We've now been through three rounds of Facebook disclosing information about political ads that targeted Seattle's 2017 municipal elections.

Round one failed to comply with Seattle law and, as I reported two months ago, failed to add up.

Rounds two and three?

They're not in compliance with Seattle law, either, according Seattle Ethics and Elections Director Wayne Barnett. They also fail to add up.

Here are just a few of the ways:

In order to comply with Seattle's unique law on election transparency, Facebook has now shared images of hundreds of ads that were purchased to influence this city's 2017 elections. It also has sent Barnett three spreadsheets (and one revised spreadsheet!) containing some data on how much those ads cost and who they targeted.

Facebook's first spreadsheet was two pages long. Its second spreadsheet was 84 pages long and extremely difficult to read. (Like, 3.5 size font difficult.) Its third spreadsheet was more readable and clocked in at 604 pages. That represents a sort of progress.

The problem is that none of these Facebook spreadsheets or image files add up to the full disclosure required by Seattle law, which states that a commercial advertiser like Facebook must provide, upon request:

1. The names and addresses of persons from whom it accepted political advertising;

2. The exact nature and extent of the advertising services rendered; and

3. The consideration and the manner of paying that consideration for such services.

This means that, for example, if someone asks Facebook to disclose all the ads bought by the 2017 candidates for Seattle mayor (as I did last year), then Facebook's disclosure should include all the Facebook ads bought by 2017 mayoral candidate Nikkita Oliver.

As many in this city will remember, the Oliver campaign had a notably strong Facebook presence during the 2017 mayoral primary.

Yet Facebook has disclosed only one paid ad from the Oliver campaign (shown above) and it has acknowledged only $100 in Oliver campaign ad payments.

According to Dujie Tahat, who was a spokesperson and ad manager for Oliver, the Oliver campaign definitely ran more than one Facebook ad and paid Facebook significantly more than $100.

The Oliver campaign's own filings with the Seattle Ethics and Elections Commission show it spent nearly $2,000 on Facebook ads. That's about 20 times the amount Facebook has so far disclosed—months after being asked for the information.

So what's going on here?

Why, more than four months after I made my request, and more than three months after the City of Seattle told Facebook to turn over local ad information, are we sitting here with only one Facebook ad disclosed for a Seattle mayoral campaign that definitely bought more than one Facebook ad?

The problem seems to originate with Facebook's success at getting Barnett to buy into a flawed process from the get-go.

With apologies for quoting myself from two months ago:

...It appears that Facebook had Barnett send the company a list of Facebook page URLs connected to the campaigns and independent expenditures involved in Seattle's 2017 municipal elections. Then, Facebook provided Barnett some limited data on political ads purchased through those URLs during 2017.

But as Facebook must know, and as the campaigns' disclosure filings clearly show, doing things this way is not going to give us an accurate picture of the total amount spent on Facebook ads aimed at Seattle's 2017 elections...

A lot of the Facebook ads targeting Seattle's 2017 municipal elections were bought on behalf of various campaigns through third parties—political consultants who handle a candidate's social media strategy. It seems likely, based on the above-mentioned discrepancies, that a significant number of Facebook ads targeting Seattle's 2017 municipal elections were not purchased through the particular Facebook pages Barnett sent to Facebook.

In the case of the Oliver campaign, Barnett asked Facebook to disclose ads bought through the Seattle Peoples Party Facebook page.

Well, what if that's not the only Facebook page the Oliver campaign used to purchase its election ads?

With apologies, again, for quoting myself from two months ago:

The law involved here, it's worth noting, is titled "Commercial Advertisers' Duty to Report."

The burden, it would seem, is squarely on the commercial advertiser to know when it's selling a political ad aimed at Seattle's elections.

By relying on Barnett to send over specific Facebook URLs, and then only giving out information relevant to those URLs, Facebook effectively transferred the burden to Barnett, making it Barnett's responsibility to know all the Facebook pages used to purchase ads aimed at Seattle's 2017 elections.

Clearly, that process didn't produce an accurate picture. Perhaps, in the future, the responsibility for knowing about all the Facebook ads purchased with the aim of influencing Seattle's elections should be on... Facebook.

We are now two months into that future.

At present, Barnett is still considering how to respond to Facebook's latest disclosures and arguments.

But it's hard to see how anyone could argue that Facebook has so far provided a complete picture of the Seattle political ads it sold in 2017. Beyond the issue with the Oliver campaign ads, there are plenty of other instances of Facebook's disclosure strategy failing to produce accurate information.

One baffling thing about Facebook's disclosure strategy is that the company began (in early February) by disclosing exact dollar amounts spent on local ads by various Seattle campaigns and Independent Expenditure groups. Then, after I pointed out that Facebook's disclosed spending amounts were different—in some cases wildly different—than the amounts the campaigns and IEs themselves reported spending on Facebook ads, the company changed its approach.

Facebook's later disclosures (which came in March) only offered a "range" for how much was spent on each Facebook ad purchase.

For example, Facebook's later disclosures reported that the campaign of Cary Moon spent between $3,500 and $18,046 on Facebook ads.

That's a huge range.

This week, Barnett told Facebook that spending ranges like this are "too broad to provide campaign watchers useful or meaningful information." The company, Barnett said, needs to offer exact dollar amounts rather than ranges.

In response, Facebook's Seattle lawyer told Barnett that Facebook has already disclosed exact dollar amounts—in that first, flawed spreadsheet the company sent over in February.

(That's the same spreadsheet that Barnett, at the time, pronounced a failure in terms of compliance with Seattle law. "Their two-page spreadsheet doesn’t come close to meeting their public obligation," Barnett said back then. "I’ll be discussing our next steps this week with the City Attorney’s office.")

What's baffling about this week's argument from Facebook is that no matter which way one looks—whether at Facebook's first spreadsheet, its second, or its third—the numbers don't add up.

To stick with the case of Cary Moon:

• Facebook's first spreadsheet put the Cary Moon campaign's ad spending at $7,812.

• Then, Facebook's second and third spreadsheets offered the previously mentioned, very wide range for the Moon campaign's spending: $3,500 to $18,046.

• But whether one is looking at Facebook's earlier offer of an exact dollar figure or Facebook's later offer of a wide range, the numbers don't match what the Moon campaign tells me it spent on Facebook ads. That number, according to Moon's political consultants and ad buyers, is more than $20,000. (And could well be closer to $30,000—I'll have a more precise figure after talking with a particular Moon ad buyer on Friday.)

This is far from an isolated case.

Take the Independent Expenditure campaign that backed Moon. It was called "People for Moon," and while it reported its spending vaguely ($25,000 in "digital advertising," with no particular platform specified) I've since gathered more details.

The political consulting firm behind "People for Moon" says it spent $3,000 on Facebook ads.

To date, Facebook hasn't disclosed any ad spending at all by "People for Moon."

These issues continue when it comes to Facebook disclosures that are connected to Seattle's current mayor, Jenny Durkan.

In his recent e-mail to Facebook, Barnett cited the wide ranges the company has now disclosed for ad spending by both the mayor's campaign and "People for Jenny Durkan," an Independent Expenditure that backed her.

By my tally, Facebook's latest disclosure says the Durkan campaign spent between $6,200 and $32,849 on Facebook ads in 2017.

Again, that's a very wide range.

It's so wide that the difference between the low and high numbers works out to $26,649—which is greater than the amount the Durkan campaign actually reported spending on Facebook ads.

That number is $23,866.49.

Further, the number the Durkan campaign actually reported spending on Facebook ads—again, $23,866.49—is more than twice what Facebook, in its initial, exact-dollar-figure disclosure, reported in connection with the Durkan campaign. That figure is $11,906.

Same problem for "People for Jenny Durkan."

Facebook says the group spent between $14,600 and $70,177 on ads on its platform.

The difference between the low and high numbers works out to $55,577—which is greater than the amount "People for Jenny Durkan" reported spending on Facebook ads: $42,006.

And the group's actual reported spending—$42,006—is considerably higher than the exact-dollar-figure disclosure Facebook has provided for "People for Jenny Durkan." That figure is $34,997.

(This discrepancy is close enough that it might be explained by commissions charged by the third-party vendors that purchased the Facebook ads for "People for Jenny Durkan." But "People for Jenny Durkan" also disclosed buying $75,000 in vaguely described "digital ads," which could well include Facebook ad purchases and make this discrepancy even greater. I've been trying for a very long time to get more information from "People for Jenny Durkan." I'll update this post when and if that happens.)

Barnett has said that digital platforms that sell ads targeting Seattle's elections should be disclosing "information on the intended and actual audiences for the ads." (In other words, each ad's targeting information.)

But Facebook, as I demonstrated in the case of City Attorney candidate Scott Lindsay's targeted crime ads, has not yet revealed the full details on the "intended and actual audiences" for each Seattle ad it sold.

Lindsay told me his ads were targeted at specific Seattle neighborhoods, and at specific voters within those neighborhoods. Facebook's disclosure didn't reveal that level of detail for Lindsay's ad buys, or for any other ad buys.

The company only disclosed that Lindsay's ads were targeted at "Washington." In terms of other targeting possibilities—whether by race, age, income, or gender—Facebook provided very incomplete information on the ads' "intended and actual audiences." It offered only what it called "Age Group Age Gender Reach" information, but that information a) failed to make each ad's "intended" audience clear and b) didn't disclose the full, "actual" audience that each ad reached.

In response to my requests, Google has disclosed ads that targeted Seattle's 2017 elections. So has Pandora.

Like Facebook, Google appears to have fallen short of disclosing full details on its ads' "intended and actual audiences."

Pandora may well have failed in this way, too.

Google is planning to send the City of Seattle a revised response soon. But all the responses from Google and Pandora so far have been, in one key respect, very different from Facebook's responses.

Unlike Facebook, neither Google nor Pandora made it someone else's responsibility to figure out who bought political advertising on their platforms targeting Seattle's 2017 municipal elections.

Pandora and Google took that responsibility upon themselves.

Even so, thanks to Facebook's most recent, still-problematic disclosures, we do now know one thing that could have major implications for how this all unfolds. The company has so far identified 648 discreet expenditures on Facebook ads targeting Seattle elections in 2017.

If Barnett ultimately finds that Facebook has failed to fully comply with Seattle's disclosure law, the fine could be up to $5,000 per violation.

If each discreet Facebook ad expenditure constitutes a violation, then Facebook could owe the people of Seattle a large chunk of money: more than $3,200,000.