Summary of long threads

[ShiftHappens]

Feels like I can't be the only one wishing for this, so sorry if this has already been settled in another thread (but I didn't find one at least...).

A feature I'd really appreciate on MR is a summary of long threads. I often see interesting topics in the forums and start reading the discussions, but often the same (more or less) opinions are reiterated over and over again and it takes a lot of time finding the gems and if you feel like commenting on a post, it's hard to find if someone else have already done so, perhaps saying what you want to say.
For MR stories, you can see the top rated comments (which is good) but I think more can be done in terms of summing up long threads.
One approach could be to have e.g. GPT write a summary of a thread with the viewpoints outlined and if there is anything close to a consensus (would be rare, I know).
One area where this would be really useful is where an issue is discussed and a solution to the issue has been presented (much like in many other forums where a solution is highlighted right after the initial question to spare others the trouble of reading the whole thread...

Thoughts?

 

[Boyd01]

Seems like this would be difficult to implement because the summary would need to be updated every time somebody posts to the thread. And, since you're talking about long threads, this would happen very frequently. I think the summaries would also need to be reviewed by a human for quality control and that could be a signifiant workload. And what criteria would be used to identify "long threads"? That would be a constantly moving target, wouldn't it?

Beyond that, I'm personally opposed to using AI to create content at MacRumors, although that's a policy issue above my pay grade.

 

[circatee]

Seems like this would be difficult to implement because the summary would need to be updated every time somebody posts to the thread. And, since you're talking about long threads, this would happen very frequently. I think the summaries would also need to be reviewed by a human for quality control and that could be a signifiant workload. And what criteria would be used to identify "long threads"? That would be a constantly moving target, wouldn't it?

Beyond that, I'm personally opposed to using AI to create content at MacRumors, although that's a policy issue above my pay grade.

...here, here!

 

[ShiftHappens]

Seems like this would be difficult to implement because the summary would need to be updated every time somebody posts to the thread. And, since you're talking about long threads, this would happen very frequently. I think the summaries would also need to be reviewed by a human for quality control and that could be a signifiant workload. And what criteria would be used to identify "long threads"? That would be a constantly moving target, wouldn't it?

Beyond that, I'm personally opposed to using AI to create content at MacRumors, although that's a policy issue above my pay grade.

Yeah, I understand that it's a leap you shouldn't do without proper analysis to start using AI generated content, but I do think it would be interesting to see what an AI generated summary of a long thread could look like (this for example=... Maybe it could be generated on request (which of course can be done without building it into MR - side project idea..! .
But I guess adding the top rated comments for all threads exceeding, say, 50 comments would be a fairly low hanging fruit?

 

[BrianBaughn]

It actually sounds like something worth testing but ChatGPT can't access or open external links…so some coding to get the content into ChatGPT would be required. OP you're welcome to experiment

 

[Weaselboy]

You can click "Sort by reaction score" at the top and that will show the most upvoted posts. I think that would often highlight the best answers to a question in a thread.

 

[ShiftHappens]

You can click "Sort by reaction score" at the top and that will show the most upvoted posts. I think that would often highlight the best answers to a question in a thread.

Fascinating I've never seen that link 🙃

 

[floral]

Whenever there's a long thread I just ignore it. At that point it isn't worth messaging in there when hundreds of people did the same...

 

[ShiftHappens]

Whenever there's a long thread I just ignore it. At that point it isn't worth messaging in there when hundreds of people did the same...

Probably a smart strategy, but at the same time - a long thread could indicate it’s an interesting (maybe more likely a provocative/controversial) topic…

 

[Scepticalscribe]

Whenever there's a long thread I just ignore it. At that point it isn't worth messaging in there when hundreds of people did the same...

If this is a topic of possible interest, are no other opinions worth reading?

And, even if hundreds of posters have already replied with posts, (some of them well thought out), is that not a good reason not to offer an opinion of your own?

 

[Scepticalscribe]

Seems like this would be difficult to implement because the summary would need to be updated every time somebody posts to the thread. And, since you're talking about long threads, this would happen very frequently. I think the summaries would also need to be reviewed by a human for quality control and that could be a signifiant workload. And what criteria would be used to identify "long threads"? That would be a constantly moving target, wouldn't it?

Well said.

Beyond that, I'm personally opposed to using AI to create content at MacRumors, although that's a policy issue above my pay grade.

I find myself in complete agreement with you on this subject matter.