Richard Feynman Side Hustles
twitter.comIf you want to read the replies without an account:
I had to read this twice to understand it. Stated succinctly, it sounds like the company's sensor measured the rate of flow of oxygen through the sensor, which would give a reduced reading if the cross section is obstructed.
Feynman's sensor, by contrast, directly measured the concentration of oxygen in the sensor, which gives the same result every time once the sensor is at equilibrium with the environment.
People are giving such bizarre examples for why it helped.
Just think of a thermometer.
If it removes heat as it measures it (consumes oxygen) then it will measure everything too cold if the system can't replace the heat that's removed (this is like having an insulated thermometer).
If your thermometer replaces heat as it removes it it solves this issue.
When is this an issue for a thermometer? If your thermometer is too large in terms of heat capacity for the objects you're measuring the temperature of.
Off-topic, but I was interested to read Carl's Twitter bio[0], "I’ve spent a lifetime switching my Special Interest every year or two. By now I’m surprisingly knowledgeable in a lot of fields— a skill now obsoleted by AI."
That hits a bit close to home!
I feel like the company might have been Yellow Springs Instrument (YSI, now a division of Xylem).
The dissolved oxygen sensor (the Clark electrode) was invented by Dr. Lee Clark at Antioch University (Yellow Springs, OH) and commercialized by YSI in the 1960s. A friend of mine worked at YSI from the late 60s thru the 80s on biosensors (glucose and lactic acid, using the Clark electrode as the basis) and worked directly with Dr. Clark.
Carl Feynman was born in 1962, according to what I'm reading, so if he was 14 that would have put this story in the time period early in the commercialization of these sensors.
Some good background:
https://derangedphysiology.com/main/cicm-primary-exam/respir...
https://www.anaestheasier.com/the-clark-electrode/
I'll have to hit my friend up to see if he knows anything about this. I'm certain he'd get a kick out of reading this, if nothing else.
Saw this twitter response,
> Do you have a similar speaking cadence as your dad? I can almost hear this in his voice.
I experience the same. Wonder if Carl asked chatGPT to write it in Richard Feynman style? :/
I troll; Regardless, it made me happy to hear Richard in Carl.
This would be cool if only it made sense.
To use an analogy with some metaphors: The sensor is like a sealed room with a screen window that only lets in oxygen. To get a reading, every molecule that enters is smashed to create a tiny spark of electricity. However, because the oxygen is destroyed to create that spark, it creates a suction effect, causing more oxygen to rush into the room to fill the void. This creates a major flaw: if gunk builds up on the screen, it slows down the flow of incoming oxygen. The sensor, which only counts sparks per second, is tricked into thinking the oxygen level outside is low, when really the window is just dirty.
By adding a third electrode to replace the oxygen every time one is smashed, you maintain a perfect balance and eliminate that suction. Because the room stays full, the sensor no longer relies on the speed of the oxygen rushing in; it simply measures the steady state of the oxygen already there. Even if gunk gets on the window, the sensor won't be starved of a reading. It might take a few extra seconds for the levels to settle, but the final number will be 100% accurate because the sensor is no longer emptying its own room to get a count.
I still don't get it. The outside is dirty, right? He said in his post "You dip this probe into beer, sewage, or canned food a-stewing". So when you say "when really the window is just dirty" I don't get it - yes it will always be, because that's what it is placed in, no?
A dirty window only ruins the reading if you are measuring the speed of the oxygen passing through it. The three electrode design stopped measuring speed and started measuring balance. Unless the gunk is a total airtight seal (which is rare on the scale of an oxygen molecule), the sensor will eventually reach the right answer, whereas the old version would fail.
So dirt as a factor that clogs up the sensor does not play into it at all? It's all just about moving it into different environments to measure?
The permiability of the membrane would still be reduced by stuff on it, but as long as there is any permiability, the inner compartment will reach equilibrium eventually.
The big gain comes from a change in how you interpret the presence of electrons.
The older approach converted oxygen to electrical current, the magnitude of current flow relating to magnitude of oxygen depletion. The assumption built into that approach is that low oxygen depletion levels meant low oxygen levels, but that wasn't the only potential cause, because it ignored variation in the permiability of the membrane.
The newer approach equates current flow to oxygen concentration, as the system doesn't deplete the concentration any longer. The permiability of the membrane in this setup only contributes to a longer initial delay as the inner chamber comes to equilibrium with the surrounding concentration.
I think maybe one thing you have to consider is that sensors still require maintenance. Software can measure the length of time the sensor requires to reach equilibrium and send a maintenance required alert and someone cleans it (like if the software expects equilibrium in 10 seconds but the reading settles at 60 seconds, it can calculate the sensor is 80% clogged and requires cleaning). There's also all sorts of techniques that can be used to mitigate gunk depending on how the sensor is being used such as physical wipers, air-blast systems, ultrasonic cleaning systems, and chemical coatings. So as long as some oxygen can get in and an equilibrium is made between the fluid outside the sensor and inside the sensor, you'll get a reading that you can trust.
The sensor is normally placed in a dirty environment. This change prevented the need to calibrate the sensor for the level of dirt.
This is a much better explanation. Thank you
I agree. It's not clear how adding a sensor "so that it adds back an oxygen molecule" works. shrug
I think this was primarily about speeding up the measurement time. With just two electrodes you had to wait for the device to achieve equilibrium with the material being measured. If the concentration of oxygen on the probe side of the barrier was higher or lower than the material side you would get false measurements, particularly in low oxygen scenarios because you have oxygem trapped in the probe.
By keeping the state of oxygen inside the probe constant and replacing consumed molecules you now can measure almost instantly.
Yes but how do you do that? that magical third electrode sounds harder to make than the original problem.
Edit: I think I get it now, it's a chemical reaction. By applying a voltage with some polarity to the 3rd electrode you can run the reaction in reverse. Still very hard to achieve because you have to make sure the reactions happen at the same rate with the same efficiency, which is far from trivial. This must be a very high end sensor for all this effort to make sense.
An oxygen molecule does some chemical reaction on the sensor electrode that releases an electron, maybe it's made of iron and turns into rust. If you supply the same current to another electrode to do the opposite reaction, maybe one made of rust that turns into iron, it balances.
The sensors must be consumable with a certain lifetime.
Because then it doesn't alter the side of the membrane where it does the reading (plus one minus one equals zero). That makes the measurement more accurate.
Specifically, if you assume a partial pressure of Oxygen and of all other gases on the electrode-side of the diffusion membrane, then you'll only see a certain number of "ionization events" per time, and you're limited in how much electrical signal you get by how fast oxygen can diffuse across the membrane. This is likely driven by maintenance of a partial pressure within the membrane. However if you re-ionize the oxygen that you deionized, then the partial pressure is much closer to equilibrium, and therefore the partial pressures are only dependent on the amount of oxygen outside of the membrane instead of being dependent on both the ionization rate and the recovery rate through the membrane. It probably makes the calculation a lot faster and more closely dependent on the environmental presence of oxygen which is what you want.
You're not really making things clearer.
What does "adds back an oxygen molecule" mean?
It means you do an electrochemical reaction that releases an oxygen molecule, like the original explanation said. It doesn't really matter what reaction it is, but it could for example be electrolysis, where you split 2x H2O into 2x H2 and 1x O2.
The point is this reaction is reversible. In one direction, you end up with fewer O2 molecules than you had before. In the other direction, you end up with more.
That's an implementation detail no? Are you asking how to add an oxygen molecule, or how this makes the sensor better?
Yeah, how do you add the oxygen molecule, and how do you know when you have to do that?
Elaborate and you'll find the issue with this setup.
How do you add the molecule? Well, you're not just dealing with single-digit numbers of molecules. Have an oxygen tank with a flow meter for example, open the valve to release the required volume of oxygen. The ideal gas law tells you how many molecules you let out.
How do you know when you have to do it? The sensor tells you how many oxygen molecules you consumed, as a proportion of the current flowing. So just let oxygen flow into the tank at the same rate as you're consuming it. Which you know because the device literally measures how much oxygen it is consuming.
I think the real issue is that the explanation in the tweet is from a physics perspective rather than an engineering one, which means it reads like it was implemented with impossible magic.
>open the valve to release the required volume of oxygen
Mega LMAO. I can assure you this is not what's going on, at all. Also, if you release oxygen in gas form into the liquid you're going to run into a zillion other problems.
One of the golden premises of measuring things is to avoid altering what you're measuring, lol.
The issue here isn't the setup, it's with people understanding it.
Enlighten us, then, savant[1].
1: as in, one with detailed knowledge in some specialized field (as of science or literature)
There are a lot of explanations elsewhere in this thread. If you've read all of them and still don't understand, I don't know how to help you further.
Mainly I take issue with the person I replied to implying "I don't understand the solution, therefore there must be some functional issue with the solution."
The current is measuring the rate of the reaction. With the two-terminal design the rate of the reaction is proportional to the rate of diffusion of the oxygen into the area where the reaction is taking place, which is related to the oxygen concentration around it but also can be affect by other things. With the third electrode, the current is proportional to the concentration of oxygen in the area around the sensor directly, which will equalise with its surroundings much more consistently than the rate of diffusion.
(A quick google brings up this document which describes the principle. No idea if this is the company in the story: https://semeatech.com/uploads/Tech_Docs/AN%20161205.pdf )
This way you're measuring change in oxygen concentration. As more oxygen comes into the compartment in order to equalize with the outside you consume and at the same time produce more oxygen. You measure the change in rate of oxygen consumption/production. It is always consuming/producing oxygen but the rate changes with the concentration.
At least that's what I assume.
I think of it differently.
Before, you measured diffusion rate of oxygen and inferred oxygen concentration from that (the concentration outside the chamber is always greater than the concentration inside). Dirty membranes etc all changed the rate of diffusion, which caused issues.
After you measure oxygen concentration directly (the concentration inside and outside the chamber are always the same).
Trust me, if we all understood Richard Feynman the first time he said something, the world would be a very different place.
That's very cool and edgy but (Richard) Feynman didn't say this.
Had to read it 3 times but it makes sense
So do you have to be a god tier Nobel Laureates to get this kind of gig where you just learn about a business and then offer random suggestions that might or might not help them and charge obscene fees for the privilege?
You definitely don't have to be god tier anything, you just need to know at least a little more than the companies you are consulting for.
This kind of work has been my primary income for the last 4 years or so. Nowhere near on the same level as Feynman, but I know enough about enough other things that I get a lot of reputational referrals.
>you just need to know at least a little more than the companies you are consulting for.
sometimes (i'd argue often, actually), you don't even need that. simply having an outside/fresh perspective and the fact that you aren't part of any of the existing groups/silos is valuable.
Often the most useful thing is just listening to the right people in the company. I wouldn't be 100% surprised if someone in the company in the story had already had the idea for the third electrode, but it took the suggestion from the high-paid consultant to get it taken seriously.
Probably true, but to get the job in the first place you probably need some sort of showy, impressive credentials.
I imagine you can also start by doing the same thing for a low cost, or for free. Find a local business that’s interested, give your advice, build reputation, repeat.
I think the story sounds fake because they listened to him.
Having the ideas is easy. Persuading and organization to change is not.
Perhaps it’s a cultural difference between the middle of the 20th century and now.
Often the highly paid consultants are there entirely to get the organisation to listen to the right ideas that already exist within the company.
That's basically what happened with Feynman's involvement in the Challenger enquiry, as he himself freely admitted in his memoir.
Why would a small company CEO hire a famous consultant only to ignore his suggestions? Absolutely not evidence of it being fake.
Not really. Just need to be really good at your shit and cut through pointy-haired BS.
Nope! There are consulting companies all over the place filled with bids and not filled with Nobel laureates!
Ergo...
None that offer that level of work life balance though…