Not with certainty, of course not. But that is not the point, and it was never claimed. And the termometer analogy is irrelevant, and you know it. :-) You need to distinguish skills from "good at guess-work", or whatever. I am talking about a researcher - or rather a team of researchers - who collectively know a great deal about this non-deterministic phenomenon known as "global warming". They do not know exactly what will happen, no. And they do not claim that they do, either. But that does not make them un-skilled. They certainly know a lot about the mechanisms behind the phenomena that we observe. And they make claims in the form of likelihoods, and ranges of effects with uncertainty bounds. I call them skilled as heck.

Well, to begin with one has to assume that whoever digs her teeth into research has a strong educational background. Hence, she is skilled in the basic theory and practice of the field of study at hand. (Otherwise, you would not call her a researcher, but an amateur, right?) Second, you are well aware that a lot of research is NOT about searching the unknown. In fact, I bet that 3/4 or more of the research that we read is of the incremental type - small additions are made to already existing theories, practices, and methodologies. I do not refer to that most common research as "working in an unpredictable unknown" area, would you? :-)

Easy: researchers know methods for discovering knowledge: observation, measurement, reasoning (induction, deduction, etc.). They, as Michael says, you build on previously discovered knowledge and exploring the cause of contradictions. Research is about discovering what was not known before but it is always in the context of what we already know.

That might depend on how you interpret "predictions", and how accurate you think that they are - or need to be in order to for the scientists to be called "skilled enough". In climate science, for example, there is STRONG evidence of a global warming trend; those predictions on the extent of global warning are very well supported by science, such as statistical analysis. That does not mean that we KNOW exactly what will happen, but you must remember that those scientists do not describe the trend with deterministic certainty. So their predictions - I would say - are definitely supported - by historical measurements, physics and meteorology, and by mathematical statistics. So there you go. Your turn. :-)

Dear Michael,

Do you think that a researcher is skilled when he/she accurately describes a pattern (e.g. a change in climate at a given scale) without knowing the exact underlying mechanisms and therefore without being able to make a reliable prediction about what will happen in the future?

For instance, any person with a thermometer at home may track changes in temperature over years. The person will be able to describe the past, but will the same person be able to predict the future based on what happened in the past?

Not with certainty, of course not. But that is not the point, and it was never claimed. And the termometer analogy is irrelevant, and you know it. :-) You need to distinguish skills from "good at guess-work", or whatever. I am talking about a researcher - or rather a team of researchers - who collectively know a great deal about this non-deterministic phenomenon known as "global warming". They do not know exactly what will happen, no. And they do not claim that they do, either. But that does not make them un-skilled. They certainly know a lot about the mechanisms behind the phenomena that we observe. And they make claims in the form of likelihoods, and ranges of effects with uncertainty bounds. I call them skilled as heck.

I do not think you need prediction at all times--a lot is discovered by just trying stuff--then when you find something you can make predictions--modern chemistry was pre dated by centuries of alchemy--they were mostly wrong but they discovered enough by trial and error to lay some ground work--now as to global warming: not so fast--this field is about 80% political--every single climate model has failed repeatedly-- there has been no average warming in 17 years--I am not saying what will happen one way or the other but the political aspect of this is no secret--if you question the status quo your career can be ruined--there is data hiding, threats to those who oppose the majority, fudging of analyses, lies--even threats from politicians--I have been around long enough to be skeptical when politics tries to trump science--

Hello again,

You are focusing the discussion on global warming. Perhaps have a look at the attached file.

Hypothesis: Clouds principally formed from water vapor creates a visible and dynamic 'green house effect' because in winter or in the desert nights are much colder without than with clouds. What is the role of carbon in the creation of greenhouse effects accepting that C represents a tiny fraction of the chemical in the atmosphere and the fact that there is a steep increase in CO2 that is disconnected from the change in average Earth temperature during the last decades?

In my opinion, they can be skilled when the person has the most relevant prediction or answer. I don't know to much about this topic, but you got me on this question, because every thing is a prediction or theory, that an intelligent question.

That's the question!

A pertinent question and also my consent to your response. In fact what you say leaves us perpetually questions - answer because no one knows the truth because they exist in a number and it is almost impossible for a researcher to possession. We can only surfer to follow these truths.

Comparative analysis presupposes unpredictability (for ex. in researching cross-cultural differences).

Answers or questions:

All is not rational or may be expressed rationally?

In the emotion of scientific inquiry can express the sterility of the researcher?

There are events that human intelligence cannot explain?

Well, research is the process  or skills of revisiting the known results to investigate the unsolved, unknown or the imagined and produce results.  In order a person to do that he/she has to have paramount knowledge and skills of different things and scientific principles with a developed power of imagination. The overall broad based  intellectual capacity of a person made him/her skilled researcher and be able to conjecture or confront the unknown to make it known or create a path to that end. 

Your skill is measured by your ability to "predict the unknown".

How often predictions are based on anecdotal observations to result in a hypothesis subsequently scientifically tested? Is this a case of 'the ability to predict the unknown'? 

A researcher is a 'skilled' one as: --- has the ability to dig out 'unpredictable unknown'!

That's why new inventions and discoveries!

The skill of a researcher lies in transforming the unknown into known knowledge

When we talk about discoveries I always have the impression that results were a great surprise (cf. discovery of penicillin by A. Fleming)? Is the skill in this case reflected in predicting the unknown?

Doing the same thing over and over again and

expecting different results is insanity

- Albert Einstein

--------------

I feel that Research requires adept skills to identify and implement rigorous approaches to destination Unknown, Unpredictable and Uncertain

Dear @Marcel

What is your definition of a skill or skilled?

In my opinion a researcher searching for something unknown or known (wants to know more about it) possesses skills of doing research. Can (s)he proceed without possessing any skill?

Skills are reflected in mastering existing knowledge to ask to best/most pertinent research direction among the many possible research directions available? 

By their previous success.

Skill areas (Generic)

• Meticulous documentation of activities
• Owns Joint responsibility in lab activities
• Follows quality in research projects
• Careful attention to ethical principles

@Krishnan: What is the importance of short-term (e.g. maximizing quantity?) versus long-term (maximizing quality independent from quantity?) success in the definition of those skill areas?

Comparing process-based climate models to empirically based clutch models isn't a good comparison of skill.  To quote myself on an article about the equally enigmatic subject of extreme flooding when applied outside of their inference space--

"There is a basic assumption that once a model that has been fit to match the observed hydrograph that all other inputs and output is therefore correct (Hooper, 2001; Singh andWoolhiser, 2002). This assumption is fundamentally opposed to the scientific method because there is no means for rejecting a model’s structure once it has been selected. Model performance is highly sensitive to structure, and alternatives may just as easily explain an observed phenomenon and structural errors are almost always present (Jakeman and Hornberger, 1993; Young et al., 1996; Cloke et al., 2003; Butts et al., 2004)."

In brief-- models of this kind aren't good for hypothesis testing because you can get the right answer for the wrong reasons (http://www.tylervigen.com/).

Thus, process models like climate models are always open to criticism.  They are easy targets as evidenced by climate debate.  There is a tendency to oversell and undersell what they tell us.  And skilled scientists are still amateur marketers (http://en.wikipedia.org/wiki/Merchants_of_Doubt)

Butts, M.B., Payne, J.T., Kristensen, M., Madsen, H., 2004. An evaluation of the

impact of model structure on hydrological modelling uncertainty for

streamflow simulation. J. Hydrol. 298, 242–266.

Cloke, H.L., Renaud, J.-P., Claxton, A.J., McDonnell, J.J., Anderson, M.G.,

Blake, J.R., Bates, P.D., 2003. The effect of model configuration on

modelled hillsope-riparian interactions. J. Hydrol. 279, 167–181.

Hooper, R.P., 2001. Applying the scientific method to small catchment studies:

a review of the Panola Mountain experience. Hydrol. Process. 15, 2039–

2050.

Jakeman, A.J., Hornberger, G.M., 1993. How much complexity is warranted in

a rainfall-runoff model? Water Resour. Res. 29, 2637–2649.

Singh, V.P.,Woolhiser, M., 2002. Mathematical modeling of watershed hydrology.

J. Hydrol. Eng. 7, 270–292.

Young, P.C., Parkinson, S., Lees, M.J., 1996. Simplicity out of complexity

in environmental modeling: Occams’s razor revisited. J. Appl. Stat. 23,

165–210.

@Mark: Comparing process-based climate models to empirically based clutch models isn't a good comparison of skill. Why? Both phenomena (dynamics in climate and clutch size expression) rely on processes that are more or less known allowing skilled people to make more or less reliable predictions?

A process model is made up of assumptions based on known phenomenon.  They are not parsimonious and subject to overspecification.  A lot of work has been done to control uncertainty, but uncertainty isn't something that modellers typically spend much time talking about.

Empirical models are based classically designed experiments that address a hypothesis and generate real data that can reject a hypothesis.  There is no way to "reject" a process model because the structure is pre-selected.  Empirical models can still be misused, but they at least have the real data to back them up and diagnostics to judge how good of a job they're doing.

Thus, modeling based on unverified assumptions is based on 'trial and error' (e.g. crossing fingers that it will work this time)?

Example

Do predictions related to Climate change have the same level of reliability as the  predictions related to Economic change? Both are large-scale multi-factorial influenced which implies that not also influences are truly mastered  I empirically observed that the predictions of Economic growth in a country have been adjusted multiple times during the last three years despite the fact that the advisory teams have been considered as skilled tops specialists in their field?

The key issue is inference space, how far out you are pushing the bounds of the prediction, what you contend, and the stakes of your conclusion.  I'm sure your clutch model isn't used to predict clutch size for input parameters that don't make sense... like a bird that weighs double the maximum bird size in previous studies... would you necessarily trust the model to answer that question?  You trust it for giving you a mean response... and less so for an extreme case.

"People" like models because they're deterministic; you plug some things in, they give you a number, easy-peasy.  Few spend enough time considering the uncertainty of inputs or the uncertainty of predictions; and unlike means, uncertainty generally only gets larger the more components the model has. And if you are trying to predict extremes your model could be absolute crap -- there are models designed for extreme low flows in rivers, normal conditions, and extreme high flows.  They are not interchangeable because of inference space.  A model designed for current climate or current economy has an irreconcilable weakness for predicting outside of their bounds, as does any model.

That being said, don't interpret my comments to mean I'm a climate denier or that I think the work is bunk.  I am not a climate expert.  I think these models are accurate about trend, and perhaps reliable on magnitude.  Modeling techniques and computing power have certainly grown since I wrote my lowly hydrology review paper.  There is a lot more being done about uncertainty.  But there is still a fundamental difference between classic, hypothesis driven designed experiments, and large-scale process models based on a preconceived notion about how a system functions.

The best situation is when the two techniques are used in conjunction.  The process model predicts this... and we design an experiment to confirm the claim.

It would be easier if climate change and economy were not so entangled; economy speaks of growth and profit, the climate change predicts a cost; models can be manipulated. The presentation of the results is also a factor.

Economy growth is not so unpredictable, everyone knows where the system is going. Why is a negative growth so hard to accept. Less instead of more.

Marcel,

In reality, all research in all areas is unpredictable. If it were, there would be no need for the research. The probably with education and any research involving thinking organisms is that there are many variables that can not be controlled. And, if you did control these variables how generalizable would the findings be?

In terms of a skillful researcher, there are well defined conventions in both quantitative and qualitative research paradigms that can be applied appropriately or inappropriately. More importantly, how well has the researcher collected data that can be used to answer the questions asked. In the end, the researcher needs to convince the reader that the conclusions reached are reasonable, given the collected data.

Norm

Researchers are skilled when they know the field of research. The skill is induction from observation and knowledge of the mechanisms involved in the prediction. A skilled research knows exactly what to expect. The scientific laws predict the outcome from the observations. Proper induction is in concert with the known laws. A skilled researcher should not be surprised by the results, because the results were predicted from known science and observation. The observation is the basis for further investigation of the predicted result. Deviation may be a surprise. The prepared observer recognizes the deviation and the surprise. Skill and knowledge allows both recognition of the surprise and the interpretation of the import of the deviation. Skill is required to know that the surprise is truly a surprise and not result of poor skill.

There is a substantial difference between 'knowing exactly what to expect' and 'the future result'!

Researchers/scientists have the capacity/competence/skill to set the just impossible objective, for lay man---not for scientists, and achieve it.

The establishment of a quality relationship is important in research situations. The interpersonal interaction is the core of research practice in making the relationship a fundamental element - human / health / psychological / sociological related researches.

Marcel

Correct. There is a substantial difference between 'knowing exactly what to expect' and 'the future result'! A result different from expected requires and explanation; equipment failure, poor timing, incorrect observation, ... , or Eureka. Nevertheless, a scientist does not stumble about twisting knobs, mixing solutions, or peeking under rocks to find something to shout, "Eureka." There is a direction and an expectation.

I once heard a lecture by a former classmate of Fleming. He claimed that Fleming was the class dummy. Everybody knew about penicillin and how it was a problem for bacterial cultures. Everyone who was any good was looking for ways to keep it out of bacterial cultures. I am sure it was a joke, but nobody laughed.

Research is a process of discovery whether you have an hypothesis or not--what is critical is to have questions and then use proper methods (e.g., look at facts, take measurements, look for causal relationships, look for contradictions, etc.

The competence is reflected in the method, not in the (future) result. Therefore, negative results also merit publication!?

It pains me that good experiments asking reasonable questions but yield non results are more difficult to publish as they still inform science and may still inspire other work; or at least keep people from pushing false leads. .

In the normal position, we are held a study to answer specific questions dealing with a problem. Sometimes you need to explore a new topic such as adapting new varieties for a certain environment. You can treat results by repeat the experiment several times which give more accurate conclusions and recommendations.

Researchers become skilled when they learn something they did not know ewfore

Marcel, hi. I wonder what will hapen if you orient the question to the "propability and stat" section. I predict some unpredictable answers.

This not just happens with science but also in business. While one can abstract all that one wants in the end its a very pragmatic issue. One establishes a baseline of being skilled by how well one knows the previous research in ones field. In any proposal one has to establish this by doing the review of literature. Using this as baseline your peers will determine whether you are just saying nonsense or repeating what others have done. This along with your "track record" allows for a "guesstimate"of your skill level. What will most likely be a good gamble is staying close to the previous work and doing incremental  research or bringing expertise from other areas.

Leeway from these rules will be based on whether your track records merit sthe gamble or not, unless your work is total genius and will depend on the "eye of the beholder" (takes one to know one).

Science is like any new venture that is filed with unknowns, and as in business, the risk is minimized by having a good team(or individual).

Just an opinion

According to me researchers can be termed as skilled because they have enough skills to work on unknown thing and discover it. I think researchers skills are absolute one which can have ability to make human lives very comfortable.