Criteria for reviews

  • 1
  • Problem
  • Updated 5 years ago
  • Not a Problem
Archived and Closed

This conversation is no longer open for comments or replies and is no longer visible to community members. The community moderator provided the following reason for archiving: Old thread

I am not new to IMDB, but I am trying to submit a review for the first time. I find it unbelievably stupid that the review form mechanically tells me that I "don't have enough lines", when I've written more than a dozen. Undoubtedly, the idiotic software designers chose to quantify number of lines based on hard return/line-breaks, rather than any real, reasonable measure. Sure, I could shove a return character between every sentence, but that's not how a good writer writes. For the love of God, will someone lash the idiot who made that decision and put someone in charge who knows the hell what s/he's doing?
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes

Posted 5 years ago

  • 1
Photo of bluesmanSF

bluesmanSF, Champion

  • 10815 Posts
  • 6431 Reply Likes
Wow. How completely inappropriate. I bet your reviews are real gems if you think that's how writers write.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
Wow, how completely ignorant. I'm a successful writer, a successful web developer, and a highly valued software quality-assurance engineer. What are your qualifications, "bluesman"?
Photo of bluesmanSF

bluesmanSF, Champion

  • 10815 Posts
  • 6431 Reply Likes
What are your qualifications
At what? Recognizine an a-hole? Just a lifetime of reading things like your post, of course.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
Fine. The vast gap of misunderstanding that exists through the Internet is a dangerous trap. Sorry to have irritated you.
Photo of bluesmanSF

bluesmanSF, Champion

  • 10815 Posts
  • 6431 Reply Likes
You're suggesting I misunderstood your ranting and cursing like a six-year-old? You meant to say something different?

You know what...don't bother answering...you're not worth the waste of the second or two it'll take to read the next stupid thing you type.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
What made you so angry?
Photo of OJT

OJT

  • 15 Posts
  • 5 Reply Likes
If you want to say something in short writing, use "message boards" instead of reviews.
A review should be a review, and that should take some lines.
Back many years ago, there wasn't a rule like this, but then we got a lot of comments which were far from reviews, and most of them today, still there, are quite uninteresting to read. 
Use message boards instead!
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
Thanks for the helpful suggestion...a proper review would, indeed, be more than a few paragraphs.
Photo of Nobody

Nobody

  • 1455 Posts
  • 707 Reply Likes
I've spent a few minutes playing with the review editor (just "testing" it, not really submitting a review). When I click the "Preview" button, the editor apparently performs a temporary reformatting step prior to checking the line count. I think I almost understand how that works (but I won't try to explain it here). ... I wonder, though, why they don't simply count the number of words instead?
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
Precisely!
Photo of Nobody

Nobody

  • 1455 Posts
  • 707 Reply Likes
Precisely!
Not quite.
We disagree on a couple of points, one being this:
... Sure, I could shove a return character between every sentence ....
That apparently would have no effect. See the following screenshots.
In this test, I entered 12 sentences, each followed by a single hard return:



In the screenshot above, you'll notice that each sentence is followed by a hard return. I did not use any double hard returns i.e. paragraph breaks. So, you might expect the system to treat the text as a paragraph with a line count of 12; but that's not the case.

Here's what happened after I clicked the Preview button:





As you see in that last screenshot, it appears that the system removed my original hard returns between sentences. The system then reformatted the text in the box, and then counted the number of lines in the box. The resulting line count is 9, not 12.
(Edited)
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
Oh, no, Mr. (Ms.?) Nobody--I agree with you completely on this particular point, and I was merely speculating about something you actually tested. Fantastic!

My beef with IMDB is that their algorithm for what's long enough to be a "review" makes no sense. Your testing helps to confirm my thesis. The question of how long a post has to be in order to be considered a "review" is entirely separate, as I'm sure you know.
(Edited)
Photo of DavidAH_Ca

DavidAH_Ca, Champion

  • 3261 Posts
  • 2917 Reply Likes
IMDb's original submission was via email, with a 72 character line for text items. I expect this was carried over to the Reviews. I believe the internal processing was done by Perl (though I'm not sure), and counting lines is much simpler than counting words, although they do now use a word count for the maximum..

Using a standard of 5 characters per word, then a 72 character line would be about 12 words (allowing for the space between words). Would a limit of 120 words be that much more helpful than 10 lines?

The limits are explicitly stated in User Reviews Guidelines which states :
Please note there is a 1,000 word limit on reviews. The recommended length is 200 to 500 words. The minimum length for reviews is 10 lines of text.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
Thanks for all the quantitative information, David. I hope you won't be offended by some suggestions:
  • The problem I commented on was the requirement for minimum length of a "review"; most of your commentary is about the maximum length--but that's an important question too.
  • I spent 12 years as a technical writer, and it all started on a new-product tabloid with a limit of about 100 words per piece. Just 100 words sometimes was barely enough to convey a fraction of the important characteristics of a simple electrical or electronic part.
  • When I was in the publishing industry, the standard measure of word count was 6 characters per word; has cable news lowered the standard from a 6th-grade level to a kindergarten level?
  • Last and most important, the problem is with how the software counts lines. It's obvious to me that the number of words doesn't count at all; the software seems only to count hard returns. This is just plain inexcusably bad programming.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
P.S. - Your guess that Perl was used to parse the text is reasonable, but only because that's what Perl was designed to do. I haven't looked at the page source code, but what I would do is use Javascript, in order to avoid having to repost the page just for a simple character-validation check.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
Geez--the more I think about it, the more disgusted I am with the incompetence that people get away with.
Photo of DavidAH_Ca

DavidAH_Ca, Champion

  • 3261 Posts
  • 2917 Reply Likes
  • When I was in the publishing industry, the standard measure of word count was 6 characters per word; has cable news lowered the standard from a 6th-grade level to a kindergarten level?
I used 5 because I remembered that as the number used in calculating typing speed. I may have remembered incorrectly.
  • Last and most important, the problem is with how the software counts lines. It's obvious to me that the number of words doesn't count at all; the software seems only to count hard returns. This is just plain inexcusably bad programming.
Actually, it seems to me that Nobody's post indicates the opposite. His test seems to indicate that they ignore single hard returns and wrap the text at about the 72 character limit I mentioned and that is what constitutes one line.
Photo of Nobody

Nobody

  • 1455 Posts
  • 707 Reply Likes
Actually, it seems to me that Nobody's post indicates the opposite.
Sorry if I my previous comments may have obfuscated some details. If my understanding may be (more or less) correct, the sequence of events would be something like this:

(1)  A user begins writing a review. Between one paragraph and the next, the user enters two successive hard returns to separate the paragraphs. The user might (or might not) choose to enter single hard returns between lines within paragraphs.

(2)  The user presses the Preview button.

(3)  The software that processes the text (I'll just call it "the software") apparently removes any single hard returns that the user had placed within paragraphs. So, at this point, there will not be any hard returns within paragraphs.

(4)  For each paragraph or contiguous text of any length longer than a prescribed line width (apparently about 70-72 characters or thereabouts), the software adds new hard returns to reformat the text to fit the prescribed line width. (I neglected to clarify this in my previous comments.)

(5)  If my understanding is correct, after such reformatting the software counts the resulting number of lines in the text box.

For typical text, on average, probably a few more than a hundred words (approximately) should usually suffice to satisfy the requirement.

However, the current implementation has at least a couple of problems:

If the user puts a double hard return following a period, the software (correctly) retains the double hard return (usually meaning a paragraph break). But if the user puts a double hard return following other punctuation (e.g., after an exclamation mark or a question mark or a colon), the software (incorrectly) discards that double hard return.

The current implementation allows users to enter shorter reviews by splitting them into short lines, or (at worst) even just single words, each followed by a period and a double return. In this way, a user could submit as few as ten words. Obviously, this is not good.

A simple word-count based limit should be straightforward to implement and could avoid the aforementioned problems.

(Amusingly, even a simple word-count limit could raise a silly controversy if some users may disagree on exactly what constitutes a "word" -- or whether some hyphenated terms should be counted as single words, or whether numeric values should be counted as words, or whether a proper name or title should be counted as if it were one word, or whether short words should be counted fractionally or given a free pass, or whether long words should be penalized, or whether certain parts of speech should be given a pass, or whether a review's average word length should be taken into account and the limit adjusted accordingly, etc. ... Some users might argue that a word count may be a bad idea anyway, and that a character count should be used instead.)

But (silly arguments aside) it seems to me that a simple word-count limit would be reasonably fair and probably would be less problematic than the current implementation.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
David, 5 characters/word isn't unreasonable--I was just being curmudgeonly. I'm not 100% up on the latest software libraries, but I'm having trouble conceiving of a variable or object method that would measure *wrapped* lines in a Text Area. But I'm not the best one to say.
Photo of Frank

Frank

  • 12 Posts
  • 0 Reply Likes
"Nobody" - that's a great detailed analysis/reverse-engineering. Some email applications and sites insert hard returns at word-wraps, but, in my opinion, that's not a user-friendly action. So, however IMDB does it, it strikes me as very user-unfriendly.

The issue of how long a post has to be to be considered a "review" is in part a separate issue, I guess. Knowing now that they have a place for shorter "notes", or whatever they call them, does resolve the problem somewhat.

This conversation is no longer open for comments or replies.