Week 1 Reflective Journal

This paper is a reflective journal on super forecasting. I expected to find ways of
developing connections between activities and expected that machines are the best tools for
forecasting. The key takeaways from the readings have been that connections are not so easy to
make because seemingly unrelated events could have a direct cause-effect relationship with each
other. I also learned that machines and humans should operate in synergy and dichotomy to
produce the best forecasting results. Finally, I learned that a good forecaster should treat
everything with an element of doubt.
One of the issues with predictability is that forecasting is a skill that many people have,
but it isn’t easy to continuously make accurate forecasts. This could be due to the nature of
seemingly unrelated events having implications on the future outlook. Tetlock (2015) gives an
example of the events that led to a revolution that led to the Tunisian President fleeing for
asylum in Libya. The revolution begins when a street vendor gets harassed by police to a point
where he douches and sets himself on firesetting in motion a chain of events that no one could
have predicted.
Becoming a super forecaster means predicting that things will happen; it also means
predicting the implication of occurrences; this is what sets the difference between forecasting
and super forecasting. The essence of this is that knowing that something will happen is abstract.
Tetlock (2015) gives an example of insurance firms’ and actuaries’ work. Insurance firms, for
example, can predict that a person at some point will fall ill or that they will retire or die. No one
knows when they will die or fall ill, but when they do, there will be implications for their
families and their finances; basically, insurance firms are super forecasters because they give

3
policyholders control, albeit limited, over events that are destined to happen at a date when no
one can predict.
Machines can perform forecasting algorithms better and faster than humans. Machines
have higher capabilities to gather and store big information, perform large format calculations,
and give conclusions faster than humans. It could be argued that machines are the ultimate super
forecasters. This could be true if the goal is to look at the conclusion without giving it meaning.
What machines give a conclusion is just an abstract that requires a human judgment to give it
meaning. Super forecasting means allowing the synergy and dichotomy of human and machine
effort (Hamid et al., 2017).
The art of becoming super forecasters should also involve changing our minds. Most of
the time, human beings make decisions based on the information at hand but never stop
questioning whether such information could be wrong. Tetlock (2015) gives another example of
a woman diagnosed with a carcinoma in her armpit, and after surgery, the doctor who discovered
it tells her that she has not got much time to live. She starts planning for her soon-to-be death,
which never does, at least not within the doctors’ timeframe; she goes on to live a longer life.
This is another indication of human bias and limitation.
Another limitation could be that human beings often use an outcome as a basis for
accepting or rejecting the practicability of concepts. Tetlock (2015) gives an example of a
medical case. If a person undergoes a treatment process and they recover, it could be seen as
evidence that the treatment procedure works. However, there could be other reasons that the
patient recovered. They probably had a robust immune system, or the disease was not so
progressed, and the patient was able to overcome it. Another patient who dies while undergoing
the same treatment regimen could argue that the treatment does not work, and they would be

4
justified; only if they looked at the outcome. Becoming a super forecaster means being able to
consider different aspects that could affect an outcome.
In the same manner, human beings err by trying to look for evidence to support a
plausible hypothesis. The challenge is that we do not look at facts and information objectively.
We may say, for example, a certain construct is true, and I will prove it. From then on, any
evidence we are looking for is meant to prove that the outcome we have developed is true.
However, there could be several explanations for an outcome, and with good research, it is
possible to find evidence that what we know to be true could be false.
In some cases, we make a judgment based on intuition, and intuitive judgment could be
accurate. Intuition is not bad per se, but it is only good if it is based on learning from experience,
opportunity, and evidence. Tetlock (2015) describes intuition as a lucky accident or magic that
does not come from learning. Therefore, a good forecaster should always treat everything with
some element of doubt because forecasting that involves little experimenting is bad forecasting,
and bad forecasting leads to bad decisions.

5

References

Hamid, O. H., Smith, N. L., & Barzanji, A. (2017). Automation, per se, is not job elimination:
How artificial intelligence forwards cooperative human-machine coexistence. 2017 IEEE
15th International Conference on Industrial Informatics (INDIN).
https://doi.org/10.1109/indin.2017.8104891
Tetlock, P.E & Gardner, D. (2015). Superforecasting: The art and science of prediction.
Retrieved from https://www.redshelf.com