Difference between revisions of "Open science and society"

From Testiwiki
Jump to: navigation, search
(first draft in English)
Line 3: Line 3:
 
:''Avoin yhteiskunta
 
:''Avoin yhteiskunta
  
Researchers are rapidly learning to use big data, and online tools enable collaboration in self-organised groups. It is time to revisit how the scientific method about openness, criticism, and falsification could be used in a better way.
+
Researchers are rapidly learning to use big data, and online tools enable collaboration in self-organised groups. It is time to revisit how the scientific method could be used in a better way. Ideally, science is open collaboration among benevolent people who share all data they have,
 +
<ref name="alldata">Publishing all data is more effective than prepublishing peer review.</ref>
 +
develop research questions and formulate hypotheses as answers. Then they falsify hypotheses using all available data. But only data and not e.g. prestige of speakers. Whatever holds against criticism is treated as the current best description about the reality.
  
Good hypotheses hold against all data, not just a single study. So, all data should be published.
+
It seems that we already can make the ideal come true.  
<ref name="alldata">Publishing all data is more effective than prepublishing peer review.</ref>
 
Open data creates much more opportunities to falsify hypotheses, so their creation should be promoted. Could there be an open website for anyone to specify research questions and to develop hypotheses as answers to them? We have tested it in small scale and it seems to work<ref name="opasnet">http://en.opasnet.org</ref>
 
  
Assume a website that obeys the rules of scientific method. All contributions are welcome, but relevance to a research question is always required. All arguments are weak except when backed up with data. Hypotheses invalidated by data are kept visible to avoid repetition of previous mistakes. Posterior probabilities are used to quantitatively describe knowledge given all data. A key challenge is to organise qualitative, verbal discussions into structured, more quantitative questions and answers, but experiences so far are positive.
+
Assume a website where scientific rules are strictly applied. All contributions are welcome, but relevance to a research question is always required and work aims at answering these questions. Falsified hypotheses are kept visible to avoid repetition of previous mistakes. Posterior probabilities are used to quantitatively describe knowledge given all data.
 +
We have successfully tested this in small scale.<ref name="opasnet">http://en.opasnet.org</ref>
 +
A key challenge is to organise qualitative, verbal discussions into structured, more quantitative questions and answers, but experiences so far are positive.
 
<ref name="lelieveldcomment">Comment paper on Lelieveld et al criticism.</ref>
 
<ref name="lelieveldcomment">Comment paper on Lelieveld et al criticism.</ref>
  
Wikipedia has shown that large knowledge bases produced by self-organised groups using specific rules are feasible. Bellingcat society has shown that using scientific approach, it is possible to rule out all but one hypothesis about the shooter of the Malesian airplane in Ukraine in 2014.
+
Wikipedia has shown that large knowledge bases produced by self-organised groups using specific rules are feasible. Bellingcat society has shown that even heated policy issues can be resolved in this way, such as the shooter of the Malesian airplane in Ukraine in 2014.
 
<ref name="bellingcat">Bellingcat</ref>  
 
<ref name="bellingcat">Bellingcat</ref>  
 
Similar approach could be used to a large number of politically and scientifically important questions.  
 
Similar approach could be used to a large number of politically and scientifically important questions.  
Line 19: Line 21:
  
 
The current systems to give merit to researchers would give zero value to such work. But the critical question is this: are there enough researchers who would contribute anyway, because of its impact rather than personal merit?
 
The current systems to give merit to researchers would give zero value to such work. But the critical question is this: are there enough researchers who would contribute anyway, because of its impact rather than personal merit?
 +
  
 
'''References
 
'''References

Revision as of 11:58, 15 March 2016

Possible article formats in Nature: Correspondence (max 300 words), possibly extended to Comment or even Hypothesis. [1]
Avoin yhteiskunta

Researchers are rapidly learning to use big data, and online tools enable collaboration in self-organised groups. It is time to revisit how the scientific method could be used in a better way. Ideally, science is open collaboration among benevolent people who share all data they have, [1] develop research questions and formulate hypotheses as answers. Then they falsify hypotheses using all available data. But only data and not e.g. prestige of speakers. Whatever holds against criticism is treated as the current best description about the reality.

It seems that we already can make the ideal come true.

Assume a website where scientific rules are strictly applied. All contributions are welcome, but relevance to a research question is always required and work aims at answering these questions. Falsified hypotheses are kept visible to avoid repetition of previous mistakes. Posterior probabilities are used to quantitatively describe knowledge given all data. We have successfully tested this in small scale.[2] A key challenge is to organise qualitative, verbal discussions into structured, more quantitative questions and answers, but experiences so far are positive. [3]

Wikipedia has shown that large knowledge bases produced by self-organised groups using specific rules are feasible. Bellingcat society has shown that even heated policy issues can be resolved in this way, such as the shooter of the Malesian airplane in Ukraine in 2014. [4] Similar approach could be used to a large number of politically and scientifically important questions.

The worst decisions are made based on flawed understanding or ignorance of scientific knowledge and/or people's opinions. Researchers could show it beforehand using scientific method on such a website. The credibility of such a website, if successful, would increase in time as its performance could be measured and evaluated. The noise in social media would not affect its conclusions unless there is a scientific signal.

The current systems to give merit to researchers would give zero value to such work. But the critical question is this: are there enough researchers who would contribute anyway, because of its impact rather than personal merit?


References

  1. Publishing all data is more effective than prepublishing peer review.
  2. http://en.opasnet.org
  3. Comment paper on Lelieveld et al criticism.
  4. Bellingcat

384 words

Nettityökalut mahdollistavat suoran demokratian.

Tieteessä on mahdollista keskittyä tieteen menetelmän perusteisiin ja hylätä toissijaisia käytäntöjä.

Kaikki data kannattaa julkaista.

Bayes-lähestymistapa on mahdollista toteuttaa laajassa mittakaavassa, jolloin alkaa kehittyä kaikenkattava malli.

Jokaisen kelvollisen hypoteesin pitää olla yhteensopiva kaiken havaintoaineiston kanssa.

Hypoteesien avaaminen kaikkien ehdotettavaksi tehostaa tutkimusta.

Data kyllä karsii mahdottomuudet. -> ei tarvita tutkijan auktoriteettia pönkittämään tutkimustietoa.

On mahdollista rakentaa nettisivu jossa noudatetaan tieteen sääntöjä. Tutkijan ei tarvitse mennä somen ehdoilla.

Se voi käsitellä myös yhteiskunnallisia asioita kuten tulevien päätösten arvioituja vaikutuksia. Vaikka tiede ei pysty kertomaan mitä tapahtuu, se kuitenkin pystyy kertomaan mitä päätösten avulla ei voi saada aikaan.

Koska suurimmat vahingot yhteiskunnalle syntyvät nimeomaan typerimmistä päätöksistä, niitä pystytään näin tehokkaasti välttämään.

Bellingcat onnistui selvittämään avoimen datan perusteella miten ammuttiin malesialaiskone Ukrainassa. Vaikka ei saatu tietää ihmisiä, selvitettiin kuitenkin osapuoli. Se on olennaista, koska samalla iso osa julkista keskustelua voitiin osoittaa virheelliseksi ellei peräti valheeksi.

Miten maailmasta tulisi parempi paikka jos tieteen menetelmän avulla ammuttaisiin alas politiikan väärät tiedot?

Tätä jo tehdään truthometer etc mutta systemaattisessa mielessä ei olla kovin pitkällä.

Somessa meluaminen on tehotonta vaikka olisi oikeassakin.

Mutta wikipedia osoittaa että voi olla tasan yksi paikka johon tieto kerätään tietyllä menetelmällä kuten tieteen periaatteita noudattaen.

Kaikki myös tietävät mistä se löytyy.

Hyökkäys toisen dataa käyttäviä haaskalintuja vastaan. Mikä lehti?

Politiikkoja voi keksiä kuka tahansa kuten hypoteesejakin. Pitää murskata harha että se olisi vain poliitikon tai tutkijan oikeus.

Ensin he eivät piittaa sinusta. Sitten he nauravat sinulle sitten he hyökkäävät sinua vastaan. Silloin oletkin jo voittanut. Gandhi.

Nimenomaan hyökkääminen on keino vahvistaa oikeaan osunutta hypoteesia. Sitä pitäisi siis käyttää enemmän.

Metavitsi: tämän käsikirjoituksen hyväksyminen kertoo tieteellisen julkaisemisen korjaamisen mahdollisuudesta, koska väitteet ovat rohkeita. Tieteen menetelmähän sanoo, että tällaisia hypoteeseja tulisi julkaista ja niitä vastaan tulisi aktivisesti hyökätä, jotta niiden kestävyys selviäisi.

Voisi viitata Lelieveldin ja Evansin artikkeliin, koska sen kritiikissä käytettiin jäsennettyä keskustelua jota tämäkin artikkeli ehdottaaa. Viite myös Eemeren ja Grootendorst.

Julkean väitteen perusteluna voidaan sanoa, ettei se liene mahdoton koska jokaisen osatekijän on erikseen jo osoitettu toimivan. Kyse on siis vain siitä, että ne pistetään tekemään yhteisttötä.