact.
@sayori3939
2 жыл бұрын
I know right, i remeber i hit my keyboard in a weird way and it made a log off/log i didn't understand anything, it shouldn't have been possible to do that from hitting some keys ( most of them were letters i may have hit shft as well)
@kenhammscousin4716
2 жыл бұрын
It's so sad because the victims were terribly ill, they were desperate, and they had to trust their doctors completely, being assured of the safety to quell their probable skepticism, or uneasiness. Even after the pain started, you know their trust kept them in the danger zone at least until the pain became unbearable.
@PMW3
4 жыл бұрын
here's an idea, maybe have a different error code for a too low of a dose and too high of a dose
@FBI-ej8zr
4 жыл бұрын
you would think thats common sense.
@Cutest-Bunny998
4 жыл бұрын
Why would you want idempotent identity for errors? Just have one single user communicated error called "UHOH ERROR" and then use use it for everything violating any precondition or postcondition and it should be fine. /s
@NGC1433
4 жыл бұрын
This is bad wording by non-programmer humans. This error means "Dose so far out of measurable range boundaries that machine has no idea what it was."
@DoubleM55
4 жыл бұрын
Nah, just have a "500 Internal server error" for everything.
@Dutch3DMaster
4 жыл бұрын
Complete guess in the dark here: maybe the amount of memory for the complete program was insufficient for those extra lines of code which caused it to get bundled up together like that in much the same way the errors in some software resort to error numbers or codes because a description of the problem would probably not fit in memory space (especially when the error causes a GUI of some sort to fall to the most basic levels of all and you are dealing with the most low-level address-space calls of all).
@steve1978ger
4 жыл бұрын
"Ouchi!" - "Doctor, the patient died." - "Bollocks."
@handledhandlehandlinghandler
4 жыл бұрын
I see what you did there.
@deadfreightwest5956
4 жыл бұрын
The patient died, but the procedure was a success.
@AndrewMerts
4 жыл бұрын
That was a reference to Hisashi Ouchi who received a lethal radiation dose in a criticality accident and was in quite horrific pain until he fell into a coma and later died. Truly horrific way to die.
@bowtiejess80
4 жыл бұрын
I stopped the video directly after the reference to read comments and see who else picked up on that. Not really funny but biggest laugh I've had in a long time.
@crazyshipper7493
4 жыл бұрын
My history teacher: There weren't that many radioactive incidents... Plainly Difficult: Well yes but actually no.
@randyrhodes3934
4 жыл бұрын
I was the service engineer on the T25 in East Texas at the time of the two fatal doses. I was employed by AECL. The unit was under service contract by AECL. There was a lot of effort put forth to find the cause of the overdoses. The first patient described a large shock like effect. Since an overdose was not suspected initially, The “pop” noise and shock effect were investigated. The unit was thoroughly checked by AECL engineers and outside consultants. No evidence of overdose was determined initially. After the second patient received an overdose, the technologists operating the unit was asked to recreate the situation. Once the pattern for the overdose was discovered, it was easily duplicated. When the overdose was discovered the unit was taken out of service. The overdose was far greater than described in the video and was not survivable. While there were legal determinations made and financial settlements no one can underestimate the emotional and human cost this caused at every level. Everyone in the business of radiation therapy takes their role seriously and incidents like this are devastating and yes fatal. In the end the fix was relatively cheap and as one might expect the T25 lost its glory as a cancer curing machine. I have as I aged had the occasion to have radiation therapy on an e-beam unit and these thoughts ran through my head. The treating technologists trained on a unit I had assisted with installing 30 years prior. The physicist who calculated my dose worked with the physicist who was on site in East Texas. Anything made by man can fail, and will one day; don’t be afraid of technology it may save your life as so much has in the past 30 years. Life is hard.
@shaneclarke6307
2 жыл бұрын
"The overdose was far greater than described in the video and was not survivable." - Yes, according to sensations of patients during and right after the "treatment" that sounds correct. Even for a local exposure overdose was too high to survive.
@disbeafakename167
4 жыл бұрын
Giant nuclear computer: ERROR! Idiot human operator: Nah, it its fine. Giant death laser: ERROR!! Idiot Human: just do what I tell you, I know what I'm doing! Killer robot: Guess we're a murder machine now.
@Foxesinthetrees7
4 жыл бұрын
XD thanks m8 for the laugh
@alzeheimersgaming
4 жыл бұрын
It was the computer that killed these people, not operator error. That's like the entire point of the video
@taotoo2
4 жыл бұрын
@@alzeheimersgaming When it throws an error you look it up. If it states "under or over dose" you take the machine out of service until it's been fully investigated.
@Cutest-Bunny998
4 жыл бұрын
@@taotoo2 Actually in most of the accidents the errors were undocumented for anyone but the company. There was no way for a user to know what any of the 64 malfunction errors meant according to the official FDA analysis. The machine also constantly threw erroneously generated errors and underdosed patients, up to 40 times a day in some cases. For example, in one of the accidents they got a pause treatment error and a Malfunction 54 that has the following exclusive documentation in the user materials: "dose input 2". You know what that means without being told? I know I wouldnt, and it is also true that they didn't, but they knew the machine constantly underdosed and it was a low priority problem since it was able to be (p)roceeded via the "p" key. That happened 5 times in a row before the forced shutdown. The machine told them it was okay though, only 6 of 202 dose units were given and it had verified that amount was true. It showedan underdose, so the operator logically proceeded until the forced shutdown after 5 times. Alternatively, in another accident it was a low priority halt from an error saying "H-tilt" and everything else indicating no dose given, so the user ignored the H-tilt message as it simply indicated an extremely common issue with the microswitches and happened all the time. It also said the patient had definitely recieved verified "no dose", and so multiple times they resumed until the forced shutdown. Note that only after this accident occurs are the operators and hospitals informed that the various tilt errors are a risk to patient health (the training and documentation did not make that previoualy clear). Also note that later on after fixes by AECL which they told the FDA were not verifiably and definitely fixing the problem, some or potetially most operators will be told by AECL to not bother with the newer and cautious response to tilt messages (or the additional visual checks) after they get the machine fixed, since thag makes the machine 500% safer than it was (but AECL makes no mention about the uncertainty). In both these examples they did the only thing that they could have reasonably done given that they had been trained by the machine to ignore the very errors it was giving in one case, which didn't indicate any severe problem and never caused this problem for that operator any other time, or they got told that there was a unspecified dose issue, but the computer told also told them only 6 of 202 dose units had been given. Even if that operator had known "dose input 2" could mean either a high or low dose was detected, they would've likely done the same thing as 6 was a low dose for a 202 prescription. Why would they have ever assumed anything other than what they did given that all the information they'd been given was pointing at doing what they in fact did? This was not user error except in that in an ideal world they should've refused to use a machine that errors constantly despite the patient inconvenience and potential to have trouble with their job, but they assumed it was error-filled because it was too safe (as it almost always underdosed), not because it was poorly made. In reality it is unrealistic for a human to be expected to do that despite overwhelming evidence supporting it as no big deal, and you only think its obviously a flaw in their behavior because of hindsight bias. This was the programmer's and AECL's fault (along with possibly their partner company CGR). You can't blame anyone but the manufacturer after they promise the safest systems, lie about making the machine 5x more safe, systematically untrain fear of errors in operators with spurious errors, trick the operators by displaying false values with false text indicating everything is "VERIFIED", lie to hospitals and regulators about there not being any reports of injuries or accidents so no official notice stating the machine has malfunctioned, sandbag the eventual level 2 recall in a way that makes it sound like there is just a fixable rare malfunction that may theoretically cause harm, do that useless fix and tell operators that they don't need to pay attention to Htilt messages or continue to do visual checks (which were only started under the recall, before that the operators were never warned about tilt messages), design a system with error handling so poorly thought out that it could even have unambiguous errors that dont have any way of idempotently identifying the problem, and give those poor operators useless documentation that doesn't tell you what any of 64 malfunctions actually even means. It makes no sense to blame the operators for not realizing the machine was broken in a dangerous way this one time (from their perspective) after the manufacturer did all those things that mislead them.
@taotoo2
4 жыл бұрын
@@Cutest-Bunny998 Sorry but if the dose is 202, you type in 202, then it says it's given 6, do you continue, knowing that you're potentially going to repeat that procedure 34 times, trusting that the machine (that you know to be unreliable) is correctly giving 6 each time?
@whiningmachine
4 жыл бұрын
I’d always wondered where Boeing got their inspiration for the safety protocols when they designed the code for the 737 MAX. Now I know. Thanks, PD
@Defosaur
4 жыл бұрын
minesweeper has better coding
@Cutest-Bunny998
4 жыл бұрын
Yeah in Minesweeper the software interlocks actually work. I have never accidentally clicked a bomb that was already flagged. Of course that doesn't mean theres no bug, but then again I'm not going to actually blow up a mine under myself if I incorrectly assume it is working without a formal verification. The consequences are a bit better if I just take the AECL approach on Minesweeper and say without proof that there is a 4x10^-9 chance of sweeper flag failure leading to detonation.
@deadfreightwest5956
4 жыл бұрын
Minesweeper is far, far more complex than an unshielded microwave oven's software.
@Lazar-w9u
4 жыл бұрын
Minesweeper is actually pretty complex,it has a decent graphics system and quite a lot of logic. This,on the other hand,is just a microwave without the shielding. The computer is useless
@Cutest-Bunny998
4 жыл бұрын
@@Lazar-w9uWell theoretically, if done correctly, the computer provides additional safety, as a fully electronic device would be using hardware interlocks only while if you add a software interlocks capable computer you add a source of redundant safety interlocking which also represents a different paradigm of interlock so it makes the overall interlock system more redundant and durable to faults that would only affect hardware interlocks but not software ones. But clearly this was not such a situation since the precondition of it being a valid and well-implemented software system is unsatisfied.
@anhedonianepiphany5588
4 жыл бұрын
@@deadfreightwest5956 Uh, microwave ovens don't produce _ionising radiation,_ and people who conflate _radio energy_ electromagnetic waves and accelerated electrons/x-ray energy, need to work on educating themselves about these basic concepts.
@kingjames4886
4 жыл бұрын
"well it can't have been us... you must have received a lethal dose of radiation elsewhere... however we aren't going to investigate that accusation."
@CheezyDee
4 жыл бұрын
Reminds me of the totally unintuitive, glitchy requisition software we use at work, except when I leave one of the 25 fields blank that should be automagically filled in from an ini file or registry key, it just throws a cryptic error message instead of killing people.
@anhedonianepiphany5588
4 жыл бұрын
I don't know what _you're_ doing wrong, but mine successfully fills in the blank fields _and_ kills people! (There's no error message, but a popup window updates the death toll.) Interestingly, attempts to power down the equipment during one of these 'glitches' seem to rapidly accelerate the fatality count.
@henke37
4 жыл бұрын
I think I heard that one of the remedies was physically removing the up key.
@Helladamnleet
4 жыл бұрын
Considering that was in no way, shape, or form the issue......
@neoqwerty
2 жыл бұрын
@@Helladamnleet no, he's right. AECL legit claimed "get rid of the up arrow" was a "fix". Let that sink in.
@thedungeondelver
4 жыл бұрын
"Hey, are these real x-rays?" "Good question! We'll get back to you on that." BZZZZZT BZZZZZZZT BZZZZZZZZZZZZZZZZZT
@garybarnes4169
4 жыл бұрын
"Arguably, the operators should have demanded equipment that operated fault-free" Your expectation of the response to "Operator Demands" clearly far exceeds mine.
@freebirthfreddy
4 жыл бұрын
Pretty sure they would have just hired an operator who "complained less." Sorta like the company culture that had a single programmer create the software. I can only assume that programmer was alone, disillusioned, and told to "just get it done"
@riibuns
3 жыл бұрын
the two incidents with the same operator is incredibly scary
@pyrotas
4 жыл бұрын
The small "m" in the animations has to be a capital "M" as I highly doubt that in the first case the patient got a 0.01V beam ;)
@laurencehoffelder1579
4 жыл бұрын
just a difference of 10^9eV XD
@spodule6000
4 жыл бұрын
I suspect Plainly Difficult may have written that software!
@RandornCanis
4 жыл бұрын
I don't know, she might have received a meV energy beam. The Therac even looks to be the right size for an 80s infrared remote. =D
@anhedonianepiphany5588
4 жыл бұрын
Uh, you mean like how you specified _volts_ instead of electron-volts (eV)?!? These are _separate distinct units_ that measure very different things, and they shouldn't be conflated!
@jsl151850b
4 жыл бұрын
So there was no 'put film where the patient goes and verify low is low and medium is medium' procedure. No dials, no "Now Emitting X-Rays" or "Now Emitting Electrons" lights. Radiation in the room not detected?
@BrilliantDesignOnline
4 жыл бұрын
Yup, microphone and camera turned off so you can't see the patient running from the machine and screaming "I have been NUKED", and the door is locked so no one gets irradiated by accident. "What is that pounding at the door of the treatment room": operator...
@deusexaethera
4 жыл бұрын
There's no need for such formalities. We know we designed the machine properly.
@anhedonianepiphany5588
4 жыл бұрын
Well, you wouldn't use "film" for such purposes anyway, but rather a detector/sensor probe of some kind, and these radiotherapy units were capable of fine increments in dose strength (not the low/medium/high type settings you'd find on a - _non-ionising_ - microwave oven). I'm not sure why you've mentioned "dials", though visual confirmation via illuminated indicators may have helped here, provided that they weren't controlled by the same buggy error-prone computer/software. There should be _multiple levels_ of redundancy when it comes to safety systems of such potentially deadly equipment.
@deusexaethera
4 жыл бұрын
@@anhedonianepiphany5588: Back in the 70's, film would've been the easiest way to check the calibration of the beam emitters, film having higher resolution than any electronic sensors at the time (and even most electronic sensors nowadays). Film also would've produced a physical record of the test results for documentation purposes.
@anhedonianepiphany5588
4 жыл бұрын
@@deusexaethera This discussion had nothing to do with beam alignment, but rather _intensity,_ which would be poorly recorded by film. Also, you're completely incorrect about the ability of electronic detectors of the time. I'm not saying this simply to refute you - _it's fact!_
@ItsBBP
4 жыл бұрын
A different type of radioactive incident, and this is a prime example of "race conditions" that can occur in software, and could have been prevented if the testing and software were more robust.
@DanielVidz
4 жыл бұрын
I'd hate the responsibility linked to a mission critical electrical circuit and/or interlock. I take pride in my work but the more basic, the easier to wrap my head around what's actually going on. Great comment, thumbs up!
@zorktxandnand3774
4 жыл бұрын
Exactly Also shows the importance of hardware interlocks, and readout of the actual status of the device, not just the set value or state.
@vinny142
4 жыл бұрын
It's not even a race condition, the software simply did not proces commands that where given during the processing of a previous command. It regisdtered the command, but did not act on it becuase the routine that was responsible for moving the hardware gave a busy-signal. Ofcourse any sane designer would add a failsafe step in which the software would probe the hardware to get it's physical position and display that on a screen. One of the first things you learn when wortking woth robotcs is that you *NEVER* assume that a mechanical object is in a particular position, you always measure it. Most modern robots will even simpy shutdown if the movement of an object is hindered or if you force it to move by hand; the system senses that the hardware is not doping what it commanded it to do and shuts down. But hey, better to keep the development process short so you can sell sell sell, amiright?
@disbeafakename167
4 жыл бұрын
Or if the operator looked into why it was giving an error message...
@DanielVidz
4 жыл бұрын
Hardware, software, firmware or even human error, a machine should make itself and its operations known ESPECIALLY when said operations are invisible like that of a radioactive nature
@callmerabbi
4 жыл бұрын
As always, super video. One teensy tiny concern: You state near the end that the operators should have demanded better equipment or something like that. Anyhow, the statement suggests that: A) Operators never did demand better equipment, or at the least bring up concerns with equipment reliability to whomever they would appropriately do so B) Such demands would even be met by clinic/hospital administrators C) Such demands wouldn't eventually cost said operators their jobs (because penny pinching administrators and executives are born evil) D) And so on, Some reference to Silkwood and to which nails get hammered. No doubt that operator complacency was at play here, but I'd bet dollars to donuts that there's a story out there about a radiologist or tech who tried and tried to get someone to listen to his/her concerns about this particular machine, and was possibly unemployed and miserable when it came out that he/she was right all along.
@michellejirak9945
4 жыл бұрын
Oddly enough, I was sitting my exam for becoming a medical device auditor while this video was posted. I can now officially offer my expert commentary. This event is one of the reasons why 1) medical device software validation requirements were implemented and 2) why the requirements for software validation are tied to the risk of the equipment, not the complexity of the software itself. Although, considering they wrote a whole new OS for the Therac, I would think that it would warrant the highest level of software evaluation regardless. This whole event is simply poor design coupled with a huge heaping pile of negligence.
@DanielVidz
4 жыл бұрын
This makes me appreciate the mouse cursor & selection a lot more than "hmm.. I think this is the right button"
@volvo09
4 жыл бұрын
Except for when you watch those people who click and drag the mouse and end up selecting a nearby field :o
@DanielVidz
4 жыл бұрын
@@volvo09 LMAO so true! That's me every time I use putty and it has right mouse button set to default paste. Crashed my cctv system over telnet trying to copy the list of commands lol
@jfan4reva
4 жыл бұрын
Burns and peeling skin are normal for radiation therapy, but not from a single dose....
@3rdalbum
4 жыл бұрын
Lots of detail in this video. Good research, thanks! This series of incidents has always fascinated and shocked me.
@PlainlyDifficult
4 жыл бұрын
Thank you!
@dan8t669
4 жыл бұрын
9:06 Error 54: Dose is either too *high* or too *low* WHAT? That's a big difference! Who would program an error message like that on a medical device.
@STARDRIVE
3 жыл бұрын
It´s like that ¨check engine¨ light. If you want to know what´s wrong, keep on driving till something gives.
@nerobernardino88
3 жыл бұрын
@@STARDRIVE So... The patient?
@yitznewton
4 жыл бұрын
I wonder how those technicians felt when the "now safe" units were redeployed and they had to do their first treatment
@ZGryphon
3 жыл бұрын
The staff of the East Texas Cancer Center (the one where the same operator accidentally killed two people in two months) refused to use theirs ever again, no matter what assurances they were given that it was fixed. The hospital had to return it to AECL and get something else.
@DanielVidz
4 жыл бұрын
You'd like to think a medical exposure to radiation would require dual key operation like that seen in James bond's 007 Goldeneye amoung others, when really it's just a single nurse saying for God's sake, why won't you just work... Computer says no
@pavelsovicka5292
4 жыл бұрын
Not to mention a typo (even when fixed in time) can send the machine into berserk mode.
@sarowie
4 жыл бұрын
how does a dual key system increase safety? In the end, both keys would end up in the two hands of one human and a dual key setup does not prevent the false system setup, as the visual indicators on screen do not display a difference between what was programmed and the state the machine is in. What helps is a hardware interlock system that checks settings against machine state and prescribed doses with applied dosage. (yeah, the measures dosis has a margin of error and it has to be very liberal, as both the machine output and the body absorption/reflection will vary - but an emergency shut off at double/tripple the planned dosage is still alot better then beeing orders of magnitude out.) I worked with X-Ray system and it had a hardwired door interlock, that was routinely and intuitively tested - if the operators wanted the machine to be in a safe state, we just left the door open. The motors acted up? Just open the door and X-Ray and motion system loose the authorization to even continue. Overwritting the hardware interlock? nope - impossible. I mean, you could tape over the door sensor, but who the hell would do that? If you want to be in the room with the X-Ray and Motion system active, just shut the door behind you - then it is obvious hows fault it is that he is standing there.
@jackkraken3888
4 жыл бұрын
That wasn't the problem here though. The creators relied too much on the poorly managed code and probably to save money removed the hardware interlocks. They assumed the code was good enough without checking first.
@erikahammer1054
4 жыл бұрын
They're not nurses, they're radiation therapists
@anhedonianepiphany5588
4 жыл бұрын
It takes only a fraction of a second to deliver extremely damaging, potentially deadly, amounts of ionising radiation. I cannot follow the, uh, 'logic' of your 2-key approach! What function _exactly_ do the keys activate?!?
@Neoentrophy
4 жыл бұрын
I like your Cue Dots to indicate ad breaks, brings back nostalgia of when I actually watched TV
@darkfent
4 жыл бұрын
Oh so the 25 refers to the MeV right? Oh no, it's 25 days left for the patients to be alive
@antisoda
4 жыл бұрын
The Therac-25-incidents are bloody terrifying. Not just the glitching but also the company's reaction to any questions regarding the faults. The "We can do no wrong!"-attitude is common in some companies even today. "The patients catch fire? That's normal. They jumped 10 meters into the air and scattered themselves over a large area? Hypochondriacs."
@MaxPower-gd2zr
4 жыл бұрын
"Surely there's nothing wrong with our software that was written by one person without much testing"
@peteraaron8626
4 жыл бұрын
AECL: Nothing to do with us. We were in the toilet.
@bificommander
4 жыл бұрын
So claimed their spokesperson, Diane T. Love.
@peteraaron8626
4 жыл бұрын
@@bificommander LOL!
@calvinhobbes7504
4 жыл бұрын
You mean "washroom" ...
@LUNUSt
4 жыл бұрын
Obviously it couldn't have been the radiation therapy machine exposing people to lethal doses of radiation. Must've been the commies... or a stray nuclear reactor wandering about
@Tiger-nm5cc
4 жыл бұрын
Wait, why is the radiation so hi-
@xiro6
4 жыл бұрын
because it needs to be much higher in X-ray mode than in photon mode.it has dual mode,thats why.
@kevin42
4 жыл бұрын
@@xiro6 doc. I cant feel my arm and there's a burning sensation in my shoulder. Is this norma....
@xiro6
4 жыл бұрын
@@kevin42 -doc,i touch my shoulder and it hurts,i touch my arm and it hurts,i touch my leg and it hurts,what do i have? -you have a broken finger.
@colonelgraff9198
4 жыл бұрын
Yes. As someone who’s studied both Nuclear Engineering and Computer Science this is a great topic...
@TinyTroglodyte
4 жыл бұрын
I found out about this when I was quite young in some poorly explained shitty top 10 list video. It scared the shit out of me. I've since read the Wikipedia article on it and looked into it more, but I could never find a decent video on it. This is a much better explanation that I have wanted for ages. So thanks for creating this.
@jeromewink557
4 жыл бұрын
I remember this on national news when I was a kid. You missed a detail about the up arrow. In the middle of the men dying they removed the up arrow key as they incorrectly thought that the problem was just the screen not updating the software. Until the magnet timing issue was discovered. There was also test where something like a glass jar of water was used as a stand in of a person and in maximum worst case scenario a visible arc could be seen flash through the jar.
@DanielVidz
4 жыл бұрын
Looks like a badass dough/kitchen mixer
@ethanlai1044
4 жыл бұрын
It mixes you innards
@TheJoeSwanon
4 жыл бұрын
1980s styling for you
@burleydad
4 жыл бұрын
Yes, it turns people into play dough.
@STARDRIVE
3 жыл бұрын
Dalek in disguise.
@evegreenification
3 жыл бұрын
I wouldn't let it within a 100 yards of my damn dough.
@_tyrannus
4 жыл бұрын
Your real talent is using funny animations and on-screen dialogue to lighten the tone, while you factually describe those ghastly incidents. Great work as always!
@majungasaurusaaaa
3 жыл бұрын
He's the opposite of Fascinating Horror. Both great channels.
@Zorro9129
3 жыл бұрын
86k rads, geez. That has to be up there with one of the largest radiation exposures.
@patrickbang3037
4 жыл бұрын
2:17 lmao it looks like a GIANT food processor like for when you need to make a 5ton bread
@blameTheDane
4 жыл бұрын
As a comp.sci. graduate, i really applaud your explanation of the computer system, you really did some reading on the subject. Well done mate!
@pavelsovicka5292
4 жыл бұрын
At 3:10 the value 25 meV should be 25 MeV. I am usually not "that guy" but here it makes for a fairly gigantic difference of value. A multiple of 10^9 if I am not mistaken.
@CalderaXII
4 жыл бұрын
"being that guy" would be if you said something about him misspelling the word "easier" 2:33. what you criticized is more then justified.
@Nicholas-f5
4 жыл бұрын
@@CalderaXII then
@sarowie
4 жыл бұрын
him mixing up units (meV or MeV is a typo not unlike the one he discribed as one of the core roots of this incident) happens regularly and I see it as an issue. Saying "em ee vv" is also a problem - it still is "Mega Electron Volt".
@pavelsovicka5292
4 жыл бұрын
@@sarowie Well thats the thing. Writing "Kw" instead of "kW" doesnt confuse anybody as electricity is not described in Kelvin-whatever (cause "w" isnt even a thing). So no reason to point a stick there. But here, "meV" is a valid unit just as "MeV". A little small but still a valid one...
@thatpersonwithamlpiconwhos2861
4 жыл бұрын
9:58 OPEN THE TREATMENT BAY DOORS I’m sorry, Dave, but I’m afraid I can’t do that
@Afishionado1
4 жыл бұрын
One of my college professors (when I was in school ten years ago) was an attorney-turned-programmer who was involved with the Therac-25 case. One of the stories from the incident goes that the company finally made one of the engineers sit and watch the nurses using the machine. He saw a nurse hit the "up" key and had a "What the hell did you just do?!" moment. It had never occurred to them that someone might do that. To echo another commenter, I believe that removing the "up" key from the keyboard was used as a stopgap fix. And yeah, the way AECL handled the situation was atrocious. Each hospital was allowed to think that they were the only ones having problems. Repeatedly, the engineers guessed at what was causing the accidents, made an update to fix the imagined version of the problem, and sent the update out with a notice that the unit was now safer than ever before. (At one point, they announced that the machine was now "ten times safer". I have no idea how they came up with that number.)
@ChiDraconis
4 жыл бұрын
There is a very much comparable extant situation; A "machine control device" intended to assist operator has a significant flaw that allows it to lockout any overrides-the design intent of the flawed assist is to remove natural dynamics from controls such that "point it there it goes there"
@Helladamnleet
4 жыл бұрын
The UP key wasn't the problem though... The fact they typed X to begin with was.....
@WildBluntHickok
3 жыл бұрын
@@Helladamnleet The fact that the machine didn't notice that the controlling computer had made a change was the real problem. Why are new instructions completely discarded when the machine is in mid-alignment? Can't it save them to a queue?
@jackkraken3888
4 жыл бұрын
I remember this case from CS class. One thing to remember was that these situations actually happened very rarely based on how often the devices were used. So it can make sense that the manufacturers thought it was a mistake. Furthermore like was mentioned there was no sign on the device how much radiation was actually administered so it further created doubts in the minds of the manufacturers. That being said if there are signs on the patient of what looks like radiation burns they should have investigated, more importantly they should at least have done so when the first patient died. I think the other problem IIRC was a replication issue, that you had to perform a set of tasks in the right order and time for the huge dose to be administered, and every other time the device would work as needed.
@anhedonianepiphany5588
2 жыл бұрын
All of the software issues aside, simple strip dosimeters or Geiger counters could have provided the important feedback they were lacking. These were hardly expensive additions and would have prompted much earlier serious investigation, preventing further death and misery.
@jackkraken3888
2 жыл бұрын
@@anhedonianepiphany5588 They also removed the hardware interlocks from the previous versions of the machine. Had they kept them this wouldn't have happened either. I guess they thought software could do everything.
@douglasjackson295
4 жыл бұрын
doctor: spams p patient: oh no
@takase5037
4 жыл бұрын
yes. The P button. man, I sure wish the override button is obvious and easy for people to press, definitely placed somewhere so obvious that people don't have to check for actual error before pressing it
@disbeafakename167
4 жыл бұрын
Seriously, at some point we have to just let evolution take its course and let the stupid ones die out. If you need a warning label telling you not to shoot the giant radiation machine when receiving an error code, you would probably die trying to blow dry your hair in the shower.
@a.dudeman7715
4 жыл бұрын
Disbea FakeName That only applies if the operator is the one getting hurt, but in this case, the ones getting hurt have nothing to do with the operation of the machine; they’re just receiving treatment.
@RCAvhstape
4 жыл бұрын
@@disbeafakename167 The operator isn't being punished for their mistake, the patient is.
@RyanTosh
4 жыл бұрын
@@disbeafakename167 You've got to be trolling, this is so unbelievably stupid I don't think I could imagine a human typing it.
@Motoko_Urashima
4 жыл бұрын
modern machines have a control panel that is NOT a keyboard next to the keyboard with overrides and such.
@atomsmash100
4 жыл бұрын
The machines shown look scary as hell, like somthing out of a "we're all doomed" movie.
@spazmaster6731
3 жыл бұрын
We studied the Therac-25 in my Operating Systems Concepts class in college to learn about the dangers of "race conditions" which are when a software process competes with another software process or a hardware process for completion and results in unpredictable behavior. The Therac-25 had major unchecked conditions in it's realtime OS that lead to it giving the wrong dosages and these were caused by race conditions. It's absolutely unacceptable for that machine to had been released when such little testing and review had been done on it.
@KarrierBag
Жыл бұрын
Only came across your channel a few months ago BUT I have to say, you from a cloudy south London you are doing a brilliant service, thank you.
@LordPyotr
3 жыл бұрын
As a radiotherapy technician seeing this for me is like ultimate hell. Great videos indeed mr Plainly Difficult!
@ZakharovProkhor
4 жыл бұрын
All software is broken. In computer science there's something known as the Halting Problem: "Can we guarantee that any given computer program ever stops?" The answer is no, we cannot ensure that a computer program ever reaches the line of code for "Okay, done." This is inherently the same problem of "Can we guarantee a program ever does or does not do X particular thing?" It can't be done except in narrow circumstances, even if (as seen here) a program works fine on one platform and you just want to use it on another. The only thing that really works is to consistently apply best practices (which are only developed after we make mistakes) and painfully exhaustive testing when safety is critical. Time and staff constraints means these are frequently if not usually ignored. When applied they only amount to "The program probably won't explode at the worst possible time." Even highly experienced outfits with gargantuan budgets can make critical mistakes like Boeing and the 737 Max. Actually the 737 Max might be a good topic to cover.
@pdorism
4 жыл бұрын
Actually it is possible to build programs that are provably correct. The halting problem only says that it is not possible to write a program that correctly identifies the halting conditions for any given program.
@ZakharovProkhor
4 жыл бұрын
@@pdorism I didn't say it was impossible to build some correct programs, but that it's impossible to prove for "any" given program. And with formal correctness there's also a confidence problem, that is you can't necessarily trust that your formal description fully covers the actual problem. Especially when that problem is a physical rather than purely logical one. We just can't make repeatable systems that verify our other repeatable systems don't have critical flaws.
@lewribaedi5997
4 жыл бұрын
Excellent video, but you should clarify whether you mean milli electron volts (meV) or mega electron volts (MeV) when you read out the units to avoid any confusion
@dosvidanyagaming4123
4 жыл бұрын
The first time he did, then he assumed people are smart enough to follow the thread
@lewribaedi5997
4 жыл бұрын
@@dosvidanyagaming4123 ahh, well I should pay more attention then. Personally I still prefer reading the full thing though rather than the abbreviation
@georgeullrich9086
4 жыл бұрын
What a civil comment section interaction
@gunnish1337
4 жыл бұрын
yup meV is milli electron volt..
@Cutest-Bunny998
4 жыл бұрын
"Beeper blooper, I'm racing to kill ya' " - Therac-25
@EvitiniaNetTV
4 жыл бұрын
That episode hit me differently. It feels quite unsettling
@deadfreightwest5956
4 жыл бұрын
Patient: "Doctor, I'm here for my treatment." Doctor: "Excellent. We'll put you on Therac." Patient: "Um, the... rack?!" Doctor: "Don't worry, we're what some term, 'professionals'!"
@STARDRIVE
3 жыл бұрын
Doctor: Please lay down on the bed with the burn marks. Patient: Do you expect me to talk? Doctor: No mr. Bond, ...
@EthanB
4 жыл бұрын
First time hearing about Yakima in a video like this. Interesting piece of history I never knew about.
@deadfreightwest5956
4 жыл бұрын
They wanted a piece of the action. Everybody downwind of Hanford got irradiated, after all. ("Atomic Harvest" by Michael D'Antonio)
@hillaryclinton2415
4 жыл бұрын
That the machine can even generate this level of radiation is down right scary
@Stoney3K
4 жыл бұрын
Even more scary is that it's effectively just a giant television tube...
@ravenID429
Ай бұрын
It’s because it had different modes where one was more concentrated
@wilting_alocasia
4 жыл бұрын
Freakin love your use of ad warning lines - like man you're amazing for doing that!
@jondobbs69
Жыл бұрын
I discovered this topic through the video that Kyle Hill made, which was perfectly fine. I definitely do like the drama that he adds to his productions. That being said, I appreciate SO much that you go into the technical aspects of these topics quite a bit more than a lot of other channels would bother to. Much respect, my brother from across the pond.
@Gallalad1
4 жыл бұрын
When I was doing software development in college this was a literal case study
@stargazer7644
4 жыл бұрын
As opposed to a figurative case study?
@WildBluntHickok
3 жыл бұрын
I bet some airplane stuff was in there too. "Cases where failure could mean civilians die".
@ninjasiren
4 жыл бұрын
I am an Information Technology student and an Associate in Computer Science graduate. This is 100% software problems. No additional layers of safety on the software.
@sarowie
4 жыл бұрын
That is the fun part: You as an Computer Science graduate call that a software problem. I as an exprineced system engineer call that "bad system design" and "horrible R&D management". Neither Software nor Hardware should assume that the other party is working correctly. Both systems (and each subsystem) need to assume that things go wrong. Like for example: Imagine the magnets/motors that switch between modes are stuck. Having an read out on the system would have allowed the software to check if its settings where applied correctly and are now reflect in the machine state.
@ninjasiren
4 жыл бұрын
@@sarowie I don't touch hardware stuff. My main focus on my studies was mostly on the software side. But yes in general, the whole system is bad, not enough R&D was done to fix problems, and incompetence of the manufacturer to accept the problems. I just don't like fixing hardware problems. It's much easier for me to do software problems than hardware.
@smort123
3 жыл бұрын
I get really strong "Get back in the robot Shinji!" Vibes
@p2p2p2p2p2p
3 жыл бұрын
lmao
@latergator9622
3 жыл бұрын
Hey man just wanna say I love everything you’ve made. Even if it didn’t really capture my interest at first you’re very calming style is what keeps me. Keep it up man 👍
@hotlavatube
4 жыл бұрын
The Therac-25 was a stressed case study in my computer science curriculum. While they didn't hold the software developers liable for the Therac-25 deaths, another series of radiation dosage deaths/injuries in Panama around year 2000 resulted in software developer Multidata Systems International being pursued in the courts in both the US and Panama. Apparently the hospital couldn't afford the manufacturer's software and used an after-market software by Multidata. The medical physicists then modified the radiation shield setup which triggered an obscure bug in the software which occurred when the shields weren't installed in a specific order. This error condition caused 28 people to be given radiation overdosages of which at least 8 died as a result (21 of the 28 died, but there's uncertainty about whether the cancer or radiation got them first). So, the allegations were, in part, the software was poorly documented and failed to ensure safe operation under all possible conditions.
@salat
4 жыл бұрын
I see there's a little dose error in this video too: Actually that thing applied MeV (MEGAelectronVolt = 10^6 eV) vs. meV (MillielectronVolt = 10^-3 eV). No worries, It's just a factor of 1000000000...
@chiaraj1003
3 жыл бұрын
I mean, it's fitting, isn't it?
@Choconillaaa
3 жыл бұрын
This channel almost always has numerous typos per video, unfortunately.
@garymeakin
3 жыл бұрын
In addition to a disaster rating, which this is relatively low, I think the introduction of a blunder rating would be complementary, and, this would hit max.
Пікірлер: 2 М.