Get rid of irrelevant attributions in ‘word salad’ statements

Leonard Mlodinow notes in The Drunkard’s Walk, that two facts can be correct at the same time where one is relevant and the other is irrelevant [1]. In this sense, there can be errors of relevance where an argument using the wrong base rate is not incorrect but has little meaning in a chosen context. Daniel Kahneman refers to this as ‘denominator neglect’[2].

Another instance of irrelevance occurs where there is no base rate and where denominator neglect is not strictly present.  For example, something that is attributable to the whole is irrelevant because it has no meaning, not because it is strictly breaks a base rate rule.

Common examples use the thought experiment of ‘everything being green’.

If there was only one colour and that colour was ‘green’ then something having the attribute of being green would have no meaning as everything in the world would be green. ‘Greenness’ would be undefinable as a ‘colour’ as colour itself is necessarily dependent on the existence of other colours to have meaning.

It would not be ‘wrong’ to say something is green; it would simply have no meaning and no possible relevance to the colour of anything[3]

Mlodinow points to the ‘prosecutor’s fallacy’ when evidence was allowed in the O.J. Simpson trial that attested correctly to the fact that 90%+ of men who beat their wives do not kill them. This evidence of course in this case is irrelevant since the proper base rate is not battered women – it’s dead women –  and it turns out that 90%+ of women found dead following a violent attack  in their homes were killed by a male partner. 

Irrelevant expressions in word salad statements

One of the common attributes of ‘word salad’ statements is this same error with its intention to steer the reader or listener to irrelevancies that are somehow meant to answer an objection.

They have various shadings that I refer to as the four ‘d’s: diminishment, definitional, defaults and distractions. Let’s look at each.


“We take all complaints very seriously” is meant simultaneously to diminish the complainant and the complaint by the attribution that all complaints are ‘serious’. Your complaint is placed in the same pool as misinformed complaints and complaints that have no factual basis. 


“Our potato chips contain no cholesterol” or that “our potato chips make a great vegan snack”commit attribution errors differently. Since potato chips (or at least most of them) contain no animal matter, they must be’cholesterol free’. The fact that they do not contain cholesterol is irrelevant since they cannot contain cholesterol. 


The Vegan claim – default error – is slightly more interesting. The same attribution error is present as well as the definitional irrelevancy. But someone might be able to make the further claim that to be vegan is something more than being a snack with no animal matter.  


The fourth is the distraction attribution error. A good example is found with no-medical life insurance and reverse mortgage products. 

The no medical life insurance places the purchaser in the same risk pool as some very sick people putting up the price of the insurance but saving the buyer from providing an ’embarrassing’ medical report. 

The distraction here is that all medical reports are characterized by the insurer to be ‘private’, ’embarrassing’ or somehow ‘disqualifying’ when it comes to being insured. The error relates to purposefully misleading information – an agnotological error. The fact that a ‘good medical’ is an easy way to obtain cheaper life insurance is not part of the equation presented even though well-known among insurers. 

Similarly, reverse mortgage providers claim that the homeowner does not have to pay ‘until you choose to move or sell’. The distraction is that most reverse mortgages are aimed at older homeowners most of whom only sell their properties when forced to move due to infirmity. Most people would not see such a move as a ‘choice’ in the usual sense. Accordingly, they are distracted by the imposition of a scenario that will likely not apply to them but which could, in other circumstances, apply to them or others. 

Regardless, ‘word salad’ statements almost always make more than one attribution mistake. In fact they are almost always characterized by four attribution errors as described.

It is not difficult find a word salad statement that contains instances of errors of relevance. I am looking for one that contains all four.

Js – July 22, 2021



“Picture two urns stood on a table in front of you.

You’re given the opportunity to pick a marble from one of them, and drawing a red marble wins a prize.

The first urn has 10 marbles in it, 1 of which is red.

The second urn has 100 marbles in it, 8 of which are red.

Which urn would you choose? It doesn’t seem a tricky decision: your chances of drawing a red marble out of the first urn are greater (10%) than your chances of drawing a red marble out of the second urn (8%).”