Thread by @alexandrosM

Media should be free to title their articles whatever they want without constant second-guessing

Readers should be free not to have their limbic system hijacked with linguistic patterns that smuggle in unstated premises

The solution to the paradox? Headline Neutralizer (TM) 🧵

How does it work?

Headline Neutralizer (TM) is a non-existent service that applies preset linguistic transformation rules designed to retain explicit meaning (which media is liable for) while silencing implicit meaning (which media is not liable for)

Let's try an example:

The Independent recently published this headline:

"The Late Show viewers call out Jon Stewart for peddling ‘harmful’ lab leak coronavirus theory"

Let's apply some transformation rules and see what happens

independent.co.uk/arts-entertain…

Rule 1: No Unquantified Sets

When referring to "viewers", readers may assume it refers to a majority of viewers. This isn't actually claimed. As such, we can clarify to:

"Some viewers of The Late Show call out Jon Stewart for peddling ‘harmful’ lab leak coronavirus theory"

Rule 2: Neutralize Emotive Conjugation

The choice of words like "peddling" and "call out" can smuggle in hidden assumptions of guilt. We can clarify to:

"Some viewers of The Late Show claim Jon Stewart is promoting ‘harmful’ lab leak coronavirus theory" en.wikipedia.org/wiki/Emotive_c…

Rule 3: Scientific Correctness

The word "theory" in popular use conflicts with scientific usage in which "theory" carries the highest evidence standard. We can clarify to:

"Some viewers of The Late Show claim Jon Stewart is promoting ‘harmful’ lab leak coronavirus hypothesis"

Rule 4: Clear quote attribution

Single-word quotes can save space, but can imply endorsement of the claim when quickly read. We can clarify to:

"Some viewers of The Late Show claim Jon Stewart is promoting the lab leak coronavirus hypothesis which one says is 'harmful'".

Rule 5: Narrow down quantifiers

If we are allowed to dip into the article text, we see that the number of viewers quoted is four. We can clarify to:

"Four viewers of The Late Show claim Jon Stewart is promoting the lab leak coronavirus hypothesis which one says is 'harmful'".

Rule 6: Specify Harm

The word "harmful" can be broad and subjective, so the type and target of harm should be made explicit

"Four viewers of The Late Show claim Jon Stewart is promoting the lab leak coronavirus hypothesis which one says is 'harmful' to an unspecified group".

In short, we went from:

"The Late Show viewers call out Jon Stewart for peddling ‘harmful’ lab leak coronavirus theory"

to:

"Four viewers of The Late Show claim Jon Stewart is promoting the lab leak coronavirus hypothesis which one says is 'harmful' to an unspecified group".

While character count grew, the second headline, at a glance, implies far less than the first. Also, once the hidden layers are peeled off, the title sounds hollow. Nobody doubts that Jon Stewart promoted the lab leak hypothesis, so the largest part is not even newsworthy.

It's only the "harmful" part that is interesting, and there doesn't seem to be any detail offered as to who was harmed and in what way.

Headline Neutralizer (TM) should not significantly affect neutral titles. For instance "Jon Rahm wins US Open" would be untouched.

To be clear, Headline Neutralizer (TM) does not exist yet, but something like it is not technically infeasible, either using traditional natural language processing techniques or more modern Deep Learning methods.

I increasingly see that the situation as it stands isn't sustainable. Our emotional centers respond to headlines as if they came from a local source with skin in the game (the only kind, historically), when they actually come mostly from remote sources optimizing for attention.

That source is only minimally responsible for the quality of the information shared, and even less so for premises implied but not stated. We need a defense layer between us and the internet that corrects for our evolutionary firmware's outdated assumptions.

Meat vs algorithm is not a fair fight. This is an arms race but one side is not yet arming at all. We must fight algorithms with algorithms.

Is anyone aware of any existing products that fill this niche of a defense layer between the human and the big bad internet?

I suppose the various adblock / tracker remover plug-ins, as well as browser privacy modes are one way in which we defend ourselves from various classes of threats, including attention highjacking, so it's worth mentioning on this list.