We should not be wary of loud truths either. Unless we push the envelope we cannot trace back to the essence. Like, as an outsider in this discussion, I am now inclined not only to assess the finding, but also the definition of an organ. The title did good work. Its refutation too. Fencing every statement in unquestionable truth does not make the pursuit of truth stronger.

But there’s rarely one correct answer. Conveying the same approximate message to multiple types of readers is extremely difficult in a handful of words. People with subject matter knowledge are annoyed by headlines written for laypeople that don’t know the jargon or have a mental model of the important concepts that jargon represents. They always seem to miss important distinctions, make poor analogies, or the like. Headline-length blurbs that successfully convey those things to the knowledgeable are often utterly opaque to laypeople causing them to ignore articles about topics they’d find interesting. The only option is trying to estimate the subject matter expertise of the audience, and despite many people in the tech world assuming their software knowledge gives them technical expertise in many non-software fields, the layperson headline is absolutely the best approach here.

This is not true. There is a complex relationship between truth and good communication. The relationship is non linear and convoluted and it depends often on the subject of what is being communicated.

There are many many cases where lies are easier to communicate than the truth.

Take tobacco for instance. For the longest time we did not know it was bad for you. It took decades of epidemiological research and statistical techniques to ascertain that it was bad.

The truth was buried in numbers and human knowledge about applied mathematics and statistics. Without centuries of human endeavor to create the concept of science unionized with the rigor of statistics the truth would have remained forever invisible.

The sheer complexity of the truth in this case makes is extremely hard to communicate. We mostly trust that smoking is bad based off of the reputation of the researchers who established that fact. But the truth was never really directly communicated to most of us as it involves diving deep into the data.

“…but LLMs can’t” is the unspoken sticking point.