The Last Serenade
Quarterly (and final) report of Operação Serenata de Amor
There is a Portuguese version of this article in here.
September 7th, 2016 — it’s the day we celebrate the independence of Brazil. But this time it was also the day we presented Operação Serenata de Amor to the world, a good timing for what we hoped to do: enhance social control of the public affairs with technology, throwing data science and artificial intelligence into unexplored seas.
This post is our last regular report from Operação Serenata de Amor, and we want to use this space to tell you everything we have achieved so far.
Today we are closing a cycle that started on September 7th when we launched to the world the crowd funding campaign for our independence: a way of stating we can cause impact from here, outside the government.
“A lot of ‘civic data science’ initiatives are gone to early, or are left behind before seeing the sunlight. In spite of that, #OperaçãoSerenataDeAmor managed to gather a serious and committed team, with disposition and skills to keep the project running; so far the results are encouraging.” — Thiago Marzagão, from the Ministry of Transparency, Supervision and Control
THE BEGINNING
It’s difficult to single out just one reason behind Serenata de Amor’s birth. In tune with its heavy weight ideal, a large cluster of facts have to be put together in order to understand the creation of such a project.
Our country was going through an extremely turbulent political moment, where the rule was to complain. “Politicians steal, government is broken, there is no light at the end of the tunnel.” And we were tired of that.
We used our project as a flag to call the attention to the other side of the coin. Serenata de Amor was made by the rest of us, citizens who were not satisfied with merely registering our dissatisfactions in a massive board — a board already full of complains from people.
We have never stood for a witch-hunt. We claim it is possible to cause impact without pointing fingers, without using our findings to create more complaints. Serenata de Amor means citizenship change by citizens.
But how to make it happen? We knew we could count on the support of those who would benefit from all that: you — that is why we opted for a crowd funding campaign.
Serenata de Amor was different from day one. We understand people and technology. The sum of these skills enabled us to make the most successful technology crowd funded project in Brazil.
Even in a country with no tradition of investing in science and technology, we got it, thanks to 1,296 incredible people who pledged on our idea of a robots supervising the govenrment. R$ 61,280.00 was our target and we gathered R$ 80,424.00 — more than 25,000 US dollars. We needed money to work full time for two months, and we enough for three.
WHAT WAS THE OPERAÇÃO SERENATA DE AMOR ABOUT
According to the Federation of Industries of the State of São Paulo, corruption diverts R$ 69 billion per year in Brazil. According to the head of Operation Car Wash, this number is R$ 200 billion. This is your money being wasted — and we want to put an end in this binging on public money.
Did you know that every Brazilian congressperson, in addition to their salary and various benefits such as accomodation benefits, gets up to R$ 45 thousand every month to use with personal expenses — theoretically, to assist their parliamentary work?
By law, this money is available for the politician and his cabinet in everything that does not require a bid, such as a meal or a taxi ride. Just last year on average each congressperson spent R$ 266,000.00. Summing it all it’s more than R$ 213 million since 2009. More than R$ 213 million of our money.
This amount is called Quota to the Exercise of Parliamentary Activity (CEAP for the acronym in Portuguese) and its expenses were publicly disclosed by law.
Operação Serenata de Amor has created robots capable of auditing government data, learning from it and identifying what is a legitimate or suspicious expense for whatever reason. Given a supicion of irregularity found by these algorithms we formalized reports to the Chamber of Deputies, our Lower House, requesting the return of the improperly used money to the public treasury.
OUR RESULTS SO FAR
In a week reporting cases, using all the tools we have built:
These numbers mean a return on investment of 371% when compared to the amount we have collected.
“This project is a tangible example of how much this new generation can do for our country. I hope it ends up as one of the many initiatives that inspire more and more young citizens to become entrepreunerus and to put their talents to transform Brazil!” — Ney da Nóbrega Ribas, President of Observatório Social do Brasil.
WHO BUILT THAT
We were not the best ones out there to do that job. We were not the best data scientists, statisticians, mathematicians, and engineers. But we did the job.
The crowdfunding allowed 9 different people to work at Serenata. Talented people who are programmers, journalists, scientists and sociologists, all with a set of unique soft skills. These are the main responsible for Operação Serenata de Amor:
Irio Musskopf
Technical Leadership and Data Science
Eduardo Cuducos
Software Engineering
Ana Schwendler
Data Science
Felipe Cabral
Administrative and mastery
Bruno Pazzim
Marketing, communication and software engineering
Jessica Temporal
Data Science
Pedro “Tonnynho” Vilanova
Communication and journalism
Filipe Linhares
Data Collection
André Pinho
Data Collection
OUR TRANSPARENCY
From the beginning, all our work was open. From the code we wrote to spreadsheets telling everyone about the origin and destnation of every penny we have raised through crowd funding. Every other week we published reports on our progress.
Keeping the project open source made us strong from day one. We have received several contributions from all around the world. Besides everything else it represented for us these contributions show that Serenata de Amor is a living proof that a technological project with a ideal behind it can gather an incredible group of people willing to help.
“Initiatives such as Serenata, initiatives in which population uses the transparecny of the Chamber of Deputies, are the greatest institutional protection and valuation we could have. The population taking the House to themselves, supervising, taking into account… This what the government employees work for” — Fabrício Rocha, Chamber of Deputies
HOW WE DID IT
To reach that we built different tools that helped us from the moment we analyzed an invoice from a congressperson, to the moment in which we were actually reporting a irregularity to the Chamber of Deputies.
1. JARBAS
Jarbas is our interface to browser the expenses of our deputies. We have collected data from several places to enrich the data provided by the government. We have got, for example, the registration of companies in which our congresspeople spent our money — we’ve got it from the Department of Federal Revenue of Brazil. Data like that enabled us to know more about any given expense. We included a photo of the place brought directly from Google Street View.
2. HYPOTHESIS
Like any data science project everything we want to teach to our Robots begins with a hypothesis validated by a human. Each of the hypothesis we have worked on were probable illegal cases to use CEAP money. We worked on the following hypothesis:
- Suspicious distance traveled
- Sub-quota limits
- Reimbursement of alcoholic beverages
- Suspicious meal prices
3. ROSIE
Rosie is the robot who processes each one of the more than 2 millions invoices available from the Chamber of Deputies. She gives us the level of suspicion of the expense and a probability together with the reason why she believes it is suspicious.
Rosie’s learned each of the hypothesis we developed in the previous step, so she can easily identify the purchase of an alcoholic beverage, for example.
She’s a smart Robot. That means she can learn new parameters and adjust existing ones on her own with Machine Learning techniques.
4. CREATING TOOLS FOR AUDITING INVOICES
We created different auditing tools that helped us to deliver to the Chamber of Deputies more than 900 reported cases in just one week.
You can read more about the auditing tools here.
5. MARATHON OF REPORTS
In one of the last weeks of the project, the whole team was focused on one thing: make beautifuly report strong cases to the Chamber of Deputies.
Rosie has cherry-picked thousands of suspicious cases for us, then we manually analyzed each one of them to report 629 cases to the Chamber of Deputies, a total of 216 different congresspeople, involving R$ 378,844.05.
You can read more about the results of this joint effort here.
6. THE END OF THE FIRST PHASE OF OPERAÇÃO SERENATA DE AMOR
This is where we are, this post marks the end of the first phase of the project.
We hope we could make our supporters proud and we hope we could made the case for the future of social control of public expenses in our country — that is to say, having cutting edge technology used alongside people, not only within the government. Open data is there for this purpose. When one opens it to the public, one gets amazing return.
SERENATA DE AMOR ON THE NEWS
“A very young group of Brazilian citizens” — William Bonner, Jornal Nacional
PUBLICATIONS
Over the course of the last three months we have produced dozens of posts about challenges, plans, and the daily taks of the project. This report is also the latest of a series that described our progress.
You can read all the other posts related to the project here.
A DATA SCIENCE PROJECT IN THREE MONTHS
Can you believe it? In three months we succesfully achieved all those amazing things. With a pride we can say Operação Serenata de Amor today has become a worldwide reference for both the use of technology for social control and the use of open government data in this country.
We have shown how we can deliver value in a Data Science project in no time, and mainly how we can creatively use artificial intelligence to solve a common problem of our country — even when this is the biggest problem of all.
Yet the open-source nature of the project made it possible to initiate new people in Data Science — something very recent in our country, contributing directly to the national development.
FUTURE
We are pretty sure we can go further, and we are pondering on different ways to keep the project running after this first step. The end of this phase is also the end of the funding we have got through Catarse. Anyway, the code is free.
We just launched a new crowd funding campaign to ensure the continuity of the project. Once more we depend on people like you, people who want to see the this project go further. Click here and support us at APOIA.se.
Click here to make Serenata go further
An alternative way to move forward with the project is through companies and foundations investment. If you know any organizations that might be interested in support us, we can inroduce ourselves, our ideas and what we have already done and also what they get in return by investing in social control of public administration in an efficient way. Talk to us.
We are also accepting Bitcoin donations at 1Gbvfjmjvur7qwbwNFdPSNDgx66KSdVB5b.
THE END
It is impossible to finish this first phase without a huge thank you to each one of the 1,296 people who believed in us when we had nothing — when we had only a proof of concept and a video talking about the project. If it weren’t for them, none of what we just wrote about would have happened. We really wanted to be able to hug each one of our supporters. If we ever meet us in person feel free to claim this hug.
For the team of nine: it was an enormous pleasure to be part of this during these three months and be able to share with all of you every step of the way. It was an incredible experience that left a permanent mark on our lives, and that made our hope shine brighter than ever. I’m gonna miss our 9am — sharp!— daily team meetings sharp.
Serenata de Amor Team❤️ 👊
This was a triumph. I’m making a note here: huge success.
