Will http://www.ncbi.nlm.nih.gov/pubmed/25525213 change deco procedures?

[shoredivr]

Great summary, UWSojourner.

It's a real study in human emotions to watch people turn themselves inside out to hold on to preconceived ideas. Once divers get a chance to read the thread and absorb the implications, they can reflect on whether to apply these concepts to their own deco diving.

 

[kr2y5]

VPM only works up to a point (...). That admission is extremely odd

Not sure why you think that's odd, pretty much every model is a simplification and works only up to a point.

The reason he didn't want to admit

You can't possibly know what someone's reason might have been unless you can read their mind, or unless they themselves explicitly stated those reasons. Until then, it's an accusation, not a fact, and as such, it doesn't advance the discussion in a constructive way.

not a comfortable outcome for deep stop proponents.

Not sure why you think it's not a comfortable outcome, could you elaborate?

 

[shoredivr]

Not sure why you think it's not a comfortable outcome, could you elaborate?

Have you read the thread on RBW? My money is on the bolded statement.

The reason he didn't want to admit VPM-B+7 is a valid VPM profile is because it is VERY similar (not identical) to the A2 deeper stop profile tested by the NEDU (which is part of what the deep stops thread showed). The A2 profile was considerably more risky than the A1 shallower stop profile that was also tested. And, it turns out, the A1 profile was very similar to GF 53/53. So we have actual dive trials showing, at least for that case, that a GF profile would likely have performed considerably better than one generated by VPM-B … not a comfortable outcome for deep stop proponents.

But, it's all here for the reading, and a good presentation from Dr. Doolette is here.

 

[UWSojourner]

Not sure why you think that's odd, pretty much every model is a simplification and works only up to a point.

Please read the Deep Stops Thread … all provided there. But just to give you perspective, Erik Baker's VPM-B program checks valid ranges. In it (you should be able to find it online) a test is made to ensure that the critical radius of N2 is within [0.2,1.35] microns. VPM-B+7 is about 1 micron.

You can't possibly know what someone's reason might have been unless you can read their mind, or unless they themselves explicitly stated those reasons. Until then, it's an accusation, not a fact, and as such, it doesn't advance the discussion in a constructive way.

That's parsing the words pretty closely. But ok. How about "The only possible explanation I can find as to why Ross and others would be willing to undermine the core of VPM-B's theory, in the context that we were debating, was to avoid admitting that the VPM-B+7 profile that we were looking into was valid."

I'm not sure why you think it's not a comfortable outcome, could you elaborate?

If it weren't particularly uncomfortable, I doubt we would have sparred for 1300+posts over the NEDU study. I do think it damaged their case more than if they'd let it be. When I started on the deep stop thread I was in the "I can't see how those NEDU profiles are much like anything I've seen" camp. If it had been left at that I'd probably still be there.

I looked into it because Ross seemed so certain it meant nothing and I had some programming skills that could look into some of the objections he raised. So during that thread I went from fairly agnostic to pretty convinced the NEDU study was on point for the kind of dives recreational technical divers do.

 

[kr2y5]

Please read the Deep Stops Thread … all provided there.

I only skimmed over that long thread, here is what I found. I would be grateful if you could help me by pointing out where I got it wrong.

1. A study was performed that asked a practical question in a specific context. The study contained a conclusion that switching to deep stops at this point was not supported by data. A lot of people misunderstood the methodology used in that study and took that statement out of context, to mean that deep stops are bad in general. Some people were much more cautions, and suggested that maybe deep stops are not as good as everyone thought they were. Everyone wanted the result to mean something they can apply in practice. To many, these sorts of general conclusions strongly resonated with their subjective experiences.

2. Of a continuum of possible profiles or exposures, a single one was tested (which is understandable, for anything beyond this would be impractical). On the other hand, the kinds of general conclusions that everyone was eager to draw required an assessment that applied much more broadly. In order to draw the broader conclusions, it became necessary to argue that the data point collected was highly representative of the entire parameter space (to the extent that you can ever make such a claim). A question was then raised to what extent the profile was representative.

3. The question whether the profiles tested were representative involved a great deal of subjective judgement. No prior agreed upon definition of "representative" existed. Some approaches to quantifying it have been proposed, but contested. As far as I could tell, it all involved a degree of eyeballing. Since it couldn't be objectively defined, quantified, and measured in the way that everyone would accept, the broader claims could not be either proven, or refuted by data. Or rather, the competing "proofs" were based on assumptions or methodologies that were themselves open to subjective judgment, and that would not be universally accepted.

4. Swimming in the ocean of subjectivity, and unable to find common ground, the fiercest opponents in the battle turned to questioning each other's credentials, honesty, and integrity, and proceeded to poke each others' eyes out. Those with their eyes poked out felled compelled to respond. After all insults have been exhausted, friends came in for reinforcement, and opposing factions formed. Resentments continue to this day.

What did I miss?

That's parsing the words pretty closely. But ok. How about "The only possible explanation I can find as to why Ross and others would be willing to undermine the core of VPM-B's theory, in the context that we were debating, was to avoid admitting that the VPM-B+7 profile that we were looking into was valid."

That sounds so much better, although I still think that Ross's motives should be irrelevant to the core of the technical discussion.

If it weren't particularly uncomfortable, I doubt we would have sparred for 1300+posts over the NEDU study.

That all just sounds like a misunderstanding, IMHO.

 

[Centrals]

But, it's all here for the reading, and a good presentation from Dr. Doolette is here.

Excellent, thank you.

 

[lowviz]

I'd like to take a brief step back, please.

I’ve been trying very hard to follow this thread for insights. For the record, I’m totally apolitical in this and I feel that both sides have insights to add.

What I haven’t been able to do is to put the whole thing into some sort of “competing dogmas” framework that makes sense to me. I am lacking a general idea of how the two competing camps are coming at this problem.

I imagine that two dogmatic extremes might be: 1) a purely probabilistic and statistical analysis of known real dives and their medical outcomes vs. 2) A quasi-physical mathematical model that is adjusted ad-hoc so that one can minimalize hits recorded from real dives over the greatest range.

I’ll try to clarify my impression of the two approaches:

Pretend that we have a database containing every dive that was ever done, exact gas mix, depth, time, conditions, and an unbiased accurate assessment of the diver’s medical condition at the end of the dive. With that in hand, I would think that a really accurate algorithm/dive computer could be developed for diving at any average assigned risk (within the limits of the dataset) using nothing but probability and statistics. No compartments, no models, nothing. But still "perfect" for the AVERAGE diver.

With a less complete dataset, one is tempted to construct a compartment model and test it against what data you have. If it holds up then it is also tempting to extrapolate it to predict dives that are outside the dataset.

In a perfect world, it seems to me that we would want to start with a perfectly complete dataset and then attempt to construct an accurate model that fits the data and now allows for extrapolation. I suspect that both camps see the solution as a mix of both approaches.

Can someone tell me where the two opposing camps would be placed on a line between these two extremes? Or is it that I just don’t get it at all and there is a different fundamental explanation as to the heart of this dispute?

 

[ajduplessis]

I'd like to take a brief step back, please.

I’ve been trying very hard to follow this thread for insights. For the record, I’m totally apolitical in this and I feel that both sides have insights to add.

What I haven’t been able to do is to put the whole thing into some sort of “competing dogmas” framework that makes sense to me. I am lacking a general idea of how the two competing camps are coming at this problem.

Thats how the parties involved here roll. They nitpick at others work by keeping the discussion fragmented, answer questions with questions, sling insults, and focus on where their perceived strengths lie with some "smoke and mirrors". One is making money with a large database/outcome, the other wanting to contribute to the industry, but unfortunately as professionals also rolled in the "mud".

The NEDU study and related topics presented here/elsewhere is debated to death with the promoters very bullish with a following.

So what now????

 

[Dr Simon Mitchell]

What did I miss?

With all due respect, quite a bit.

If you skim a thread like that you will form the impression of roughly equal numbers of posts for and against, and find ample superficial justification for composing the nice egalitarian story you have presented.

If you really want to understand the deep stop issue PLEASE read and carefully appraise the information presented by both sides. Start at the start, and end at the end. Same for the helium kinetics issue in the CCRX thread. You will find that far from "swimming in an ocean of subjectivity" many of the arguments are grounded in compelling logic, often with supporting data, and much of it published. Of course it has been contested; that's why it became a debate. But you need to dig deeper than that simple observation before concluding that the debate was not resolved with a reasonable degree of certainty.

If you are not of a mind to do that because of the magnitude of the task, then you will also understand why I have no desire to repeat the exercise here.

Simon M

 

[Storker]

Pretend that we have a database containing every dive that was ever done, exact gas mix, depth, time, conditions, and an unbiased accurate assessment of the diver’s medical condition at the end of the dive. With that in hand, I would think that a really accurate algorithm/dive computer could be developed for diving at any average assigned risk (within the limits of the dataset) using nothing but probability and statistics. No compartments, no models, nothing. But still "perfect" for the AVERAGE diver.

That would be a purely empirical model, nothing more than fitting a random equation to your datapoints. You would really be out of your depth (pun not intended) if you used a model like that for extrapolation. It's a flawed model no matter how many datapoints you fit it to.

Both supersaturation models and bubble models have known physical/chemical processes as their fundament, so by nature they should be more robust as you venture outside the data points that were used for fitting the model.

If I understand correctly, we don't really know if it is the supersaturation or the microbubble stability that limits bubble growth. And, of course, that using any model - whether it's purely empirical or semi-fundamental - for extrapolation model in a range where it isn't supported by data is inherently risky. That is what the experiments we've discussed have tried to amend, by collecting at least some data in a range where we've only had extrapolations and assumptions.


--
Sent from my Android phone
Typos are a feature, not a bug