3D Warehouse stats out of whack

3dwarehouse
tallies
graphs
stats
downloads

#1

Here we go again, except worse; The downloads counter just credited me with as many downloads in a few hours as usually occur in a few weeks. It’s been off since last month’s tweaks, but has gotten much worse this evening, to the point of now being completely unreliable. On top of which, the UI even disagrees with itself.

This is one example, but they’re all off by a wide margin. The likes jumped too, but without the usual E-mail notifications that accompany them, which means they’re but the product of glitches as well.


#2

:slight_smile: Couple weeks ago asked about the doubtful downloads counting and got a very “manly” answer: “So what?” or something to that effect. Hope you get a better response.


#3

I would ‘hope’ the lower figure reflects actual ‘unique’ user downloads…

i.e. it ‘should’ exclude any repeat downloads to the same address and all downloads to the file owner…

if that’s not the case then the system is open to people rigging the stats…

john


#4

Thank you for your fast answer @john_drivenupthewall,

It is the 3rd time that I get an off-topic strong, not very polite answer to a simple arithmetics question:

  • started the topic on https://blog.sketchup.com/article/3dwarehouse-materials as i thought the bad counting was related to the 3DW upgrade. The answer i got there from a titled poster (not sure about the real title) was “Not here, move this to the forums”;
  • so i moved it here and i got a fast “Who cares” from another non-ordinary poster;
  • I noticed the new crazy behaviour of the counter and I joined a topic that mirrored my doubts and the answer you (another non-ordinary poster) give us is “maybe the counter is screening out the self-downloads of the creator” implying (willingly perhaps) that we may repeatedly download our own “creations”, but again no answer to the real thing. Is it so hush hush?

For your information i attach the image of a “Staff picked model” statistics:

As you can see, very few downloads for 25 days (not sure if May 17 was the new update launch, I noticed the bad counters on May 18), then the “downloads” explode and culminate with 5,474 “possible self-downloads” of the author, as per your theory. Hmm.

Edited -> very few downloads for 25 days - the correct number is 390 days


#5

You already opened a thread on this issue.

The proper category is Technical Problems > Warehouses (where I’ve also moved this thread.).
If you check the correct category, you’ll most likely post in the right place and avoid duplicate issue threads.

Ie, your previous thread:

… in which @Barry told you that the graphs are not yet sync’d with the property box on the right and that they are working on things.


And also a few others threads on the same issue:



#6

I’m experiencing the same problem. Something’s really haywire with the way the stats are presented.


#7

I was not implying that either of you where rigging your own stats, in fact, @franquin makes it clear he is surprised by the unexpected upsurge in his own stats…

I was making an on-topic observation of one reason they may differ, if it is not a programming error…

If it is an error, as it seems it may well be, then I hope any fix can show both total downloads and unique downloads…

john


#8

I opened a second thread since I got a “so what” on the 1st one, and the situation took a turn for the worse…
I didn’t think anyone’d read a post in a thread that had been inactive for a couple of weeks.


#9

Thank you Dan to pointing out those posts; one of them is even mine and you can see that it has no real answer.

Until about yesterday (or Sunday maybe) it was like this: uploaded new model, checked it after couple of hours, saw 4 downloads, checked it again in other couple of hours, found 0 downloads but the graph side showed 5 downloads … repeat, same going on with both new and old models, etc. Seemed like download numbers are baaad but graph numbers are good. Asked about it, got the “Not here”, “Who cares” answers, swallowed it and adapted.

What i was joining to is signalling a new behaviour of the counters: since yesterday the graphs (not the “Downloads:
XX” like before) went berserk: they show large jumps in both views and downloads and this is why I took the liberty to post again. I did wrong it seems…

You are now the fourth “exceptional poster” that cares not for the issue but the presentation.


#10

No you did not get a “so what?”.

@Barry (who is responsible for the Warehouse) gave you a precise answer:

It is likely that different programmers worked on the graphs and the property box, at different times.

The answers boil down to:

  • They are working on 3DW features.
  • They know the graphs and the property box are out of sync.
  • They want to get them sync’d up.
  • It takes time (via established engineering protocols) to process changes to be done.
    (The Warehouses are themselves a product. It is not like one guy tweaking his personal webpage.)

Yes you did wrong.

Duplicate posting is prohibited because it just confuses the issue and spreads the discussion about the issue all over the place, instead of one thread where everyone can participate. And it makes makes more work for SketchUp Team members. Now they have to respond in multiple threads instead of one master thread.

A forum sage answered you immediately, telling you at that time it was already reported (ie, you had posted a duplicate issue.) He also asked whether it was a “big deal” ? This is what you call the “so what?” reponse.

Even though Dave, doesn’t have control over the 3DW, sages often ask questions to attempt to get posters to clarify their problems or state their case better. It might be he genuinely wanted to know why it was so important an issue for you. But you never answered this.

@Lorand and @franquin, both of your badges pages show you have not earned the “Read Guidelines” badge. When you joined the forum, you were asked to read the guidelines.

How can you know how to present your issues and topics correctly, if you do not even read the forum rules ?


#11

This is what left me with the “so what” impression:

To which I replied that the margin was 0% until the then-recent tweaks were made to the site. I’ll admit to not having read the guidelines, but will do so now. I just hope I’ll both understand and remember them, which isn’t always the case with me, and the reason why I hadn’t done so up to now.


#12

I don’t know how you got “Who cares?” from my question. You misread it. I acknowledged that there seems to be an issue with the numbers and I genuinely wanted to know what difference the accuracy of those numbers makes to the way you are using the 3D Warehouse and SketchUp. Do those numbers change whether or not you upload models/components to the Warehouse? Do they affect something else?

I asked because I’ve never cared about the number of views or downloads any of my shared models or components.

No need to answer my questions now, though.


#13

It most definitely was not intended as a “so what”. I was speaking in general terms there on any site. Whether it’s internet advertising, TV or radio numbers, people will ALWAYS wonder how many people truly saw their content. That doesn’t mean “so what” or we don’t take it seriously or that you’re wrong, that means “this is a normal occurrence in this space”.

There are some numbers that have changed a bit in the past 12 hours. Specifically, as of GMT midnight today, we pushed some new code to 3dwarehouse that may bump downloads a bit. In short, there are a number of ways people can download models (any browser, Window->Components, Window->3DWarehouse, etc…), and some paths were not being counted accurately (we under-counted).

You didn’t notice that Likes are on that chart? That’s new.

But specifically for Wall-E: that’s what happens when you’re featured and have an interesting model. People view it and download it. There’s absolutely nothing screwy about that. Look at other featured models to see the same affect. Note: because I don’t accept your premise that the Wall-E model data is wrong, it does not mean “so what”. You can find models that were featured, then rotated off the featured list, and their numbers drop as well.


#14

Your post of May 28: “There is a couple of other threads on this topic. I agree it’s odd but is it really a big deal?”

I read it as “it is not important”; two other people read it as “not important” and answered you accordingly.

But if you really think I misread it, I must apologize.


#15

Finally a real answer, thank you very much.


#16

And no one noticed the “daily” data as well?

And an example of a model that was featured for a little while (June - Sept of 2014, when we had ported 3DWarehouse from Google internal infrastructure to our own). Note that 90% of it’s views and downloads came during that 3 month period:


#17

I certainly have noticed the newly available daily readouts, and have used the functionality to try and sort out the discrepancies between the different download readouts. That worked until this latest event, where the numbers grew so far apart that no reconciliation can be done anymore.
Another thing I’ve been wondering about for a long time: Do our own views of “my 3d Warehouse” count as visits, or is there a distinction made between genuine browsing and simple “tabs”-keeping?
Should I open another topic in this regard? Or, for that matter, regarding the suggestion I’d make that maybe a moderator should decide what category a new topic should belong to, as this is often not cut-and-dry.


Should moderators decide what category a topic/query belongs to?
#18

does it screen for multiple downloads from the same user?

john


#19

No, and when we left Google, we did get download data but not users tied to it, and that’s continued at Trimble. Unlike click fraud, we assume users don’t do a ton of download duplicates. If someone has a use-case for that, let me know.


#20

Let’s start another thread. This description is fairly broad, and interest is growing. That also buys me time to get the answer right. :slight_smile: