|
Finally! I thought it might have been overlooked when DD last reported it and I was about to report it myself. Looks like I can post comments now though.
|
|
|
|
|
Been a busy week with lots of code branching, so I've only just had a chance to upload a new stable build.
Well, as stable as they get 
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Chris Maunder wrote: stable build
Oxymoron. Perhaps "a build that meets an arbitrary minimum level of stability" would be more appropriate.
|
|
|
|
|
Maybe just "build"
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Has a certain ring to it. 
|
|
|
|
|
Chris Maunder wrote: Maybe just "build"
indubitably[^]
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
As you know questions can have an accepted answer. The OP (and only the OP) can choose which answer to accept. This hasn't been working out as expected. Authors simply don't use this feature. Either they just don't want to be bothered or don't understand the system.
We were thinking of removing the "Accepted answer/alternate" functionality altogether. Instead we'd have the idea of "Best answer" or "Viable alternate" (wording undecided) for questions and tips/tricks, respectively. An answer/alternate would be marked as such if it had at least 2 votes with a rating >= 4.
So basically the OP would no longer be responsible or even have the ability to mark their question as accepted. It'll be decided by the community on votes. This has the added benefit of allowing newer answers to overtake an older slightly less accurate answer - not to mention far more questions having good/top answers.
|
|
|
|
|
I don't see much use for such feature at all; not the "accepted" one, and not the "best" one.
"Accepted" refers to the original enquirer, and isn't very relevant to the community.
"Best" is debatable and transient; whatever is best today, tomorrow may bring a better answer using the same basic technology.
Luckily there still is real voting with a range of values, from low to high; the average score really says all. IMO whatever you'd add to the score, it does not add value, it just confuses people, and opens meta discussions.
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
Yep I'm with you. I wasn't clear enough in my original post. No one will explicitly choose the "best" answer. It will dynamically be determined based purely on the votes/rating. So if an answer happens to have 2 votes of 4 it will also be highlighted (on the view page). And on list pages we'd have the answer count bordered in gold to indicate that a good answer exists.
|
|
|
|
|
Thiru Thirunavukarasu wrote: Yep I'm with you
I'm not so sure; I think you didn't get my drift.
What I said is I don't need anything but the average score, so no "best" either, no gold mark. All those qualifications add little or nothing to the value of the one question and the answers it has got.
But now it seems you want an automated way of splitting questions in two categories: those that need more attention (because no, no good, or no good enough answers are present), and the others (that probably have a sufficient number of sufficiently good answers). I can understand such need, I don't agree with your proposal.
IMO you can't solve that purely based on average score, you need to use a weighted quantification too. Having two beginners each say perfect answer, doesn't make it fully answered.
This is what I would use for sorting questions by their need for attention:
for each answer calculate sum of voterWeight*(vote - 3)
of all these, pick the largest value; use that to sort questions
Example:
question 1 has 2 answers:
answer 1 got two white fives. Valued at 2+2 = 4
answer 2 got a single platinum five. Valued at 8*2 = 16
the "answered value" is 16
question 2 has 1 answer:
answer 1 got two white fours. Valued at 1+1 = 2
the "answered value" is 2
question 3 has 2 answers:
answer 1 got 5 white fives, and a platinum 1: Valued at 5*2 + 8*-2 = -6
answer 2 got a single silver 4. Valued at 4*1 = 4
the "answered value" is 4
since it scores lowest, the second question needs most attention.
This proposal does not use absolute values; it only quantifies the degree of "answeredness".
Final thought: whatever you do, people will try to abuse it. You may end up in a silly situation where all (including bad) answers on other questions get up-voted to make those questions go away, so the own question gets more attention (the opposite of your "latest best picks" phenomenon for articles on the home page). Including voter weight will reduce such risk.
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
Luc Pattyn wrote: You may end up in a silly situation where all (including bad) answers on other questions get up-voted to make those questions go away, so the own question gets more attention
You are a devious person Luc!
I would never have thought of doing something like that.
Thanks for the heads up. 
Henry Minute
Do not read medical books! You could die of a misprint. - Mark Twain
Girl: (staring) "Why do you need an icy cucumber?"
“I want to report a fraud. The government is lying to us all.”
Why do programmers often confuse Halloween and Christmas?
Because 31 Oct = 25 Dec.
|
|
|
|
|
Life becomes somewhat easier when you consider the potential side-effects of some decisions or actions before committing... It is part of the analysis phase, don't you think?
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
Luc Pattyn wrote: It is part of the analysis phase, don't you think?
Indubitably!
Henry Minute
Do not read medical books! You could die of a misprint. - Mark Twain
Girl: (staring) "Why do you need an icy cucumber?"
“I want to report a fraud. The government is lying to us all.”
Why do programmers often confuse Halloween and Christmas?
Because 31 Oct = 25 Dec.
|
|
|
|
|
Henry Minute wrote: You are a devious person
devious, another word Google Translate has trouble translating into other languages such as Dutch and French. All it gives would translate back to unwieldy, verbose, lengthy, winding, sinuous, divergent, aberrant; while I have the impression you rather meant cunning, maybe even deceptive.
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
These things haven't improved greatly in something like 20 years. (Well, they have really, but I'll ignore that for the sake of the story. )
Roughly 20 - 25 years ago I read an article in a computer magazine, can't remember which one, where the author (might have been Guy Kewney) had been invited along to a University who wanted to show off some translation software they were working on.
After they demonstrated he asked if he could have a go. He translated "to be or not to be, that is the question" into another language and back to English. What came back was "to shall or not to should, that is the interrogation".
Henry Minute
Do not read medical books! You could die of a misprint. - Mark Twain
Girl: (staring) "Why do you need an icy cucumber?"
“I want to report a fraud. The government is lying to us all.”
Why do programmers often confuse Halloween and Christmas?
Because 31 Oct = 25 Dec.
|
|
|
|
|
Automatic translation of sentences must be very hard; every language has its idiom and idiosyncrasies, and so has every author.
I use Google Translate many times a day to look up single words (mostly the concoctions of a couple of British CPians, such as DD and POH), and most of the time it is sufficient to understand what is said and meant; it typically is no good at translating sentences, it does not know enough about grammar, and it has an annoying tendency to just copy every word it does not recognize, so each time a word looks the same in two languages, I'm uncertain whether it exists at all in one of them.
I also checked indubitably, to make sure my leg wasn't being pulled; but it was what I thought it was. 
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
Luc Pattyn wrote: to make sure my leg wasn't being pulled
As if  [^].
Henry Minute
Do not read medical books! You could die of a misprint. - Mark Twain
Girl: (staring) "Why do you need an icy cucumber?"
“I want to report a fraud. The government is lying to us all.”
Why do programmers often confuse Halloween and Christmas?
Because 31 Oct = 25 Dec.
|
|
|
|
|
An unfamiliar quote. I'm afraid I have been watching Gophers in Dutch only, in ancient times.
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
question 1 has 2 answers:
answer 1 gets two white fives. Valued at 2*1*5 = 10. Normalised = 4, Rating = 5
answer 2 gets a single platinum five. Valued at 1*8*5 = 40. Normalised = 16, Rating = 5
the "answered value" is 40, Normalised = 16, Rating = 5
question 2 has 1 answer:
answer 1 gets two white fours. Valued at 2*1*4 = 8. Normalised = 2, Rating = 4
the "answered value" is 8, Normalised = 2, Rating = 4
question 3 has 2 answers:
answer 1 gets 5 white fives, and a platinum 1: Valued at 5*1*5 + 1*8*1 = 33, Normalised = -6, Rating = 2.54
answer 2 gets a single silver 4. Valued at 1*2*4 = 8, Normalised = 4, Rating = 4
the "answered value" is 33, Normalised = 4, Rating = 4
question 4 has 1 answer:
answer 1 gets 10 platinum ones. Valued at 10*8*1 = 80. Normalised = -160, Rating = 1
the "answered value" is 80, Normalised = -160, Rating = 1
Summing the votes normalised over 3 certainly provides a benefit when sorting like this, and is definitely easy to implement.
The main discussion point, however, is that
1. Members do not hit "Accept Answer" consistently enough for it to be of reliable value
2. A better answer may be posted after the accepted answer
3. We need a way to identify questions that need some loving.
4. No system is perfect. Let's aim for the 95% of cases.
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
I think we agree on this subject!
Chris Maunder wrote: The main discussion point, however, is that
1. Members do not hit "Accept Answer" consistently enough for it to be of reliable value
2. A better answer may be posted after the accepted answer
3. We need a way to identify questions that need some loving.
4. No system is perfect. Let's aim for the 95% of cases.
I think I covered 1 to 3 in my message?
1. I see no need for "Accepted"; it does not work, and even if everyone would use it correctly, it would still offer dubious value.
2. Indeed (marking it accepted only makes it less likely to happen).
3. So that is what a sorting criterion like my "answeredness" could do: offer a way to sort questions, at the reader's discretion, by "lack of good answers", "attention deficit", or whatever you want to call it.
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
It shall be done.
Thank you, Luc.
[Edit] Hang on - I just need to clarify something:
1. We still want to mark a question as having been addressed. We are currently rely on two indicators: An answer (any answer) and the "Accepted" (annointed) answer. We want to simplify this into a single "Acceptable Answer", hence Thiru's original question.
Your proposal gives us a good way of sorting anwers. However, we still want a threshold above which a question is to be deemed to have an acceptable answer, regardless of whether the original poster agrees.
Peers who understand the topic and vote accordingly have just as much, if not more, ability to regard an answer as "correct" as the person who asked the question (and who, by implication, doesn't know the answer, and can only treat answers as suggestions)
So what is "acceptable"? A Gold or above voting 4? 3 Silvers voting 5?
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Chris Maunder wrote: It shall be done.
![Jig | [Dance]](https://codeproject.freetls.fastly.net/script/Forums/Images/jig.gif)
Chris Maunder wrote: We still want to mark a question as having been addressed.
If you insist on having that, I would use the same calculation used for sorting, and apply a threshold value, or better yet two thresholds. Anything below the first is "not answered". Anything above the second is "experts agree", and everything in between is "answers are still being gathered". A color code could represent such state; that makes readers aware of it not being black-and-white.
For "experts agree" I would recommend the value that corresponds to:
a platinum five plus a platinum four plus two bronze ones.
So two experts don't have to agree on a five, and a couple of morons are allowed to down-vote without a net effect. And this might be one of the few formula you keep secret (and of course change when you feel a need to).
Now whatever absolute value you impose becomes a risk. Some day all platinum people will be on holiday, and you'll need a lot of silver & gold to jump in.
I'm really not convinced we need it.
And you will get new questions, such as: how many questions (in %) are not answered? fully answered? how does that compare to other sites? etc etc. There will be no end to it.
Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]
I only read formatted code with indentation, so please use PRE tags for code snippets.
I'm not participating in frackin' Q&A, so if you want my opinion, ask away in a real forum (or on my profile page).
|
|
|
|
|
Baby steps first. Let's work on a "acceptable answer" (lower limit) then we can add a "Experts recommend" limit (upper). But yes, I do like that idea.
cheers,
Chris Maunder
The Code Project | Co-founder
Microsoft C++ MVP
|
|
|
|
|
Thiru Thirunavukarasu wrote: had at least 2 votes with a rating >= 4.
I don't see much of 2 votes either! Which again would look like nothing has changed... In this case, system depends on viewer voting which too doesn't happen much in a Q&A forum!
Thus, i don't think this too would help much...
Alternative: Limit enquirer's on asking question until they close the last one... (lets say there can be a pool of maximum 3 open question of an enquirer at one time)... now cases available that needs discussion:
1. 3 open questions, none of them replied yet
2. 3 open questions, 1 or 2 replied but not appropriate answer
If OP wants to post new question, they need to close any one of them. Apart from 'Answer Accepted', there should be an option of 'No Answer/Answer not appropriate' or so (wordings can be decided later.)
This would give OP an option to close a question even if they don't get a proper reply.
Further, in a given period of time.. lets 2-3 days... if a question can be answered then generally it is answered... it rarely happens that question is answered after a week or so. Thus, its an auto reply to OP that no one currently can reply them back on the issue!
|
|
|
|
|
I think it's still a big step up. I'm sure there will still be a number of questions without an answer with at least 2 votes but that number will be much lower than it has been till now.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
