On Preferring A to B, while also preferring B to A(marginalrevolution.com)
marginalrevolution.com
On Preferring A to B, while also preferring B to A
https://marginalrevolution.com/marginalrevolution/2018/07/preferring-b-also-preferring-b.html
8 comments
Would giving things "star" ratings count as evaluating them separately? I've always been bothered by star ratings. If I give ItemA 5 out of 5 stars, and then later find ItemB which is even better, how do I indicate that?
I made a simple proof of concept ranker app that shows you two power metal songs and asks "Which is better?" Then it uses your answers as the comparator to binary search each items' positions in the ranking list.
http://rcfox.ca/Power-Metal-Ranker/
I made a simple proof of concept ranker app that shows you two power metal songs and asks "Which is better?" Then it uses your answers as the comparator to binary search each items' positions in the ranking list.
http://rcfox.ca/Power-Metal-Ranker/
Here is the citation for the actual paper, including a link for its download without fee or login:
Sunstein, Cass R., On Preferring A to B, While Also Preferring B to A (March 21, 2018). Forthcoming, Rationality and Society; Harvard Public Law Working Paper No. 18-13. Available at SSRN: https://ssrn.com/abstract=3132428
Sunstein, Cass R., On Preferring A to B, While Also Preferring B to A (March 21, 2018). Forthcoming, Rationality and Society; Harvard Public Law Working Paper No. 18-13. Available at SSRN: https://ssrn.com/abstract=3132428
I’ve become increasingly skeptical of appeals to “rationality”... the idea works when you’re approaching reasoning as a formal system with clear inputs that everyone agrees on, but that doesn’t apply to most of the decisions we make on a daily basis.
What’s “best” is going to be determined by the specific nature of your values. Rationality, in my opinion, is value-relative, and therefore it makes little sense to speak of rationality as an objective property of a decision. For toy examples where all of the variables are agreed upon, sure, but not much else.
What’s “best” is going to be determined by the specific nature of your values. Rationality, in my opinion, is value-relative, and therefore it makes little sense to speak of rationality as an objective property of a decision. For toy examples where all of the variables are agreed upon, sure, but not much else.
I think that at the core of this problem is:
1.) Separate evaluation is done in absolute terms.
2.) Join evaluation is done in relative terms in a specific model. Which specific model? The model that only takes into account the union of features of both items. So this is similar to a relative comparison with conditional probability of occurrence.
The relative evaluation is done thinking about: In which situations would item B be more valuable than item A. All those situations are weighted according to the avaiability heuristic: easy to recall situations seem more frequent or important. To give a simple example: Perhaps a dictionary with 20000 words is enough for you, and you don't need a greater dictionary. But you can think than may words are in the 40000 words dictionary but not in the 20000 word dictionary, so the difference seems to be more important that what really is for you.
The relative evaluation is done thinking about: In which situations would item B be more valuable than item A. All those situations are weighted according to the avaiability heuristic: easy to recall situations seem more frequent or important. To give a simple example: Perhaps a dictionary with 20000 words is enough for you, and you don't need a greater dictionary. But you can think than may words are in the 40000 words dictionary but not in the 20000 word dictionary, so the difference seems to be more important that what really is for you.
This is interesting to me because it gives one a way to evaluate if they are valuing something fairly (usually in comparison shopping) like mentioned with the monitor example. Keeping this in mind reminds you to determine the value of a factor independent of other products, thus assigning a significance to you on its own and at different levels. It's also an indicator of how knowledgeable we are about the metric.
For example: One can visualize 128 colors as being not that many. Once we get into the thousands this gets more and more abstract & less important.
Another: A has a 5ms response time, B has a 10ms response time. A seems better, but on its own each doesn't seem to make much sense - lets go learn more about what this metric means.
Another : A is 5kg and B is 10kg. A might seem better but when we look at each independently we look at why weight matters. We realize both are OK - a monitor isn't going to be moving around much.
For example: One can visualize 128 colors as being not that many. Once we get into the thousands this gets more and more abstract & less important.
Another: A has a 5ms response time, B has a 10ms response time. A seems better, but on its own each doesn't seem to make much sense - lets go learn more about what this metric means.
Another : A is 5kg and B is 10kg. A might seem better but when we look at each independently we look at why weight matters. We realize both are OK - a monitor isn't going to be moving around much.
A 10kg monitor is probably better because it’s gonna feel less flimsy and wobbly any time you kick the table. Depending on how the weight reduction was achieved.
Any monitor that doesn't bend your desk is light enough. Even ones that do bend your desk are OK if they're a Sony Trinitron. :P
Unless your goal is to mount it on a wall. Better/worse are really not meaningful by themselves w/o context.
>Sellers could take advantage of the bias of joint evaluation by emphasizing information that consumers might think is important but actually isn’t–our computer screen has 1.073 billion color combinations while our competitors only has 16.7 million–while making less salient 6 hours of battery life versus 8 which may in practice be more important.
Is there a specific computer manufacturer that this is taking a shot at? Apple maybe?
Is there a specific computer manufacturer that this is taking a shot at? Apple maybe?
Top two things that come to mind are camera megapixels and 4k screens on phones. Apple always used a smaller number of megapixels but ultimately a better sensor, while other manufacturers touted the raw count. Similarly there were some 4k phones with an unnecessary dpi. I think Samsung had one and although the screen was 4k the OS resolution was defaulted lower to save battery life.
> Congressional Candidate A: Would create 5000 jobs; has been convicted of a misdemeanor
> Congressional Candidate B: Would create 1000 jobs; has no criminal convictions
>
> In each case B tends to have a higher value when evaluated separately but A tends to evaluate higher with joint evaluation.
Wait, what? The other examples make sense, but this one doesn't. Even with the joint evaluation, the choice is still highly debatable depending on the actual crime as well as the jobs. Creating jobs doesn't automatically make up for assault, which is considered a misdemeanor in some places.
Wait, what? The other examples make sense, but this one doesn't. Even with the joint evaluation, the choice is still highly debatable depending on the actual crime as well as the jobs. Creating jobs doesn't automatically make up for assault, which is considered a misdemeanor in some places.
One study even quantified it[1]. For example, they found that people perceived the value of an iPod at $108, while an iPod with plus a free music download was valued at $86. Side-by-side, the choice is obvious, but individually humans tend to evaluate bundles based on some average rather than a sum.
This has immediate, practical applications for us in everyday life:
* If you're going to buy someone an expensive gift, don't add any extras or bonus accessories.
* If you're sending out resumes, mention your most relevant experience without padding it out with less relevant ones.
* If you're up for a review, "I fixed serious issue X" will make you seem more valuable than "I fixed serious issue X, and this unaligned icon on the web site"
[1] http://bear.warrington.ufl.edu/williams/Mar_3503/MAR_3503_Ho...