Jessica Green: Are we filtering the wrong microbes?

[rodswebdesign]

This is a great video

[SuperDogbrown]

love the work here

[sprattysy]

interesting video and very informative

[MrJonkelp]

love the video really good

[Sic7777]

Houses made of yoghurt. Just what I always needed.

[aarongrooves]

I felt like she was just getting started, then it ended. I really wanted to hear more. How disappointing.

[Eucalypticus]

discussion has gone a long way otherwise too. You seem an interesting and likeable person, so this was not distasteful. Take care. 

[egokick]

>13 data points

>attempt to draw meaningful conclusion

lol.

[twistedbass15]

Genius



[thegardenerspolemic]

It is possible that the undesirable microbes are being controlled by the rest of the aerobic microbial community. We're discovering the same is true in our soils, and that salvation might lie not in disinfecting, rather in infecting wisely.

[TheFounderUtopia]

I get noticeably sick once, maybe twice a year. Usually when the weather changes drastically enough to confuse my body. You know what would make me get sick more often? If I spent my life as a paranoid germaphobe hiding from every microbe - then my immune system would not get the opportunities it needs to keep generating immunities and I would be susceptible to every virus no matter how weak.

Hospitals are one thing, they SHOULD be as sterile as possible. Otherwise, relax. You'll live longer.

[jafro8]

huemens

[beatboxpeej]

solution...camping

[Vortex42]

@Digitized Self - If you can't understand that graph, I can't help you. It's a small sample, but it leads to a very clear and obvious conclusion. Larger data set? Yes - no problem - worth doing... But even this small test showed a working system. There were enough samples done to see a clear stratification on the axes, dependent on the altered factor in each sample.

[DigitizedSelf]

@Vortex42 In what way is the data sample statistically significant? Sure, the data is indicative of some sort of grouping of the data point but how sure can you be that this hasn't just happened by chance if you don't have a larger data set? Is a linear model really the right hypothesis? What would the correct slope be? What's meant by "outside" and is it really surprising that indoor hospital air isn't that healthy? No doubt there's a "very clear and obvious conclusion" - but how correct is it?

[Vortex42]

@DigitizedSelf As I mentioned in my first response, I agree that a larger data set would definitely be an advantage. However, even the small random data set these initial findings were based on point in a clear direction... Many questions you ask, I dare say, would be clearly defined in the writing of the researchers (ie : what is defined by 'outside'). Also, it doesn't matter if something seems like it wouldn't be surprising... You still do the tests.

Anyway - more tests? - don't see why not.

[DigitizedSelf]

@Vortex42 I agree that you still have to test it - simply thinking you have the correct answer leads nowhere. That being said I see the linear fit to be the crudest model possible to put forth and it's agreement with the data in this case is rather limited (unless the data points have rather large error bars) - and finally, I don't see the data as being suitable for TED. It's not inspiring or surprising and it's not even backed up with significant statistics :-/.

[srgwarcock]

@DigitizedSelf oh i bet you feel real fucking smart dont you, how about you do something useful with your time rather than dogging on other people research and hard work faggot!

[DigitizedSelf]

@srgwarcock Wow, someone seems to have a problem :-/. No use in name calling - it's only degrading to yourself and, since you chose 'faggot' to be your insult of choice, shows you to be rather ignorant.

[srgwarcock]

@DigitizedSelf Well obviously you need to show all the research you but into your work in a full report that will be seen by actual Scientists, but for a presentation for the public needs to be simplified to get the point across. Anyways i hope you enjoyed arguing with someone who couldn't even abbreviate Sergeant right... so Fuck you and thank you kindly.

[DigitizedSelf]

@srgwarcock I fail to see how showing more data points would confuse ppl - if anything it should only make the trend clearer. In any case; nice discussing with you despite your slightly abrasive words (oh, and didn't realize srg was an abbreviation of sergeant or in any case how to abbreviate it correctly - English isn't my first language so had to look it up)

[srgwarcock]

@DigitizedSelf really, i thought i suits you actually. Your talking like you know everything, theres a reason shes up there talking and not you and thats because she is a perfessional in her field, and im sure she has more than enough data to prove what she is saying.

[DigitizedSelf]

@srgwarcock Must be nice to live in such a simple world! ...jokes aside, I too assume she's competent and as such probably doing some nice research - problem is she's doing a pretty poor job in presenting it. Science isn't built on the conceived social standing of those presenting it; it's based on the actual data and should be convincing because of the information put forth. So you can be 'sure' and all but that doesn't really tilt the scales the slightest.

[nomusician]

After a supposed successful operation (to treat CML), my father caught a lung infection, while still hospitalised, leading to his death shortly afterwards.

[W0ngster]

@nomusician thats sad. sorry to hear that.

[shrek4ever2003]

I felt more of the research money was put into the animations for the slide show than in making a larger study

[IvanMarinIvkovich]

The more microbes, the better. Strenghtens my immune system. :)

[dookiecheez]

Seems like an unchecked portion of the reason why hospitals have such high infection rates.

[dylanlawless1]

Microbiology rules. This talk didn't.

data points.. (in a sarcastic voice, with a look of disgust)

[deadswans2000]

several years too soon, for both topic and speaker

[rx7777777er]

Is the y axis proportion? learn to put units on your graph!!! you are a professional scientist...

[rakoon3434]

@rx7777777er yeah, the y-axis was proportion... which is a dimensionless quantity and so it doesn't have or need any units. However, the x-axis just said "diversity" and I'm not sure what units that was measured in, although I would guess the number of species.

[rx7777777er]

@rakoon3434 Thanks

[Waltham1892]

hey, at least she isn't going on, and on, and on, about her vigina...

[kiddhitta]

i couldn't even watch that video because of the way she carries out the last word of her sentence. she sounds like a family guy character.

[2528997]

Germaphobe much?

[xjustamem0ryx]

@2528997

would be hilarious to see an OCD germaphobe try to cope with microscopic vision >:)

[goldfishlaser]

liked for "architectural yogurt". that sounds like a FUN science fair project!

[brahman8908]

Yeah, I love the continual look back of her ass. No other TED talks has had as many ass shots as this lady...

[pragette1]

My mum already told me you are not getting sick in winter by the cold outside, but by staying inside all the time without letting new air come in.

[klimaxg]

Hospitals are palaces of disease. It seems reasonable that there would be a lot of pathogens about, many more than outdoors.

I'm sure there are lots more compelling reasons to open the windows and let the fresh air in.

[xjustamem0ryx]

lol we're suffocating ourselves?

[alphakristjan]

How dare you accuse me of filtering the wrong microbes...

[dzntms144]

If I used her first "graph" in a talk in the same way she did, my professor would brake my neck...

Wait a second, she thinks probiotic youghurt works?

Well, there still is a long way to show significant effects...

[Sylocat]

Another problem is that overuse of antibiotics is breeding superbugs, thanks to that pesky evolution thing (drug-resistant pathogens are popping up since we use too many drugs against them so only the drug-resistant ones survive).

That, plus this... I hope her ideas get implemented.

[csrtitus]

I have been living in "un-healthy" buildings for my whole life.....I am still alive and am in good health. A "healthy" building is just a cheap gimmic.

[NjC121]

@Formosus111: The point of the talk is that hospitals are fostering the pathogens, despite how obvious it's become that being within has become worse than being without. She didn't pose a serious solution, which irked me when I watched this, but that's because we dont really have one available yet. Will planting a few plants remove the germs? No, that doesn't help cycle out the air with the cleaner air outdoors. It doesn't change that all that air is regurgitated through the same system over an

[randomizer1666]

Probiotics have never been proven to do anything at all. There's no evidence that suggests you can change the microflora in your body by introducing new ones, generally the new microbes introduced just get destroyed by the resident microflora.

[themeloninc]

clean air and... yogurt??  love it!

[fibrousfeces]

She's dressed like an extra from Grease.

[Formosus111]

Wait. Were the two indoor conditions obtained from hospitals? So, without allowing outside microbes in, microbes found in hospitals are the most densely pathogenic? STOP THE PRESSES, new headline! Diseases are found in centers of disease!

[Primalxbeast]

I don't think the point of keeping the windows closed is to filter microbes. It's usually too hot or too cold to have the windows open, and the outdoor microbes may be more likely to be harmless, but where is her proof that they are helpful?

[apocaRUFF]

Kinda sounded like she was reading out of a book and that it was her first time reading about it.

[sklunt]

GO OUTDOOR! NOW!! 

[ranasingh]

We need plants everywhere in buildings as we depend on them aswell as they do on us. Let's grow plants everywhere in buildings.

[YouHolli]

This is the worst TED talk I ever saw.

[CerainLoLs]

higher chance of hitting a pathogen... but a lower chance of hitting a microbe. Does not mean more pathogens. Her analysis of the data seems a bit lacking.

[anticliche321]

ocd

[PR0H0LDEM]

wow, that was short and I didn't get information about open window environment

[akazombie]

That's not what I'm getting from that graph at all... :/ She needs more data :P

[ExtreamThrills]

the reason they keep the windows from opening is so that way the old folks grandchildren don't fall out, least thats what i heard. other than that it was a great speech. i don't believe in most of it but i agree with the part about her saying stuff about microbes and stuff.

[bweazel]

We have immune systems. Next!

[Alkoholwioslaidziwki]

IMO it's the issue of incapable ventilation systems and overcrowded buildings. I can't really see how adding more bacteria can help lowering the rate of infections by airborn pathogens.

[drche420]

Did anyone else think that graph needed at least a dozen more data points per ventillation method in order to draw a "best fit" line?

Least impressive TED talk so far.

[xjustamem0ryx]

@drche420

seriously.

underwhelming samples with overwhelming generalizations. there should be stricter laws regarding drawing inference/correlation/causatio­n from statistics.

[Darkaero199]

We have the longest average lifespan in all of human history, I don't think being indoors is hurting us too much.

[maggru91]

Herpa derp, I hear Anthrax is an amazing air refresher, derp derp.

[tculig]

So she is saying that we should have more plant germs inside...how does that help in decreasing the number of human germs??? There is no solution here other than to kick everyone out of the building!

[kevintype]

@tculig - The vast majority of microbes are not germs, because they do not cause diseases. However, when we reduce the diversity of microbes in our environment, leaving behind only the ones that thrive in and around humans, this increases the percentage of disease-causing germs among all the microbes we encounter. Increasing the diversity - by letting in more microbes from outdoors - forces the disease organisms to face greater competition, which may reduce our risk of illness.

[tculig]

@kevintype Exactly, percentage decreases but the absolute number does not.

[tculig]

@kevintype Nice argument...I'm sure you were the head of the debate club.

[tculig]

@kevintype Touche sir, touche.

[xjustamem0ryx]

MAD MICROBE DISEASE!

[vazeh]

TURN DOWN YOUR INTROS!

[SMFApples]

@vazeh Turn down your volume...

[carefulcarpenter]

Reminds me about pasteurization, and the standards set by otherwise intelligent people. Are we sterilizing our human experience to the point of killing our healthy diverse culture?

[xjustamem0ryx]

@carefulcarpenter

"Are we sterilizing our human experience to the point of killing our healthy diverse culture?"

yes. and one need only look at western aesthetics/architecture to see just how deeply sterilization has penetrated.

much of what makes something aesthetically interesting is non-optimal. but the industrial optimization complex paired with the path of least resistance eventually bound toward uninteresting degrees of conformity.

[carefulcarpenter]

@xjustamem0ryx Well put~! Had I not worked for the aesthetically-challenged highly-credentialed power elite, I might not have awakened in time to understand that office buildings can be dangerous to one's healthy cognitive development and species diversity. ;)

Give them 3 choices and they can put a plan together; give them free reign to use their imagination and you discover a void.

[DigitizedSelf]

Okay, so outdoor air contains microbes from outdoors, indoor air from humans - no d'uh :S

Oh, and WTF was that straight line @2:11? Yeah, I get the point that it claims that higher diversity means lower risk of pathogen but the fit isn't really that convincing... especially without errorbars.... without knowing much about it I'd assume it gets a terrible chi-squared value :S

Moar data, broader study - nao! (additional datapoints, effects of indoor plants, alternative ventilation etc)

[DigitizedSelf]

@DigitizedSelf Okay - actually I DO have a question; the claim, according to the description, is that the ventilation is "getting rid" of microbes but she doesn't seem to actually touch upon this subject... To me it seems much more likely that the outdoor microbes are just kept out and indoors in. I mean, it's fairly simple to test; you just check the time-dependency of microbe populations indoors. Furthermore there's a HUGE difference in what is meant by outdoors; right outside, a forest, what?

[gaiagale]

@DigitizedSelf did you mean 'more data' and 'broader study - now! ?

[DigitizedSelf]

@gaiagale Yeah, that's what I wrote :P. IMO YouTube (and teh intarweb in general) isn't really suited to be serious all the time; you know, someone gotta bring the lulz once in a while ^.^ (it was either that or crying after that video)

In general I'd appreciate the corrections to my grammar etc. but when it's an obvious, deliberate display of incorrect spelling corrections become rather redundant - anyways, glad you managed to figure it out ;-) (and, if done in good will, thx 4 teh thought)

[gaiagale]

@DigitizedSelf yes my intention was understanding ...I'm feeling a bit naive now though ...oh well ;-])

live and learn

[VivekVish]

@DigitizedSelf Chi-squared or t-stat (for the coefficient) or F-stat (for the R-squared)? What will the Chi-squared value measure in a linear regression? (I'm not being facetious. I'm just wondering if I'm missing something.)

Also, just eye balling that graph, it is almost certainly significant on a 0.05 level. Not sure about stricter significance. There are many places where TEDTalks could include p-values, but if the paper has been published on a claim, it is surely significant.

[2plus2make4]

@VivekVish It is probably not a linear model.

[DigitizedSelf]

@VivekVish Not sure I get the question? Are you implying that Chi-squared measures something different for a linear regression or just asking what it measures? Assuming the latter: Generally, if the uncertainties in the data points are Gaussian distributed then Chi-square minimization would lead to the most probable values of the coefficients. The chi-square value can be mapped (in a degree of freedom dependent way) to a probability that can tell you if your function is correct.

[DigitizedSelf]

@VivekVish TBH I don't know the other tests you mentioned (other than if t-stat is the same as Student's t-test) but Chi-squared is a pretty general (and popular) approach - even when the uncertainties are not Gaussian it gives a fairly robust (although not necessarily the most probable) fit result and intuitive measure of the total fit quality.

...I'm curious though - how can you say anything about significance without errorbars?

[kikosemmek]

@DigitizedSelf agreed. However I typically give TED the benefit of the doubt when it comes to me wanting more data and more clarification, because the prestige involved in just getting to be in that conference tells you that the research must have been great to start with. It's maybe that she simply did not have the time allocated to her for a more full presentation or w/e. It's always good to look up these people's names regardless :)

[srgwarcock]

@DigitizedSelf Hey how about you use proper english and real words. Besides i think she was trying to make a straight-to-the-point presentation.

[DigitizedSelf]

@srgwarcock Hey, how 'bout you spell "I" with a capital letter? :P Srsly though, the problem lies in the cases where ppl don't KNOW how to spell words properly and not when incorrect spelling is deliberately used to achieve a specific effect.

That being said language isn't static so how do you define "proper" and "real"?

Don't get stuck on formalities - it serves little to no purpose and often just provides a hindrance for a discussion with substance ;-)

[Eucalypticus]

@DigitizedSelf This is not a 'no d'uh' situation. Without data it would not be immediately clear that indoor environments can harbor pathogenic microbes that are distinct from outdoor environments. This is not a mere curiosity since these bacteria are found more pathogenic.

The straight line was just a visual aid. There is no purpose to attempt to mathematically fit such a small data set; no underlying pattern is sought and no interpolation is done. The line demonstrates a very important...

[Eucalypticus]

@Eucalypticus ...correlation. Claiming the data is flawed due to lack of error bars or lack of samples is fairly narrow-sighted. An important and non-trivial result has been demonstrated in this instance. If there were 100 people in a party, and one drank the punch and died instantly, I would reason that something is up--not request more data.

[Eucalypticus]

@Eucalypticus I would add: further data would be useful. But given that these experiments seem costly, likely time intensive, and that the results are intuitive upon analysis, I think it worthwhile that these results are presented as they are, in a ~5 min. format, so that their content is spread sooner than later.

[DigitizedSelf]

@Eucalypticus 1) You mean to seriously tell me you didn't know that indoor air and outdoor air was different? ...or even less healthy? 2) What is the difference between a fit and a "line that demonstrates a correlation"? Oh, right - you just kinda 'feel' one of them. 3) What does data from a measurement even mean without uncertainties? It can be whatever! ...and how can you claim the non-fit, correlation thing (whatever that is) to demonstrate anything when you can't even trust the measurements?

[Eucalypticus]

@DigitizedSelf 1) Knowing a difference exists does not imply knowing how the difference manifests. Also: if I were to claim o. air were different than i. air because it seemed reasonable, wouldn't I be jumping to a conclusion while lacking data, which is exactly what you're arguing against? 2) A correlation indicates a relationship; indicates dependency. A fit minimizes some error and provides a functional relationship that is often a direct numerical correspondence between both variables. ..

[Eucalypticus]

@Eucalypticus Since no error was minimized requesting error bars is meaningless. 3) You are focusing too heavily on what was meant to be a simple visual aid. It was just meant to highlight the pattern. I am not saying uncertainties would not have made her work more convincing, but consider: a) perhaps she has them but opted not to use them; b) using them would obscure the main point and might not be a good 'fit' for the audience. Even if measurements were taken without uncertainty, which I doubt

[Eucalypticus]

@Eucalypticus it is highly likely that she's onto something. What is the other explanation? The bacteria organized itself this way in a manner of pure chance? I don't think I need error bars or a confidence interval to convince that that is *highly* unlikely.

[DigitizedSelf]

@Eucalypticus 1) Sure, but I'm really questioning whether this 'discovery' is significant enough to be a TED talk. 2) Glad you agree that a fit doesn't make sense, but I don't find 'visual aids' to be better (I find them to be mostly deceiving). Correlation measures a linear dependency; the most probable of which would be found through a fit. The errors I'm requesting is on the data points, not the potential fit. 3) You DO in fact need the error bars to say that it's highly unlikely :-/.

[Eucalypticus]

@DigitizedSelf I appreciate: a) your attitude, b) your desire to be mathematically rigorous, but the latter is too extreme here.

1) A non-obvious result relating simple actions and pathogenic microbes was shown. The result contradicts conventional practice, it dispels an illusion by showing the opposite. Most importantly, it demonstrates this in a hospital, where one least expects it..

[Eucalypticus]

@Eucalypticus ..and where one least needs it. This is very significant. TED organizers felt similarly. 2) My agreement that a fit would be meaningless is in disagreement with your claim that it's a bad fit. If you now agree with this then I suppose you abandoned your initial claim. Deceiving visual aids? With due respect, you seem to be reiterating bullet-points found at the back of some statistics textbook. It was simply a straight line used to quickly emphasize a relationship in a..

[Eucalypticus]

@Eucalypticus ..time-bound presentation. My senses were not deceived by some woefully complex device. It was just a line. Correlation needn't measure linearity. I could be more precise about that, but not here. The crux is that dependency was shown. Not opening windows can make you sick, period. The lack of error bars on the data points have already been addressed. To reiterate: their lack does not strongly alter her conclusion (related to 3) ); their presentation would've obscured her point..

[DigitizedSelf]

@Eucalypticus 1) I'd expect it quite a lot in a hospital actually, but let's let that rest :-/. 2) A terrible fit or no fit at all - this seems more like semantics than actual meaning to me, but yes, I was working from the premise that it was a fit... if it's not a fit, it's not a fit (this is not a matter of opinion). I'm quite intrigued by your claim that correlation isn't only measuring linear dependency - there must be some def I'm unaware of (?).

[DigitizedSelf]

@DigitizedSelf 3) Your claim that the lack of error bars doesn't alter her conclusion is based on the assumption that the bars are small, which we don't know (although it doesn't seem unlikely).

Meh, in any case, nice chatting with you - doesn't seem like we're moving further presently due to simple difference of opinions. Always nice to discuss stuff with someone who offers an opinion and is willing to argue for it. As stated I'd be most interested in the correlation thing! ;-)

[DigitizedSelf]

@DigitizedSelf Ah, missed the last reply since it was marked as spam (for some odd reason :S ...I 'unspammed' it but dunno if that makes a difference). Regarding error bars it seems like we're in quite the disagreement; I would claim that you'd ALWAYS need error bars, especially when dealing with measurements you have no idea how is performed. How do you get thermodynamics into the picture? Simple probability would tell you if it's likely or unlikely, but only when the errors are known :-/.

[Eucalypticus]

1) Alright. 2) In general it would just be dependency, i.e., altering one variable affects the other. Two things may be correlated but their relationship might not be linear. It is just a matter of defining our terms properly. if we carefully laid out definitions I'm sure we'd both agree. 3) See below.

It was refreshing and surprising to have a cordial e-discussion. The crux of my arguments and actions were to defend research that I thought useful. You, or someone, removed the..

[Eucalypticus]

..main offending response (with 56 thumbs up), so there's not much else left. If you remove it that is noteworthy.

The final point: I concede error bars would add validity. My argument was that I see no valid reason to call the entire operation into doubt. This is not an approval of the lack of error-bars, but rather a disapproval of the claim that everything is meaningless without them. It seems clear something is amiss. I believe you're requesting presentation of measurement uncertainties

[Eucalypticus]

..and/or an analysis of statistical significance. While useful, they would not make or break the greater purpose of this presentation. The only purpose of introducing TD is to support the idea that these bacteria would not spontaneously organize themselves, that other causality exists, and that her conclusion is not at all unreasonable. The work is likely published. It is reasonable to assume that work of this caliber did not use methods that would introduce gross uncertainties in measurement.

[Eucalypticus]

The reason I began commenting is because I felt that a derisive comment, whose effect was amplified by others, unjustly called her work into question. Worse yet, those who were not familiar with mathematical analyses might not question this in the slightest and take for granted that her work was flawed. I think a greater than 10% dislike count at this point would be 'correlated' to this.

These were my motives. The comment is gone. This..

[greob]

Good talk! The CGI animations were nicely done too.

[adolfog316]

wat the hell is this new reaction button, there should be an epic option

[rlclaveau]

Great TED talk... and yes i would tap that

[quaxk]

@rlclaveau she's not playing that team dude, get your gaydar fixed

[ereignisfelder]

@rlclaveau legs worth spreading XD

[TheHickstead]

The intro is extremely loud!! nice animation!

[VideoNewZ9]

no i was just sick