I like the Hemingway app, but it is very much designed to make you write like Hemingway (ie short, clear sentences _only_). Obviously most great writing is not in this style.

The Hemingway app is good for work emails and marketing copy, or if you happen to like that style. It is hardly the universal arbiter of "poor or confusing writing".

Oh wow it's almost completely okay with Hemingway. I pasted some of his work in and it had almost no complaint at all. Good point.

I was coming to say the same, and the descriptors being removed actually just points mote to its usefulness for writing 'in a Hemingway style'.

However, having said that I don't think the author of this article used it well, as for many reasons most any journalistic article is going to be very descriptive and use things like puns to get/maintain a reader's attention (something I personally usually hate, but not always).

I think it would be fun to create a version of this which pushes writing towards other recognizable styles like Raymond Chandler or David Foster Wallace. Maybe you pick a style from a dropdown and the analysis changes. You could possibly take it even further and automate some transformations of the input to stylize a piece of writing.

To satisfy a DFW style it'd need to understand 2 page long sentences with multiple footnotes per sentence. Haha. It would be a fun one to do though! (I say that loving DFW)

It's Hot Dog/Not Hot Dog but for creative writing.

When it comes to newspapers isn't this the goal, though? Hemingway's writing is journalistic writing. "The [Kansas City] Star’s style guide formed the basis for his own style that ran against the elaborate tendencies of 19th century writers."[1] So if you're judging the style of a journalist's writing a preferable guide is how similar to Hemingway it is.

[1] https://mediahq.com/famous-authors-also-journalists/

> elaborate tendencies of 19th century writers.

reminded - on some estate sale here in US i randomly picked up and opened a small end-of-19th century book of some American writer that i never heard (that just means that it is definitely not a Tier-1 writer as i'm not an American). The first large paragraph consisted of just 2, yet pretty large, multipart sentences with several great ideas and observations masterfully woven together. I was awestruck. In Russia we call it "Tolstoy" (War and Peace) style and that beat even the Tolstoy. I immediately closed the book, quietly put it back. It was a very uncomfortable reminder of what we lost. The times has changed.

I would say, no not really. The goal is to make writing readable to human, not same as Hemingway wrote. Those are two different goals, despite Hemingway being also readable.

Practically, I don't have problem to understand American major newspapers.

There are plenty of writers - I know several - that seem to think Hemingway's style is the "correct" one. And it may be, for business use-cases, where clarity is more important than beauty, insight, nuance, individuality, or consonance with the meaning of the text.

You will write like a dog for no good reason.

It’s a great example of how people will use an easy quantification just because it exists. There’s a popular mindset that seems to think you can’t argue against numbers and math because they’re cold and scientific. Who cares if its a stupid number, it’s mathematically/scientifically correct!

A measure that becomes a target often stops being a good measure.

Yes, but sometimes it was garbage in the first place, too.

Huh. This is actually quite insightful to me. I wonder if this applies more broadly? What from your own life makes you say that?

It's a variant of Goodhart's law. Can't speak for the person you're replying to, but absent good management, I've seen review goals turn into nonsense. "You had a goal to fix 5 bugs this quarter, and only fixed 4", "but one was a massive bug that would have cost us $x!", "but you didn't fix 5 bugs, so no bonus"

You see it happen on different levels a surprising number of times.

Anecdotally, when I worked in customer support, we were explicitly told that we had to hit a certain number of replies to customers in our ticket system. So I played by the rules: I very clearly grabbed all the easy-to-reply tickets, sent the easy reply, and dropped them if they got hard.

Appropriately, a manager spoke to me telling me this was unacceptable (though he did concede the goal was stupid), and gratifyingly the official goals soon changed to reflect an appropriate balance of consistency with a customer as well as some degree of overall throughput.

The drawback there is that accurately reflecting what they want in hard metrics might not really be possible because it does come down to a judgment call where the exact balance between support quality and throughput lies (and it's often a shifting balance as well, depending how much support load there is). Which is probably why managers fall back on stupid but easy-to-measure metrics.

It does in my experience, I work for a gov public safety agency and the numbers they show you are one part of the story. In public policy we learned how you determine what the thresholds are (When does good become excellent?), as well as things like what does improvement mean (a ten percent decrease in raw number? ten percent decrease from last years average?) mean that the framing of metrics must always be examined. Not to mention the litany of inside baseball stuff regarding how various things like 911 calls, car crashes, crimes, and population figures are captured, analyzed, and understood.

Outside of numbers. The same logic applies to fact checking sites which is why I always question why so many people use them to refute arguements. The authors decided what was deemed a good source.

A dirty secret I’ve noticed of the most controversial issues at the end of the day can be argued successfully both ways, so the fact checkers effectively choose the side they agree with. For example, should Texas be demonized for removing their mask mandates? Lot of areas in Texas are very remote. Texas is more insular of a state than say New Jersey. Federalism and US legal history typically affords states the latitude to make decisions for themselves based on their populations desires. To what extent does interstate commerce (flying out of Texas for example) justify outside/Federal force to upend Texas sovereignty? Lot of people try to argue Texas is just wrong but it’s deeper than that.

Not GP, but that's a formation of Goodhart's law [0], which comes up in a lot of situations.

Notable potential Goodhart's law situations:

- Testing in education

- GDP (or a number of other economic indicators) as a measure of economic performance/target for economic performance

- Healthcare outcomes along a number of axes

[0] https://en.wikipedia.org/wiki/Goodhart%27s_law

New tools, same old tricks. We've just upgraded from turtle shells to web apps.

that's not how it works though...

you can't argue against numbers, you can absolutely argue against arbitrary numbers (like in this example the premise is that the app produces reliable results, which it does not and anyway it should be proved with good, hard to dispute, numbers)

OP's point is that all numbers are arbitrary (to varying degrees). The act of quantifying always imposes some sort of framework on what you're trying to look at, and we live in an age of pretty vulgar scientism where this is associated with 'objectivity'.

From 'personality scores' for applicants, to social credit scores, and in this case the quality of writing in a newspaper.

Right. The number 3 is inarguable. That 4+2=6 is inarguable. The applicability of a give statistic to a given situation or argument is a much more difficult proposition, but people too frequently treat a statistic as relevant just because it's easy to compute.

After all mercury is in retrograde, and that's a scientific fact.

and they are all non arbitrary (to varying degrees), but arguably some are objectively more arbitrary than others.

So I don't know anything about this app really or its evaluations but I think applying it to acclaimed fiction is not a criticism of applying it to a newspaper article per se.

News articles are meant to convey concrete information relatively concisely. So metrics like sentence length and complexity and adverb count are at least in theory valid things for it to measure and aim to reduce.

Now whether it effectively measures those things I have no idea or opinion about. I didn't know it existed until ten minutes ago but if I had I definitely would have predicted that most highly respected fiction would break its rules all over the place.

This isn't a news article: it's a column, written by someone who's explicit remit is columns about the media industry.

It absolutely conforms to different rules than a news article on page A1.

I mean, I think it's bad writing too! But it'd be bad writing if it were in something like the New Yorker, too.

I am a bit surprised about: " The Lottery is considered a masterpiece of a short story, and is widely taught in American high schools."

Really? I have difficulties understanding this. Is it because it is interesting for the kids through its shock value and manageable length? But this then looks to me like declaring ketchup to be vegetables so enough veggies are eaten at the canteen.

Well, I suppose you could read some critical analysis of it if you're interested in why it's considered great; I'm no expert. For what it's worth I agree with its status in the canon: it's a powerful and carefully articulated story in a very compact form with a unique, exciting, and thought-provoking premise. The perfect short story.

Yes it is good entertainment, but wouldnt then the kids read those anyways?

That’s not a given, unfortunately. I believe that the intended long term strategy is to give kids a glimpse of something that is outwardly interesting, in order to increase their internal motivation to seek out other good reading material. Not necessarily to get it “right” the first time.

I don't claim to know the current thinking but, to me, it makes sense that the focus should be on "classic" literature (for which The Lottery qualifies) that's also approachable. I'd argue that a lot of Hemingway and Fitzgerald do as well--though I don't fully understand the popularity of The Old Man and the Sea.

I liked reading in high school and most of the 19th and 18th century works I read bored me to tears and I probably ended up at least partly just reading the Cliff Notes. Today, I expect I would go to Wikipedia.

No. Kids dont like to read all that much these days, they moved to another entertainment options.

One of big goals teachers and schools are having currently is to make at least some kids like reading enough so they do it.

Kids need to encounter something before they're motivated to seek it out on their own.

Are you questioning that it’s a masterpiece or why it should be taught in schools?

For the latter it’s because issues students deal with in school directly parallel those in the story.

You school has something like a death lottery?

The parallel is in the perpetuation of destructive societal practices in the name of tradition, and in people being in favor of those destructive practices until it affects them: then it's suddenly not fair.

You can find that situation in so many places in modern culture that there's no point listing any, it's like looking for air.

I have no intention of using Hemingway or anything like it, but I think it is obviously targeted at business communication, not literature, so it's unsurprising that it would find issues with literature excerpts.

English is not my first language, so it astonishes me that some people really think that short sentences make good writing. Anyone can write bad prose using short sentences. Moreover, by writing in this way the language loses any kind of spice, so it's not a surprise that journalism and technical texts are so unpalatable. Another consequence is that even educated people never stop sounding like they're talking to a ten year old.

But in a sense I realize why this is a problem in English: German an some latin languages have grammatical structure to make it easier to create and understand long phrases. English lacks much of that structure. So, long phrases are difficult to parse. Compare for example Proust novels written in English and in his French original. In English it is almost gibberish, even in good translations.

That's probably true. I would have had to work to parse the first sentence in that article. The other thing going on is that, for many situations, the use of punctuation within long sentences to help the reader has sort of fallen out of fashion relative to breaking long sentences up. It's probably also more acceptable to use sentence fragments in certain types of prose.

> The other thing going on is that, for many situations, the use of punctuation within long sentences to help the reader has sort of fallen out of fashion relative to breaking long sentences up.

Semicolons, colons, and dashes, are rarely used now days.

Heck I am impressed when I see a dash properly used, I myself cannot lay claim to such knowledge. I am almost 100% certain that their usage was not taught to me in either school or college.

I use em dashes a fair bit (and have to remember the spacing style of the publication I'm writing for). But I'm much more inclined to break a sentence in two rather than use a semicolon unless there are two very connected thoughts.

Any tool that uses passive voice as a heuristic for good versus bad writing is immediately suspect.

And any writer who uses such a tool for validation is unlikely to be a good judge of writing to begin with.

Writer here, reasonably competent, because people pay me for writing (in Czech; forgive me my English errors).

Passive voice in narration is ... strange most of the time. It kicks the reader out of the situation, like Out-of-body experience, only it is an Out-of-book experience.

"John was killed by the assassins, his body was later discovered by a cleaning lady, carried out by morticians and dissected by pathologists."

It is pretty hard to feel any emotions around this almost technically cold description. A sentence like this could have come from a police dossier.

I am not a friend of software validation tools in this regard, though. Capturing the essence of beauty in words isn't a solved problem.

Just for comparison, in the active voice:

"The assassins killed John, the cleaning lady discovered his body later, the morticians carried him out and the pathologists dissected him."

It is better, though not massively so. It's tighter. (Personally I'd break it into four sentences, but that's just me. I also would have used an Oxford comma, for that matter.)

The real problem is that the passive voice can be used for:

"John was killed, his body was later discovered, carried out, and dissected".

It's definitely punchy, and as the beginning of a thriller novel its vagueness could be exciting. But any place else you're left mostly with questions. It would be terrible in a newspaper account. And in an actual police report, it would sound like somebody was covering something up.

Your written English is superb. It's indistinguishable from that of a literate native speaker.

Glad to hear that, thank you!

Slavs in general tend to struggle with "a/an/the" in English, because most Slavic languages lack articles. Out of the entire language family, only Bulgarian uses definite articles.

Articles feel unnatural to me, too. I got used to them, but in a similar way that one gets used to a prosthetic arm: you can do things, but it feels mechanic and distant.

> it feels mechanic and distant.

Just like the passive voice, then?

*mechanical

Is what you meant that it is suspect to have passive voice being used as a heuristic?

The article mentions "8 use of passive voice" in the app's report.

All these types of apps do that, sometimes with a recommended max guideline. It's not a terrible heuristic like "Never use passive voice" is. Over some threshold, there's probably too much of it.

But I've always wondered: why should business writing be like Hemingway's writing? He wasn't writing for that audience. Seems like it would be better to find examples of good writing for a business audience and use those. Same thing for journalistic writing: that's not what Hemingway was doing. Maybe take all the Pulitzer winning journalism from a few decades and use that or something. But it just seems weird to pick one arbitrary literary writer for this kind of exercise.

Won't speak to literary/journalistic writing, but business writing definitely has a problem of quantity over quality. The exercise of really dumbing down the language helps with making sure you're actually getting the point across instead of just talking in vague abstractions. It also helps for communicating with ESL colleagues. In that sense, "more like Hemingway" is probably going in the right direction.

I guess it's that I've always thought this is a misunderstanding of what Hemingway was doing. He wasn't trying to write simple language comprehensible to busy business people trying to get their jobs done. He was trying to write good literature as he saw it. Any alignment of the two things is accidental at best but often just a misapplication. He left things out in order for them to exist in the reader's own imaginative journey through the book; his spareness was a way of welcoming the reader into the story. That is not a good thing in business writing!

It’s a good question and one I don’t know how to answer absolutely.

The reason why I think it most generally applies to “business writing” is because this area is most likely to value efficiency over others.

I don’t think it’s meant to say business folk should write like Hemingway, but that Hemingway is the most popular writer in this style.

>>this area is most likely to value efficiency

Agreed. The general style values simple sentences which helps reading speed and comprehension, especially for audiences that don't share your native language. And it reduces fluff. Adverbs, for example, generally don't help make a point in business writing - data does.

See the example in that last sentence? I snuck a premise instead of stating it as a hypothesis or providing data to support it. At best that causes confusion for the reader and at worst it can mislead people that aren't carefully watching out for pitfalls like that.

What was the sneaky premise?

A premise is a hypothesis. We don't explicitly tag all our premises; it's up to the reader to decide if they agree with the premises.

Fair, it wasn't a great example.

>We don't explicitly tag all our premises

True but we should aim to tag more of them in business writing. It's helpful to clearly state them so 1. You as the writer are crystal clear on the assumptions you're making, and 2. if there's any disagreement on premises you can resolve it before you spend any time discussing which course of action to take.

But my quibble is that I don't think Hemingway valued efficiency. It would be much more efficient to write "There was a man in the hospital who had lost his reproductive organs and this resulted in a tragically failed romance" but that is not how he wrote the story, because it's bad literature. But if someone at work asked me to write them an email about it, that's what I would write, because it's much more efficient than a whole novel.

For kicks, I threw a few paragraphs from The old man and the sea to see if the editor displayed any deference to its namesake. Red all over, "very hard to read", "use of the passive voice".

I didn't try that one-- I put in Hills Like White Elephants and it was mostly okay.

The reading level is probably the best gauge for whether an average person will understand what you're saying. If it's reading level 14, that means post high school. Most people in the US graduate from high school with an 8th or 9th grade reading level. So a piece with a 14th-grade reading level will be completely beyond them.

> Knowing full well what it would think of James Joyce, I skipped over him

I like to imagine that Finnegans Wake would result in a crash.

That Shirley Jackson example is particularly galling because her prose is considered by many people to be about as close to perfect as a human can get. Better than Hemingway's. IIRC, her editor said in an interview that she was their easiest client to work with, because her work didn't actually need editing.

I probably agree about the tool, each writer adds their own flourishes and richness to their prose to capture vividly in your mind a scene or emotion. Hemingway probably doesn't understand this.

But the reason he arrives at using it is because he rightly calls out a bunch of prose that is terrible writing and is trying to quantify this in some way. And whilst I have no problems reading long form articles, many these days do tend to suffer from confusing and downright tedious language.

Yup, any kind of automatic quality assessment tool is made with the assumption that quality is even something that can be measured objectively.

Zoe Bee recently made a pretty good video about why this falls completely flat for grammar (rather than just appealing to "X is good, Y disagrees, Y must be wrong"): https://www.youtube.com/watch?v=Q5rB9jDbTPU

Well said. It’s like people think they have to get the machine to validate them instead of the other way around. I was coaching somebody’s writing recently and trying to tell them not to modify their style based on Hemingway feedback, but they just had so much faith in it. Why would so many people use it if it wasn’t reliable, right?

I watched https://www.youtube.com/watch?v=Q5rB9jDbTPU yesterday which goes into details about problems with a similar service, Grammarly.

What Bee said really resonated with me. I used to be a stickler for grammar rules and "proper" English, but as I've progressed in life I've come to realize that those rules are all secondary. All that really matters is that you are understood in the way you want to be understood.

Grammar rules help educate people about common idioms and structures so that they can understand others and be understood themselves. Additionally, by following a certain set of grammatical conventions, you can change the way you are perceived. By ignoring the rules, one might be seen as erudite and sophisiticated. OTOH u dont always want to seem stuckup :)

It is somewhat ironic that a blog about machine learning cherry-picks results from a black-box model to prove a point.

Sounds like a set of training wheels. Helpful for those begging to learn, but extremely limiting if you don't learn to take them off once you have the basics. I can definitely see the benefit of such an application, but the risks are also quickly apparent. Looking at it from one level higher, I think I would be in favor of such an application because those who don't ever take the training wheels off will, on average, gain more than had they never had it while those who would be held back by it are likely to outgrow it on their own pace. The only real threat is if it becomes a tool of authority, like teachers that depend upon it and mark down writing that the app doesn't like.

Can anyone shed light on how it works? There is another one that comes to mind, PTE exam.

The PTE exam is fully computer graded (as far as I know), including written essay. I wonder how they do it. The algorithm has to account for topic relevance, grammatical errors, style, clarity etc etc. Topic relevance alone can be quite hard, isn't it? If the question is "What is your opinion on making voting mandatory? Support with examples", I can talk about voting and elections without actually giving an opinion. How does the algo detect it?

PTE, Hemingway app etc - really intriguing to reduce human language to logic and calculation.

You might enjoy this researcher's automated high scoring essay generator

> Our hypothesis was simple. If the AES machine consistently gave high scores to machine generated gibberish, we could surmise that 1) the construct being measured by the machines is not an essential component of human communication; and 2) students could be taught similar strategies to achieve high scores on computer scored writing tests by sprinkling their prose with long meaningless sentences composed of pretentious and irrelevant words.

> Our greatest surprise was how easy it was to fool all of the machines. We succeeded on our first try, demonstrating that rather than being elegant and complex manifestations of state-of-the-art artificial intelligence, these engines could best be characterized as crude stupid machines.

EDIT (an example):

> We have now used the Babel Generator over twenty times to generate essays for the site, which, when submitted, receive top scores with comments such as articulates a clear and insightful position on the issue in accordance with the assigned task and sustains a well-focused, well-organized analysis, connecting ideas logically” for essays that read like this following opening paragraph:

> Careers with corroboration has not, and in all likelihood never will be compassionate, gratuitous, and disciplinary. Mankind will always proclaim noesis; many for a trope but a few on executioner. A quantity of vocation lies in the study of reality as well as the area of semantics. Why is imaginativeness so pulverous to happenstance? The reply to this query is that knowledge is vehemently and boisterously contemporary.

https://lesperelman.com/writing-assessment-robo-grading/babe...

I believe red and yellow just mean the 'reading level' expected. When writing for a very broad audience, or writing things that need to be short and easy to read like e-mails, you want to aim for 'lower' reading levels.

I pasted some excerpts from a paper I was reading and it gave the 'Post-Graduate' reading level. Does not mean it's bad writing, just hard to read (it's probably inevitable at that level).

It's probably a great tool for daily work, some people at my workplace used Grammarly, but it's paid, so this could be a good alternative.

I think you’re misunderstanding this tool. You’re shoving a Monet into a tool for graphic designers and complaining that it flags it as blurry.

Writing for news should be boring, dry, precise, clear and near painfully explicit with as little subtext as can possibly be achieved. Hemingway’s writing style just happens to be a good example of it.

I think you and people in this thread are confusing news with storytelling/newstainment like John Oliver, lots of Gimlet, Vice, etc. who take the news and up the production value by infusing their personalities and commentary. Despite the crass name this is a valuable service. Good storytellers are the bridge between the dry academic work of historians, journalists, scientists, philosophers, etc. and people outside their respective fields.

tl;dr news isn’t art and can’t be judged by the same measures.

So you don't like the writing in The Economist? I do think it aims to be precise and clear but I certainly wouldn't say it aims to be boring.

Specifically, Hemingway was a journalist. His style is one I would expect to be a reasonable standard for newspapers in particular.

Hemingway is designed for genre fiction - in fact a certain terse style of genre fiction - not for journalism.

Although I somewhat agree with the premise, especially with science and tech journalism.

Most is fine, but every so often there's an article which looks like it was written by someone for whom English is a fifth language.

Blog posts can be hit and miss too, but that's not such a surprise.

You should see what it does to The Old Man And The Sea!

8 of 263 sentences are hard to read. 15 of 263 sentences are very hard to read.

Did you try throwing any Hemingway in there? :)

I agree, Hemingway is a useful tool for editing, but it's not any kind of authority on what "good" or "bad" writing is. It's great for reducing your word count during final edit, but that's about it.

I sometimes write for one of the online tech pubs. The tool you publish through has a reading difficulty score which I always fail. (I mostly ignore it; at most I might tweak a couple of things.) Also all sorts of SEO-related stuff.

Yes, the tool isn't great. But it provides a sanity check to confusing run on sentences or use of adverbs and passive voice.

I also pasted the first page of The Lottery, and it scored it a Grade of 11, Readability: OK, compared to a 14 on the Times article.

Do you find the times article mentioned easy to read or well written?

That Times article is not great. Not sure if I'd have made it through the first paragraph. That said, these types of tools often seem to be about dumbing things down to increase "engagement."

Indeed. When it first showed up here, I tried feeding some Hemingway to Hemingway.

It hated it.

> Seriously, don't use this tool.

Horrible take.

I use the tool for all of my blog posts. It’s a stellar tool with a couple of killer features. I 100% promise if you the tool it will improve the quality of your writing.

That said, it does not identify “good writing” or “bad writing”. It is not sacrosanct. And it definitely can’t write for you.

It’s perfectly valid to leave in yellow or red sentences. But the tool helps make it a conscious decision when to do so.

It’ll improve your writing if you’re a lazy, inattentive, careless writer. If you have any writing ability and pay attention to what you are doing, it’s worthless.

I bet you don’t use spell checkers or intellisense either

Feel free to read my blog. I used Hemingwayapp on almost every post. I’ll let you decide if I’m a lazy, inattentive, careless writer.

https://www.forrestthewoods.com/blog/