DRMacIver's Notebook

Safety Advice and Judgment in Emergencies

Published

2024-01-09

Another draft bankruptcy post. I don’t know why I didn’t finish this one. I barely remember writing it - it’s from a full year ago - and it looks pretty close to complete to me.

Please note that improper application of what follows can literally get you killed.

This is not a joke. I’m about to explain how safety advice works, and the reason it works the way it does is so that you don’t die. If you read this article and go “Ah I see, safety advice is for dumb people, but I am smart and know better so I can disregard it”, you will probably die.

I am emphasizing this because that is absolutely a risk factor associated with this article. I would like you not to die and, equally importantly to me, I would like not to get blamed if you do.

In fact, my official adviceIn my very official capacity as that guy wot writes on the internet. is that you should always follow safety advice to the letter. I’m now going to explain to you why this isn’t always a good idea, and give some examples where I have not done so. If, based on this, you sometimes don’t follow safety advice, please don’t blame me. This newsletter is for informational purposes only etc etc.

So, let me be clear: Under normal circumstances, you should follow safety advice. Under moderately abnormal circumstances, you should follow safety advice. Under exceptional circumstances, after a careful balancing of the trade offs and proper consideration, it might be correct sometimes to not follow safety advice, but as far as my advice is concerned, you should follow safety advice and if you don’t then that’s on you.

Now, with those caveats aside, let’s talk about how safety advice works.

Back in 2018, I went to A&EAccident and Emergency. British English for the ER / Emergency Room. twice in one week.

Why?

Well, at night I was having a bit of trouble breathing. I wasn’t gasping, I wasn’t really struggling for breath, but my chest felt a bit tight and it felt like my breaths were never quite sufficiently deep.

So I called 111, the non-emergency medical advice helpline, and had the following conversation:

Me: (describes above symptoms)

Operator: Yeah you need to go to A&E right away.

Me: Are you sure? It doesn’t feel dangerous, just unpleasant.

Operator: We’re sending you an ambulance right now.

So I went to A&E, waited a few hours in the middle of the night, they duly did an ECG, found nothing wrong, and told me I was probably fine. I got told to take some ibuprofen and an antihistamine and it’d probably make the symptoms go away.

Then it happened again a few days later, and basically the same story repeated.

The third time this happened I did the obvious sensible thing: I didn’t tell a medical professional about it.

Also at that point I figured out what was triggering itA buckwheat pillow I’d got for correcting neck posture at night apparently set off some sort of allergic reaction, though I am not as far as anyone can tell allergic to buckwheat. and removed it from my bedroom and after that I didn't have the symptoms again.

But, if I had called 111 for advice that third time, I would have found myself in an ambulance, because as far as 111 is concerned, if you’re having trouble breathing, you go to A&E.

This is a completely reasonable thing for them to do, because they are not in the position of doing a detailed medical diagnosis, they are in the position of dispensing safety advice.

Safety advice has three key properties:

1. It is designed to help you avoid situations in which you die, or other terrible things happen to you.

2. It is designed to be used with minimal information or skill, on either the part of the person receiving it or the person giving it.

3. It has a very limited set of concerns and treats everything outside of those concerns as irrelevant.

When you call 111, you are not talking to someone who has a medical degree. It’s a standard call center job with standard call center job requirements. The entry requirements are as follows:

• Experience of dealing with members of the public over the telephone

• Ability to remain calm when working under pressure

• Ability to work to defined policies and procedures

• Ability to work on own initiative

• Age 18 years or older (in line with legislation)

There is presumably on the job training, but they are very much not medical professionals, they are there to go through a decision tree as to whether there is something simple you can/should do, or whether you need to escalate to a medical professional - either “go to the GP”, “go to A&E”, or things of that ilk.

In this decision tree they are working through with you, some of the branches involve you dying. They don’t want that to happen.Most importantly, they don’t want to get blamed when that happens.

You can think of this part of the decision tree as the big that decides “Do I send them to A&E or not?”. There are two types of errors they can make here:

• They can send you to A&E when they don’t need to.
• They can not send you to A&E when you do need to.

In the second one, you probably die, and they almost certainly get in trouble (either personally or as an organisation).

The first isn’t free. If it were free they’d send everyone to A&E. But it’s a lot preferable to you dying.

Basically, there are three costs here:

1. Risk to you.
2. Burden on medical system.
3. Inconvenience to you.

From the point of view of the 111 operator, these are in decreasing order of importance, and their importance decreases sharply with each item - they very much don’t want you to die, they don’t want to create unnecessary work for doctors, and also they probably care a little bit about not inconveniencing you but let’s be honest it’s barely on their radar.

This means that in any circumstances where the risk to you is high, they will tell you to go to A&E, and they will right to do so. In particular when there is a non-negligible chance that you might literally die, they will tell you to go to A&E.

Now, here’s the wrinkle: “chance that you might die” is not purely an objective property of the world. There’s “genuine” uncertainty - given the true facts about your state, how at risk you are - but there’s also epistemic uncertainty - given what we know about your state, how at risk you might be.

What you want in an ideal case is to make all the judgements based on the genuine uncertainty. If you’re actually at risk, go to A&E. If you’re not actively at risk, take care of yourself at home and see your GP when necessary.

The problem is that you don’t know the genuine uncertainty. You’ve got some approximation of it cobbled together by your (unreliable) self reports over the phone and someone who’s had a couple of weeks of training and has access to an internal knowledge base which probably isn’t much better than the NHS website.

What this means, is that they are very very strongly incentivised to err on the side of caution. They have simple rules, and rather than try to refine them, they offload them to people who are competent to do the actual risk assessment. Thus most of the uncertainty in their assessment is epistemic, not genuine, and they hand you over to a medical professional to reduce it when it seems like it needs reducing because it looks uncomfortable far from zero.

In the case of difficulty breathing, this sounds sufficiently like I might be having one of several rapidly fatal problems, that they very much want to reduce that epistemic uncertainty down to zero and find out if I’m actually going to die. Off to A&E with me.

This is good. It’s how the system is supposed to work. I want to be clear on that.

BUT it does mean that you can make better decisions than the system under two particular types of circumstance:

1. You have significantly more applicable knowledge than they do.
2. Your preferences are significantly different than the ones they’re optimised for.

In my case with the A&E visits, both applied.

I at this point had significantly more applicable knowledge than they did - I’d talked to doctors about it and they’d examined me. Twice. I literally had more relevant medical knowledge about my problem than the operator at 111 could ever acquire, because I’d talked to medical experts about this.

My preferences were significantly different than theirs - being in A&E for 3 hours starting at midnight sucks. I didn’t want to do it. They didn’t care about this.

It’s tempting to say that they should be able to have a conversation with me that takes these preferences into account. They shouldn’t! That’s not what they’re for. They’re a human interface to the system. If I want to have a proper conversation about the problem, I need to talk to an expertThe experts also tend not to be very good at respecting patient preferences and knowledge, but that’s a separate problem., the phone operator, in trying to act as such an expert, is likely to make significant errors, and those errors are likely to kill people.

The solution, unfortunately, is to not put them in a situation where they are forced to give you bad advice. This is not always available, and it requires you to have good judgement about when that is.

Anyway, I’m writing all this because I ran into another problem like this recently, in which I flagrantly ignored safety advice, which is that my house had a gas leak recently.

We’d had a smart meter fitted on WednesdayEditor’s note: This was written over a year ago, so the Wednesday in question was in December 2022.. The person who fitted it did a sufficiently bad job that the person who fixed his work had a legal requirement to report him.

We noticed this problem on Friday when I noticed that operating our gas stove caused a strong smell of gas in the living room. I turned off the stove, checked the meter cupboard, and sure enough there was an even stronger smell of gas coming from there.

I turned off the gas, which solved the immediate problem, and called the national gas emergency line, as one does when you have a gas leak.

They informed me that a gas engineer would be with me as soon as possible (“as soon as possible” turned out to be about 50 hours later. We did not get gas restored until Sunday evening), and then very insistently informed me that we must:

• Not operate any electrical appliances.
• Keep doors and windows open to ventilate the property.
• Stay out of any attics.

I earnestly informed them that yes I would absolutely do this. Then I opened the back door, waited a bit, verified that I could not smell gas anywhere, and turned on a space heater.

According to the official safety advice, I’d just behaved horribly irresponsibly and was now going to blow up my house. This didn’t happen, in case you’re worrying.

Why did I behave in such a seemingly flagrantly irresponsible way?

Because:

1. It was -6C outside (about 20F).
2. There was no longer a gas leak, because the gas was off.

From their perspective when writing the safety advice:

1. Gas leaks are very very bad, and you should not trust inexpert judgement as to how bad a particular gas leak is.
2. The cost of gas leak mitigations is relatively low, because there will be an engineer with there to deal with them rapidly.

From my perspective:

1. I was extremely confident in my ability to assess how bad this gas leak was, because I knew exactly where it was occurring, I had turned off the gas supply, ventilated the property, and sniffed around to confirm a total absence of the smell of gas.Technically of mercaptan, the additive they put in gas to make sure you can smell it.
2. When someone tells me “I can’t possibly comment” when I ask them whether “As soon as possible” means “Some time today”, I know I’m in for a long wait, and I was absolutely not going to have a long wait in sub freezing temperatures with no heat.

The result was that the next 48 hours, while extremely unpleasant, were not actively dangerous to my health. If I had followed their safety advice, it would have been actively dangerous.

Honestly, this is not good safety advice. It might be good safety advice in the normal course of things where you really can count on a two hour response time, but even then it’s not great.

The problem is that this had two assumptions about the world encoded into it:

1. I am incompetent to assess whether there is a gas leak.
2. They will respond fast enough that any inconveniences are minor.

The first assumption is, honestly, probably fair enough. Gas is scary and most people are indeed incompetent. I was happy to trust my own judgement, but they probably do not have the information to determine that my judgement is trustworthy.

If their advice was “If you think it’s safe, treat it as safe”, this would probably get more people killed, because most people probably don’t have judgement you can trust.As this whole episode shows, you can’t even trust gas safety certified engineers to reliably have good judgement.

But, eh, regardless of whether they trust my judgement, I trust my judgement! And I’m pretty sure I’m right to do so. The problem is, of course, that many other people who also think this about their own judgement do so incorrectly. This is part of why my official advice is not “trust your judgement”.

I don’t really know what to tell you about how to tell whether you should actually trust your judgement, but one partial test is that you know when not to trust your judgement. For example, I was at the time basically 100% confident that I knew what the problem was and that I could fix it. I was proven correct in retrospect too - the problem was exactly what I thought it was, and doing what I thought would fix it would have indeed fixed it.

I sat on my hands for the full 48 hours and refused to touch it. This was correct. Everything would have been fine if I’d tried to fix it, and I knew that it would be fine if I tried to fix it, and I also knew that I would still think it would probably be fine in many scenarios that end with us being blown up or poisoned.

Afterword

I think all that stopped me finishing writing this piece was I didn’t have a very satisfying way to wrap it up other than to say “learn when to trust your judgement, butplease don’t get yourself killed”. So uh, learn when to trust your judgment, but please don’t get yourself killed.