Hacker News Reader

Comment by elgenie | Hacker News Reader

FB audited a bunch of metrics and found issues with 4 out of 220. The narrative being pushed here is that all of them nefariously favored FB. The only problem with that is that it's not true.

> """Elsewhere, Facebook admitted to exaggerating the number of full views that video ads received"""

You can tell they didn't do any homework, since the issue was that full views were under reported when the audio and video tracks were not the exact same length. The result is that full views were actually roughly 35% greater than what page managers saw reported.

http://newsroom.fb.com/news/2016/11/an-update-on-metrics-and... goes into the details of what the issues with each metric calculation is. You can draw your own conclusions as to whether each was (a) important and (b) favored FB. That link also goes into FB's plans for next steps in making metrics externally audited and more transparent.

View on HN

"FB investigated FB and found FB not guilty"

Fwiw sounds like they are opening up to independent third party review with viewability vendors like Integral.

Now if only they'd let me use DCM tags to measure impression data and VTs.

DCM tags? VTs?

'DCM' = DoubleClick Campaign Manager tags

'VT', in this context, probably meant 'viewability tracking' or 'viewability targeting'.

And people worry we won't have enough jobs when the robots take over.

Actually in this case VT refers to View-Throughs, ie. post-impression performance or conversions that occur from someone who sees a display ad impression, doesn't click, but later converts.

Ah, that makes better sense. Thank you!

A valid point, but it's ludicrous to report it as "FB found guilty".

Well, yes, it was just a paraphrase, maybe a bit exaggerated in this context.

So what? That's what FB can do when they find an issue. You are free to sue them if you feel that you have lost money because of this. Go ahead, that's what courts are for.

But I guess what many really want is for government to step in and use government force to punish FB until people feel that some kind of justice has been done. (burn the witch!)

That's why democracy is fundamentally incompatible with liberty.

On a side note: The media is attacking FB because they feel that FB doesn't censor and steer public opinion in the way they would like it to be. They are all up in arms claiming that Trump is FB's fault after they themselves have reported 24/7 on Trump. (but so biased, even Stalin would have cringed) That's the only reason why this article even exists.

> You can tell they didn't do any homework,

It's not a question of "doing their homework". That's the naive cop-out. The inflation of metrics and numbers is company-wide, and I am pretty certain most - if not all of it - is deliberate. Especially after 'Zuck' took the Company Public.

The NY Post are the ones that didn't do their homework. Their source for this article is FB trying for transparency.

If you have proof of inflating metrics and numbers, especially deliberately, you're welcome to post some. Posting detailed reports of errors discovered during an internal audit when under no obligation to do so, on the other hand, is not in keeping with your conspiracy theory.

It's not a 'conspiracy theory'. See my other comment. I worked at the Ad Team for a few months. Saw it from the inside.

>Posting detailed reports of errors discovered during an internal audit when under no obligation to do so

You dont think a public company is obligated to audit internal controls and keep shareholders and regulators in the loop?

A public company is obligated to report quarterly, in a prescribed format, a bunch of high-level information material to the business by Regulation S-K and Regulation S-X.

A public company is not obligated to rush out details like "We’ve determined that the average time spent per article had been over-reported by 7-8% on average since August of last year. This was caused by a calculation error: we were calculating the average across a histogram of time spent, instead of reflecting the total time spent reading an article divided by its total views."

Forget about this article and consider the video metrics they inflated. They counted a 3 sec play as a video view and that's a big thing. 4 out of 220 is still a big number for a big company such as FB to be wrong at.

> They counted a 3 sec play as a video view and that's a big thing.

3+ seconds representing a single FB view has been the definition of a video view on FB since the metric became public.

And it was considered dishonest way to count views since then too:

https://medium.com/@hankgreen/is-facebook-lying-about-video-...

this is subjective evaluation, but they have been open about it and not lying on our faces.

I get the big brother corp hate, but let's be reasonable and factual with it if we must go this way

> I get the big brother corp hate, but let's be reasonable and factual with it if we must go this way

I fail to see how calling FB out on cherrypicking criteria for counting conversion on videos is invalid because those criteria were made know ealier. This is IMHO alike of arguments over e-bay auctions doing "image scam" not being scam because you knew what was sold, after all it was there in auction's description. It's still dishonesty.

It wouldn't be a problem if they discontinued autoplaying. The average FB user needs more than 3 seconds to find the X button.

What I don't understand is why people think FB lying about metrics helps them. Aren't these people keeping track of sales? Those should show that if FB was getting all these views they aren't worth much as their sales didn't go up.

Course the clients might be running several campaigns at once and can't tell what is driving sales. If only they knew what half of advertising dollars worked!

'Actually roughly' is the best

If you need help parsing, it's "actually (roughly 35% greater)".

Reply on news.ycombinator.com