BlaBlaMeter detects how much bullshit is in your text

I tried some Hegel, who scored 0.18. Clearly broken.

Schopenhauer got it right, where BlaBlaMeter gets it wrong:

"If I were to say that the so-called philosophy of this fellow Hegel is a colossal piece of mystification which will yet provide posterity with an inexhaustible theme for laughter at our times, that it is a pseudo-philosophy paralyzing all mental powers, stifling all real thinking, and, by the most outrageous misuse of language, putting in its place the hollowest, most senseless, thoughtless, and, as is confirmed by its success, most stupefying verbiage, I should be quite right. Further, if I were to say that this summus philosophus ... scribbled nonsense quite unlike any mortal before him, so that whoever could read his most eulogized work, the so-called Phenomenology of the Mind, without feeling as if he were in a madhouse, would qualify as an inmate for Bedlam, I should be no less right."

Excerpt is cited as fair use, namely the educational value in evaluating a tool with three sentences of buzzword heavy english from a real site. Specific citation info omitted only so they won't be embarrassed, but may be found with Google.

Or just esthetic rules + word dictionary.

The text was:

"Möchten Sie SAP-Software vor Ort installieren oder über die Cloud darauf zugreifen? Wir bieten in jedem Fall umfassende Services, zugeschnitten auf Ihre individuellen Anforderungen. Wir verfügen über eines der größten Expertenteams weltweit. Unsere qualifizierten Mitarbeiter beraten Sie gern bei der Konzeptionierung, Implementierung und Optimierung Ihrer Systemlandschaft. Profitieren Sie innerhalb kürzester Zeit von Ihrer SAP-Lösung."

It my old pet project; like BlaBlaMeter, but for licenses.

I still hope to experiment with the idea of improving licenses in future.

It appears our internal bullshit meters work just fine as well :)

- It uses a unigram language model. You can take the same text, randomly permute the words, and you get the same score. This means it also can't be using things like POS tagging, phrases, etc.

- It normalizes words by making all letters lowercase. The exact same text in all upper case has the same score.

- The score is eventually normalized by the length of the text. The same text copied multiple times gets the same score.

- It does not form a valid probability distribution, as someone's managed to get some 1.16's. This makes me believe it's not a Naive Bayes classifier giving you the P(Bullshit|Text). Though this is what I originally thought it would be.

For example, if you take the score from the Oracle Pricing blurb posted by BitMistro, and change 'strategies' to 'goals', you drop down to 0.8 or so. If you add an extra random 'strategy' somewhere, it bumps up to 1.4 or something.

I actually suspect a bug on their pair for strategies... probably a decimal error when building to BS level tables.

But similar things happen with other 'bullshity' words, just to a lesser degree.

"Politics are great, come buy our new, brand spanking awesome banana phone, apple, steve jobs, cripplingly epic banana phone. Just great phones, with bananas, no apples to be found here. Samsung can suck on our banana phone. Android is better than iOS."

"Your text: 251 characters, 43 words Bullshit Index :0.03 Your text shows no or marginal indications of 'bullshit'-English."

Edit: oh its happening today - I thought it was old news and no-one had told me. Apologies for the exclamation marks now gone.

---

Your text: 129 characters, 29 words Bullshit Index :0 Your text shows no or marginal indications of 'bullshit'-English.

BlaBlaMeter prefers an extra word over a comma. Makes sense given the goal here.

High-quality journalistic texts are usually in the range of 0.1 and 0.3

* Are there any texts with an indexof 0.0?

This is rare, but occurs. Too low of an index is rather suspicious though, and might also indicate stilistical deficits.

* What is the highest index value?

Thile the BlaBla Meter was designed for index values between 0 and 1, but in rare cases the scale can be exceeded. The highest measured values so far have been over 2.0 even.

* Does Google use a similar algorithm?

We don't know, but if we were Google, we'd probably try! Our samples for highly competitive search terms show that highly ranked sites often show good index values.

* How does BlaBla Meter work?

BlaBla meter checks texts for various linguistic traits, e.g. it checks for exceeding use of Nominalstil[1]. In addition, the text is checked for various phrases [buzzwords] with a certain weighting. We don't want to reveal the secrets though ;)

[1] Heavy use of "nominal style" is a German particularity. You can replace most verbs by nouns, adding suffixes like '-ion' or '-age', and transform the sentence into passive. This is often found in legal texts, where the author doesn't want to appear as a suject.

* Why does my scientific text have such a high index?

Nominalstil has crept into scientific language at large. This often serves to 'beat around the bush' - in this regard, there are analogies to typical PR tongue.

* My mindless text gets a good score, why?

BlaBla Meter can't really understand the topic, it looks after linguistic traits. While rainbow press can be attributed all kinds of substantial weaknesses, their language is usually short and concise.

* My text is wrongly devalued!

As every computer algorithm, BlaBla Meter might be at fault - in doubt, human beings must decide of course. Generally the hit rate is quite large, so you should usually tweak the wording upon receiving high indices.

* But humans read texts completely differently ...

When BlaBla Meter crys havoc, one can usually assume that a human reader is alerted too. People are trained every day to tell authentic messages from artifical advertising messages - it'd be naive to belive they couldn't!

* What happens with entered texts?

Texts are private matter of course, and are neither saved nor processed in any way or shape.

I put in a page or so from a scientific paper I recently wrote and got 0.19.