A Bad Ad

—as far as the writing goes, anyway. It’s ambiguous. I ran into this:

Stop Wasting Money With PayPal’s New Money-Saving Tool

(Find it yourself. It’s around.)

What’s the relationship between the tool and money?

• You waste money if you use the tool? This is how I took it when I first read the line—It was an ad for a competing product.
• You waste money if you don’t use the tool?

The problem is what “with” means. Less ambiguous would be to say, “Stop wasting money. Use PayPal’s new money-saving tool.”

You can probably come up with a few other ways to say it. Feel free to improve the line with your own line in the comments.

Subscribe to this blog's RSS feed

Not much of a lesson today, but if you’re into typography it’ll be funny. I suppose I should mention that font is a single style of letters (roman, for example) and a typeface includes variants of that font (roman, italic, bold and so on).

It’s funny because the business card uses the Papyrus typeface, roundly disliked by many professionals.

One of my rules is to choose the simpler of two choices when you write. Hence, don’t say “upon” when “on” will do. Same for “in” and “within.”

And “based off of” is definitely wrong! Aak!

He came up with the correct solution: get rid of all the unnecessary words.

Harrumpf.

I thought this was interesting. Sorry it’s so academic.

MIT researchers have built a system that fools natural-language processing (NLP) systems by swapping words with synonyms:

The software, developed by a team at MIT, looks for the words in a sentence that are most important to an NLP classifier and replaces them with a synonym that a human would find natural.
For example, changing the sentence “The characters, cast in impossibly contrived situations, are totally estranged from reality” to “The characters, cast in impossibly engineered circumstances, are fully estranged from reality” makes no real difference to how we read it. But the tweaks made an AI interpret the sentences completely differently.
The results of this adversarial machine learning attack are impressive:
For example, Google’s powerful BERT neural net was worse by a factor of five to seven at identifying whether reviews on Yelp were positive or negative.

Abstract: Machine learning algorithms are often vulnerable to adversarial examples that have imperceptible alterations from the original counterparts but can fool the state-of-the-art models. It is helpful to evaluate or even improve the robustness of these models by exposing the maliciously crafted adversarial examples. In this paper, we present TextFooler, a simple but strong baseline to generate natural adversarial text. By applying it to two fundamental natural language tasks, text classification and textual entailment, we successfully attacked three target models, including the powerful pre-trained BERT, and the widely used convolutional and recurrent neural networks. We demonstrate the advantages of this framework in three ways: (1) effective — it outperforms state-of-the-art attacks in terms of success rate and perturbation rate, (2) utility-preserving — it preserves semantic content and grammaticality, and remains correctly classified by humans, and (3) efficient — it generates adversarial text with computational complexity linear to the text length.

From the May 15, 2020 Cryptogram

Today’s post is a vocabulary lesson. Metaphor is a generic term used for about any figure of speech (technically, a metaphor is when you say that something is something else). A mixed metaphor is when you combine two figures of speech, often with humorous results. The last two panels have two mixed metaphors:

Don’t do this in real life if you can help it.