In response to:
- I feel like the quality of posts in Dev is deteriorating
- Improving the way DEV elevates the good stuff
This is quite an interesting case for study.
On one hand, we have websites that optimize for popularity (engagement) of posts, like hacker news and reddit. Users can upvote and downvote content. Those sites tend to be toxic and elitist. It is hard to get on top, but if the author gets on top they will get a lot of views and comments.
On the other hand, we have dev.to which the main goal is to keep it as beginner-friendly as possible. Commitment to CoC makes it non-toxic. But the average quality of content tends to be low (personal opinion). Because of a big number of posts and no "real" rating, good authors lose motivation to write - the good article will be buried in a flood of bad ones.
Is there a way out? What are the alternatives?
- Is the reputation (or karma) system tends to make community elitist? In my opinion, the absence of CoC makes the community toxic (not a downvoting function).
- What if we would allow downvoting, but will show counter only to the author (so there is no public shaming)? The same way as dev.to doesn't show the number of followers. And as soon as the post passes some positive number of likes it would get some badge (50 likes = 🦄, for example).
I don't know what is the solution here. I'm throwing some ideas to discuss.
Beginner friendliness is a nice goal, but some authors didn't bother even to put in some content and send this post to feed of thousands of users. How is that different from spam? There should be some reasonable line between beginner friendliness and respect to the other users, right?
Quality
I also feel like the quality of the average article on dev.to is quite low.
Important: I'm criticizing it because I write for dev.to myself. So the critics can be applied to me as well. Only a ginger can call another ginger.
But there is a trick to it. How would you define the quality of a post? If you read the comments of the first post you would see that there are different opinions.
So let's discuss options - what we can consider as poor quality. Options are arguable - not a final judgment.
Violation of CoC
Not arguably bad.
Spam
spam - unsolicited usually commercial messages (such as e-mails, text messages, or Internet postings) sent to a large number of recipients or posted in a large number of places
I guess everybody would agree that spam is bad.
Ways to fight: restrict a number of posts for new accounts, restrict a number of outgoing links for new accounts, spam detection algorithms (like, akismet or similar), moderation, spam reporting system.
In general, this is a well-known problem for ages on the Internet, and there are well-established ways to fight it.
Note: it feels like it is easier to post spam on dev.to than report it. To post you don't need to pass CAPTCHA, but to report spam you need to prove you are not a bot 🤷♀️.
Plagiarism or copy-paste
plagiarize - to steal and pass off (the ideas or words of another) as one's own: use (another's production) without crediting the source
Not arguably bad.
Note: do not confuse with cross-posting.
Lack of content
Posts without any sense, like unreadable text or only title with "test" or similar.
Ways to fight: give caution to the author "you about to post an empty article, while you can do it, we recommend to add at least some text"
Factually wrong
Posts that do false claims about facts. The same way as Twitter and Facebook tries to fight misinformation, we can add a small disclaimer to the post, like an official source on the subject ...
Misuse of terminology
Sometimes happens that terminology in software development is misused.
Opinion presented as fact
A lot of things that can't be proved with facts are presented as if they are facts, for example, "how to become a better developer by avoiding else statements".
Poor text quality
Posts with (big number of) grammatical errors, poor style.
Highly arguable, because there are a lot of non-native speakers. Maybe we can allow moderators to help fix grammar 🤷♀️.
Poor formatting quality
Posts with poor text formatting, for example, code without fences or fences without language specified, lack of text structure e.g without headers.
Ways to fight: gamified introduction to the markdown editor
Shallow content or poor retelling
For example, "Array methods in JavaScript", where 5 random methods were selected and showed, but no value was added compared to reading an "official" manual, like MDN. Value in such articles could be an attempt to explain array methods with real-world analogies, for example.
Arguable: those articles are fine if they presented in the vein of "today I learned".
List-articles
Everybody likes lists, but reading "top <Number> of <Things> for <Subject> in <Year>" again and again is kind of counterproductive.
The problem with those articles is that they lack any kind of research - you can find the same list or even bigger one in "awesome repositories" or with a simple Internet search.
List-article would make sense if it is exhaustive or contains some kind of comparison or this is unique content.
Tisers
Posts with partial content with a link to the original article.
Arguable, but I don't understand the reasoning behind this tactic. dev.to has canonical links, so SEO should be okay. What is the sense to make users click one more time?
Offtopic
Something that has absolutely nothing to do with the subject at hand...
Arguable. There is even a tag for it #watercooler
Top comments (31)
For the stuff about off topic posts, it should only be off topic to a certain extent. There is a fine line between "Is it ok to only have computer related hobbies" and "How to know if your wife is cheating on you".
How about "How to know if your wife is cheating on you using node.js"?
okey ..... this made my day.
Honestly, we, instead of trying to make 100% of the posts quality, could try and break it
into 50/50, in a non-toxic way.
I could see this being done in one simple way, promote people with low follower numbers, AND high quality posts. And handle (in any general way) posts with low to no quality.
What defines quality could be user-dependent, or, could simply be controlled by the staff, but, whatever determines "quality" should be (and this is almost a given), fair.
Maybe typos don't set off some kind of "Bad Quality Detector", but, maybe certain wordings that may be incorrect, or lazily put together will (though, this would probably work against who's native language isn't English, though, writing in native language would be preferred, then).
Or, maybe posts would be manually reviewed if it gets reported enough for low quality.
However "quality" may be implemented, or interpreted is up to anyone who makes any changes regarding this topic.
But, my personal opinion is, "Quality doesn't matter, community does".
If an other feels like their work isn't represented as much as it should, or, in the example you gave, they feel like their rating doesn't properly reflect what people think, then, they can voice that in a post, and ask for people's honest thoughts and feedback.
As a community, if we voice, and work together, we can build "quality" from what we already have. Like making a
class
in C fromstruct
s.Cheers! <3
@ben already created a post talking about how their are set of users called trusted users who have the ability to mark articles as low quality and high quality.
Yes, we are working hard to fight spam by flagging the bad post as spam to admins and marking the quality (high/low) of the posts.
Well, thank you for the information.
Oof! That's a hot conversation here.
The problem with quality is not the quality of the post but the target audience
By keeping the bar high - you are making site better for seniors, but completely not friendly for juniors. I know a few sites who went this way (habr, reddit, hackernews), and result is not great.
By not doing it - you are doing completely the opposite. And here we are - I am able to see articles in a language not known to me at the main page.
As the result I am reading stuff from the people I follow, and just not opening the main page ever.
And from stats of my own stories I can clearly see that others does the same.
Frankly speaking - I've lost any hope to get any good reads via dev.to and writing something here mostly to explain some things to myself 🤷♂️
hmm you raise some valid points. I just happened to ask a bunch of my co-workers about DEV posts and a couple of them replied with a similar vibe: "isn't that just full of newbie JS articles?"
I think the DEV community treats beginners openly (which I think is good), so it's natural it will be a saturated with content tailored to beginners and FE developers.
Where do you find good reads?
I've lost all hope on twitter to find good reads...
First problem to solve is to find a good writer, as they produce a good read.
The second problem is that just a few good developers you can learn something from are good writers, or writing anything even if they can.
The third problem is youtube and online sources, luring so many authors to a different format of education...
In short - I am a lucky to find a good reads from time to time, but it's usually based on talking to people, and be given a good link to read during the conversation.
In this case your duty is to share a
good read
further. It will be forgotten if you (yes you, whoever read this sentence) do you best to spread it.Just to reflect my own point - this is a few good reads (and one watch) I was given last month:
Oh dear, I feel you @jennrmillerdev I've recently created a Twitter account to connect with more engineers. I find Tech Twitter to be JavaScript or React.Js Twitter. I feel left out, it's like there's nothing else other than JavaScript. 😞
I'll have to jump in and defend these platforms here: HN isn't toxic most of the time and with reddit it depends mostly on the community. I've seen many users here on dev start to become toxic and they get shut down instantly, which makes a much bigger difference than any up-/downvote system.
As for elitism, that's just completely untrue. Content on either of these platforms (in the case of reddit I can obviously only judge the communities I interact with regularly) is ranked almost exclusively by its content. There's no such thing as an elite that gets special treatment or more relevance.
This. I like reading about users learning and figuring out new things, but when those sorts of posts are framed as tutorials, I feel kind of cheated. It's a similar thing to clickbait: the content might not even be bad, but it was advertised as something else.
If they are framed correctly however, I might just skip them on some days, and on others have a look at them and maybe even add a bunch of extra info or just my opinion if it's a topic I am familiar with.
It really depends. Sometimes people actually do make good lists of things that aren't widely known / used, and I do appreciate those. However, many of them also fall into the "shallow content" category. If your top 10 software of any kind list includes VS Code, vim and git, you're not providing any value.
EDIT:
As for a rating system, instead of a simple up-/downvote system, I would prefer having a 1 to 10 rating system. Hick's law is probably the best if not only reason to have a simple binary rating system, but I'd love to have a bit more of a spectrum to rate content. Some posts are just pointless while others might just need a little improvement here and there.
It'd also be cool if giving a bad rating would tell the user to explain their downvote à la "You've downvoted this post. Why not leave a friendly comment explaining what could have been done better?"
100%. Just be honest about it - "I learned this yesterday and here are my thoughts..."
True - to write this kind of list author needs to do a research. But a lot of list articles are just result of 10 minute internet search.
Yes. Would be nice
I totally agree with some of this some people write some really nice articles, software and more but this gets flooded by 100 version on how to use array.map in javascript for instance.
I finally became a trusted user and will really try to promote those hidden gems.
I had also suggested but to no avail a top 10 / top 5 hidden gem posts every week. Were the dev.to staff could pic 10-5 posts which weren't that popular but were very interesting!
Just because lists or opinions aren't appealing to you personally doesn't mean that they should be subject to content moderation. Just scroll onwards.
As you linked at the top, there have already been changes to the platform to fight low-quality articles. Consider just taking a few days and let the changes take effect. If you scroll even now I think you'll find the quality of the feed has improved dramatically over the past few days.
I don't think that is what the author meant. I believe they meant that lists should be subject to content moderation because they usually lack research effort and are usually low quality articles.
But my point is that you can't just say "low quality articles" as though it's an objective fact and not largely just a personal preference.
Don't get me wrong--I know exactly the kind of article to which the writer is referring, but even at their worst, they're still helpful starting points for absolute beginners, and we have a lot of those on the platform.
Yeah, I think we all know of the 'kind of articles' the OP was talking about, and I can see how people might judge them of having low research effort. To be honest, I also grow tired of seeing them too :P
One problem is see is this:
I think it's hard to judge what is low quality or not. We can see people trying up their following counts simply by pumping out 'low quality' articles. The community seems to accept them because they have high followers on twitter...
yet for the beginner with no followers, we judge that same kind of content as being a lower quality "list post".
What's the difference between the two then? Just my 2 cents.
Yeah, I tend to agree. Though I can understand why people are becoming tired of seeing the same kinds popping up all the time, those are still valid content to those who are new or interested.
I don't know if it has already been proposed, but I think what I would like to see is some sort of elected representational boosting algorithm.
There's people who write really good articles and they are people I like to follow. But it would also be nice to have articles they have saved and liked boosted in my feed as it's likely I will be interested in similar articles to them and given that they usually write really good articles it's likely that the articles they read I ones I would also like to read.
It's also highly likely that I would want to read articles saved, etc. by people they are following in the same manor.
I'm not saying this should replace the current model, but would it be nice to have it as another top category available (like Feed, Week, etc.).
I would like to add some sort of showing posts that I like based maybe on posts that I click. Even tho I have Javascript as a tag I'm tired of seeing post like "How to use arrow functions", I mean there's no problem about article for people who are new at javascript even that, as you said, don't offer anything new related to MDN, but why should that appear on my feed?
I have to scroll a lot to find a valuable post
It may be too intensive, but maybe some form of buddying/mentoring system for authors (peer article writing anyone?) to help with the syntactical parts (grammar, formatting, etc) and the tagging and subsequent filtering, even perhaps content writing advice.. a #needhelp tag? Do we have a tag for #contentwriting advice too - "how to write a good dev.to article" exists I'm sure!
This lifts those who want to do better, but doesn't address the intentional junk though, for which I hope the trusted member ability to push back a little will reduce their impact sufficiently that spammers go elsewhere - the joy of being a popular platform!
It's a perennial problem in scientific journals too, trying to balance the effect of 'quality checking' / peer review that results in following the 'authority' biases (and creates echo chambers) against openness, that can result in junk papers being published with 'journalistic authority'.
I wonder if the fact that DEV has a "saved" articles feature contributes to the problem? If you take a look at the top articles of this month, a lot are list-articles that tell the reader "you need to know XYZ as a developer".
It could be like people see this title, go "oh, I better save this article for later, it could be important" or "this list of project ideas could come in handy later" but it's not really indicative of the quality of the content.
I believe book marking an article counts towards the reactions on an article, and the number of reactions on an article would most likely spread the article around more on DEV. If the number of reactions would spread an article more, they should probably remove book marked reactions because as you said, people thinking that it may be important but haven't been able to look at it shouldn't make it more popular.