Word average count keeps on dropping even though I write logs.

Question

In the past week, my word average has dropped from 69 to 64 words.  Why?

Answer 1

We recently found bugs in the code that processes logs that would count things that were not really words as being words. Since that has now been fixed, some word counts may be somewhat lower.

Comments

Somewhat?   I've lost almost 80,000 words.  Not all at once - but rather about 12,000 each day.  i.e.  1,919,000 words yesterday  1,907,000 today.  My author badge has gone from level 88 to level 85.

And shouldn't that have been a one time reduction?

It might be nice to have an example of what was being counted - but is no longer being counted.

---

This change means that all logs (millions and millions of them) have to be reprocessed, which is expected to take a month or so in total. I don't know exactly when this began, but I think there's still some work to be done here. Thus, it's not surprising that the change happens gradually.

---

OK - that makes sense.  However it would still be nice to see an example.  The 80K or so words that I've dropped over the last week or so amounts to over 4% of my total.  I would prefer not to use "non" words in my logs.

---

I agree with gcstraggler. Until recently, it was clear what counted as a word, and we created our notations accordingly. Without warning, the old words are now lost to us, and the levels achieved are also lost. And it is not clear what is a word and what is a non-word.

---

It’s about as clear now. There were adjustments before as well, only that some didn’t work properly.

---

Okay.  Let me ask then.  Would this be an example of text / notation / entry no longer counted as a word.  I tend to, perhaps, over use shortcuts that may be understood locally (southern California) like SFV in place of San Fernando Valley or SGV in place of San Gabriel Valley.  Would SFV,  SGV still be counted as words?  or would they be not counted?

---

I assume abbreviations like that would be counted. Numbers and smileys would not.

---

Besides properly removing numbers and smileys as pinkunicorn mentions, we now also remove html and markdown more properly.

If you happen to know what log entry you had before which had a very high word count, and have lost it. Look at that log and count the words yourself, to see how many words you think there are. It's very likely that you will come to a number that was lower than Project-GC thought it was before.

Counting words aren't easy. It's not black and white, it's very much grey areas all the time. Logs that will be affected the most with these changes are most likely logs on Challenge caches with "proof" that are fairly long lists of stuff, especially if D/T was listed.

---

By the way, about 35% is now processed. The counter is up to ~370 million. The total is a bit over a billion.

---

I wish you would implement something that just ignores these meaningless copy&paste signatures from the word count.