Julie Rubicon. Confession of a Former Facebook Employee

2024 Author: Seth Attwood | [email protected]. Last modified: 2023-12-16 15:55

Writing it all down is the last thing I want to do, but it's necessary. Partly for the people who need to know what's going on with their Facebook posts, but mostly (99%) for Julie Rubicon and that spike on the chart.

My former colleagues from Facebook Inc. in Menlo Park, California - hi Jane, hi Neil, hi … Mark? - will immediately understand who wrote it, and the company will probably follow me, but I think they will turn it over quietly. The SEC won't just quietly investigate if the relevant rules and regulations are indeed violated, but honestly … hardly any such rules exist.

I am writing this on February 27, 2016. Today is my last day on Facebook. I turned in my badge, my laptop and walked out onto Willow Road, with the flash drive and the screenshots you see below. On the street, he silently watched as corporate applications disappear from the smartphone screen, one after another.

It was weird to feel like I wasn’t a Facebook employee, even though I was expecting to be fired most of the time. I started in the development group, it didn't work out very well there, moved to the advertising department, where it was even worse, and ended up in PIG (Partner Intellectual Group). The PIG department is where my story begins.

Each Facebook user sees only a narrow, personalized snapshot of the system. For Facebook itself, there is a much broader picture available. From my PIG work terminal, I could run queries on all posts with comments, open and closed. According to personal messages too. I might ask: How many people on Facebook mentioned the US presidential election today? How many people wrote something about Donald Trump? How many of these messages contained an emoticon? (On the morning of February 27, if you're interested: 65 million; 42 million; 32,541).

Advertisers love these statistics, but obviously Facebook can't give them direct access. (This is really obvious, right? It should be obvious). Therefore, there is an intermediate link. If you spend a lot of money on Facebook advertising - I never got to know how much - the PIG department will prepare special reports for you about how your brands / products have been discussed throughout the system. As of last October, these numbers included mentions from Instagram and WhatsApp.

There is nothing sinister about such statistics; all internet companies do it. The data is carefully anonymized. This is a view of a billion users from a height of 10 kilometers.

Just for the sake of example, here's a timeline I prepared for Adidas last year, with some helpful notes.

You can see the peak on the graph in mid-February, just after the launch of the Yeezy 750. Here's a simple story: cause and effect. My partner in the marketing department of Adidas showed this graph to the bosses and said: look - we have bent reality!

There were three people working for us in the PIG department: me, T and Julie Rubicon, whose real name I give for reasons that will soon become clear. I came to the department with minimal expectations, but it turned out to be great, because T and Julie turned out to be smart and interesting people. Tee was really passionate about the different brands and / or products we analyzed; our clients really liked her, and it was clear to me that she belonged to a different camp, demanding reports, not preparing them. (Keep it up, Tee!)

Julie Rubicon was different. She came to PIG from the department of work with users, got rid of the monotonous routine and yearned for new tasks. I was lowered into the PIG, and Julie clawed her way in here. This difference was visible to everyone, including Julie, and on the second day at the PIG she called me an asshole. For this I will always be grateful to her. My resentment turned into anger, which hardened to determination and, in the end, I learned to really do my job. Julie and I became friends.

In the spring of 2015, I got a very strange graph.

All PIG requests are processed by an internal application called Enchilada. These queries have two parts: a comma-separated set of keywords ("yeezy 750, yeezy boost, yeezy 750 boost") and a date range: start date, end date.

I was preparing a report for Vernix, a brand of children's shoes. I entered keywords as usual (““vernix, babyboots, vernix babyboots” ), but when setting the time range, I made a mistake and did not specify the end date.

I sent this request almost a year ago, in April 2015. It's strange to remember that.

Enchilada was supposed to return an error message, but something went wrong in the data center, and instead of an error, a graph came, which ended for some reason in October 2016. The data in meaningless squiggles continued into the future.

Future peak

I found this to be a strange bug and repeated the request. The parents discussed Babyboots vigorously.

On June 1, 2015, Vernix announced a purchase from Nike, and our partners ordered all PIG statistics to be consolidated under a common account. Something clicked in my head. I rummaged in the only overflowing folder where I dumped all the prepared reports, and dug up that erroneous graph.

Nike buys Vernix

The peak on the chart matched perfectly.

It turns out that I had a graph with a demo of the discussion of buying Vernix, received two months before this discussion began.

I spent almost an entire day on the roof of the MPK20 [Facebook headquarters building] feeling uneasy and anxious as I gazed out over the bay.

That evening, I sent a request to the system that was not associated with any client. The NBA playoff final was about to begin, the Golden State Warriors v Cleveland Cavaliers, so I asked Enchilada for a schedule of reports for each team. This time I again did not indicate the end date, now on purpose.

Enchilada did not predict the winner as such. She just predicted the number of discussions. But discussions strongly correlate with real events, which is why advertisers are primarily interested in them. The peak of discussion happens when the winner is determined. When companies merge or go bankrupt. When politicians get into scandals. When people die.

On the day of the sixth game, I showed Julie Rubicon's schedule. A series of rising peaks predicted the outcome perfectly, with the largest and sharpest peak indicating that Golden State would win the final in great style.

The next morning Julie came to my desk: “Come with me,” she said. She and I crossed the Bayfront Expressway, and there, on a path near the salt marshes, with a loud voice to shout down the wind, she said that we were starting to trade stocks.

We took a list of the largest companies in the world. We acted with caution. Every system on Facebook is monitored, and the flow of stock symbols will surely raise suspicion from one of the admins. Typically, the PIG department generated about a dozen inquiries per day; we decided it would be safe to add two more queries every day.

We didn’t tell T about this, and T, if you’re reading this, forgive me.

As we worked out the list of companies, we put forward different theories about the country of the new "feature". Facebook developers have been able to connect powerful neural networks to many of our systems in recent months. Could it be that Enchilada was hooked up, intentionally or not, to something like advanced artificial intelligence that not only analyzed but extrapolated the data, and not only plausibly, but perfectly? May be. Or a rat gnawed at the cable. Or it could be that this rat was magical.

We received charts two at a time. Apple and Exxon Mobil. Berkshire Hathaway and Google. For the most part, they looked like a random set of values. Incoherent. The future was boring.

And then in August I saw it.

Just in case you didn't understand from the y-axis, it was a very large peak. A real giant among the peaks. It was September 2015.

Can you guess?

The graph shows all posts and comments on Facebook, open and closed, from the recent past and possibly the near future with the mention of … Volkswagen. I told Julie that it must be something terrible. I mean, a car company? We have compiled a lot of reports on the release of new car models, a huge number. They never looked like this.

Julie with me, each of us, bet $ 2000. I read a bunch of blogs and learned how to short a stock and bet all the money against Volkswagen. We made $ 1000 that September day. Happy Julie gave me a high and we celebrated our success by walking in the salt marshes.

We were eager to continue, but peaks of this magnitude were elusive. In November, something appeared on the Walgreens chart. But I was wrong. As easily as we earned a thousand, we lost it.

Everything continued until December: small wins, small losses. We found the magic lamp - there was no doubt about it - but I had the nasty feeling that the genie inside does not speak our language. Come si dice, "I want to get rich?"

Julie Rubicon was quicker as before. She called me outside again and said: forget about this intraday trading. “Let's become data brokers. Let's launch a secret company hidden inside Facebook. It will be possible to access it only on a website in a hidden network through Tor, it will only give out raw data, without interpretation, payments only in bitcoins.

Julie has read the novels of William Gibson.

“There will be word of mouth about us,” she said. - Hedge funds will want it. They will come to beg us."

Julie was eager to start.

I read even more blogs, learned how to make sites on the hidden network Onion, and just launched such a site on the day Julie, on the last Monday of January, for some reason unknown to me, sent a request to Enchilada about … us.

She asked for our names.

Here is a graph for me.

Exactly what it should be. I am a normal person, not a celebrity or a politician. Not a brand.

And here is Julie's graph.

Julie's chart

The peak of discussion happens when the winner is determined. When companies merge or go bankrupt. When politicians get into scandals. When people die.

What in the world could have caused the peak of discussion for Julie Rubicon?

That day, looking out into the distance of the salt desert, she said with confidence in a trembling voice: "They know."

No, I assured her. They cannot know.

“They will find out. They will look at the logs. We're going to get in trouble. They'll jail us."

“I don’t think there are any laws against this,” I said. What laws can there be against something impossible? I asked her not to worry.

But not only this peak worried me.

January spilled over into February. At the office planning meetings, Julie looked normal - crisp and confident as always - but every time I quietly asked if we could talk about something secret, her eyes turned cold and she answered: later. I'm busy with the Puma.

It was always Puma, for a reason.

I didn’t enter any tickers anymore, didn’t short the stocks. There were no visitors to our dark web site because no one knew about it. Occasionally, I have run requests for presidential candidates with no deadline. I will not say what was there.

And I saw Julie check her name over and over again. Each time on the chart, there was the same mid-March peak, towering and unforgiving.

The morning after Valentine's Day, smart red napkins were still hung around the PIG department. Tee and I got together for a planning meeting. Julie didn't show up. We waited 15 minutes. There is still no Rubicon. Tee sent her a message. I dialed her number. Nothing.

She missed a call from the Puma that afternoon.

Soon, our manager, Jane, locked herself in a meeting room with an HR member, examining a missing employee file.

I understood perfectly well what had happened.

She contemplated this peak on the chart, coming in two weeks and rapidly approaching. It is impossible to imagine what could have caused such a large number of Facebook users to call her name, but it is absolutely certain that they will call her - Enchilada's predictions, if they were clear, always came true. In her 13 months at PIG, she knew that sharp peaks meant significant disaster and shame … in contemplation of all this, Julie Rubicon did an absolutely sensible thing.

She ran away.

At first I was angry, mostly because she didn't ask me to run away with her. But soon my anger gave way to anxiety as I presented possible options for what dire reasons Julie’s name would be on everyone’s lips in two weeks. Abduction, plane crash, bomb explosion - my imagination painted all the terrible pictures. I could hardly sleep.

And all of a sudden, I got an idea how to fix it.

The peak of discussion happens when the winner is determined. When companies merge or go bankrupt. When politicians get into scandals. When people die.

And maybe when people tell the truth.

There she is.

The Affiliate Mind Group at Facebook used an app called Enchilada to scan and summarize all posts and comments on the system, public and non-public. She delivered the results of these scans to advertiser clients, not for malicious purposes, but simply for some person to tell their boss, look, we did something right.

In the fall of 2015, Julie Rubicon and I used the undocumented and unexplained feature of the Enchilada app to execute multiple trades in the US stock market, netting a net profit of $ 162.

I am sending this post to a few journalists and writers with the assumption that most of them will dismiss the story as a joke or fiction. But I know how Facebook works; I mean, I really know. If only a few of my addressees publish this, it will go viral.

And I will, if not forgiven, then at least be deprived of this burden.

And the dev team will fix the Enchilada app (Seriously Neil. Fix it).

And this message, all the various copies of it, condensed or retold versions, will begin to circulate in the system - this could be Julie's peak. Not a scandal. Not a disaster. Just a true story.

Julie, if you're reading this - on Facebook or repost elsewhere - I think it means you're safe. Hence, all the people who published this text actually worked together to revise the Enchilada prophecy. This means that if you want - and I understand, if not - you can go to our site on the dark web and use that mailbox to tell me where you are. My bag is complete.

It's all.

…………..

Although it might be worth writing her name a few more times. Each mention is counted separately.

This is for you, Enchilada:

Julie Rubicon.

Julie Rubicon!

JULIE RUBICON!