Clinton’s data-driven campaign relied heavily on an algorithm named Ada. What didn’t she see?

Washington Post ^ | Nov 9, 2016 | John Wagner

[HarleyLady27]

Inside Hillary Clinton's campaign, she was known as Ada. Like the candidate herself, she had a penchant for secrecy and a private server. As blame gets parceled out Wednesday for the Democrat's stunning loss to Republican President-elect Donald Trump, Ada is likely to get a lot of second-guessing.

[Trillian]

Hehe..

[b4me]

that could be. my first thought was she didn’t see the people of the area or the country worthy of the fireworks whether she won or lost!

[RitaOK]

I can’t seem to sign up for the Washington Post. I wanted to read this column. Bill Clinton had been warning them for some time. They were in denial.

[raisetheroof]

ADA was a computer programming language that was developed back in the late 1970s. Could that be what they were using?

[Billthedrill]

Ada is a pretty mainstream programming language, and one used in government.

Yep. Early 80's stuff, a Pascal variant IIRC. That was running through my mind when I read this and I checked the credentials of the author - political reporter. Whaddaya know.

By the way, one thing Ada Lovelace was working on with Charles Babbage was a program to predict horse races. It failed. The irony abounds.

[onyx]

Where's Ada?

[CommerceComet]

I wonder did Ada get it wrong or did some in the campaign know but not tell. They cancelled the fireworks on Monday, what other reason would cause that?

She also seemed to concede pretty early. I thought it odd. I wouldn't have faulted a candidate in that situation to wait some more and hope for some improving returns, even if the odds were pretty long.

[AppyPappy]

If she had even suspected she might lose, they would have ramped up the voter fraud.
The whole world was shocked when Trump won. Even Trump looked shocked.

[Myrddin]

I've worked on a project since 1992 that has a mix of Ada, C, C++. FORTRAN IV, PL/SQL, Java 8, JavaScript, make/maven/gradle build tools. We've phased out all but one 3rd party FORTRAN routine. The Ada has been replaced with Java. The C++/C code has been rock solid and has only required a port from Forte C++ on SPARC to GNU g++ on Linux to keep up with the technology platforms. The Ada compilers were always a problem. We used enough of the language to expose all the dirty little defects and paid for ongoing maintenance to get the compiler defects repaired. Our DoD customer directed us to replace Ada with C++ on projects that fell far behind due to the defective Ada compilers. We're done. C++ and Java 8 are what we use going forward.

[Crolis]

I think if I were to design an analytics program to perform similar functions, the first rule or metric I would establish is (1) take no voter for granted and (2) generate a metric about how loyal a voter type is.

That way if I saw support softening in a particular area, I could quickly respond.

This all of course is predicated on good data going into the system.

[FreeAtlanta]

They came much too close in Florida.

That would have made it a very long night.

[kingu]

That is exactly what the USC/Times polling was doing, measuring loyalty and enthusiasm. And when the pollster came on the local news the morning after, he said - ‘Hey, I’m not surprised at all, it was as our polling showed, if anything, we slightly undercounted some areas which we’re looking at to fix our next polling.’

[az_gila]

If I were a campaign manager, I’d be seeking out the NY Times election night predictor and the USC Dornan polling methods. Both, when given accurate data nailed the results.

---

Yes, but only after the votes started being counted. It's easier then.

The big laugh was the NYT on-line predictor that started of with really biased predictions (90+%) for the witch...

It was 730pm in AZ when the tables turned on the predictor.

[kingu]

GIGO - The NY Times plugged in their flawed polling data into the system (and SLIGHTLY adjusted them from some of the exit poll data), so early predictions were skewed with their wishcasting. Once real TRUE data went into the system, it quickly switched.

[JediJones]

[ViLaLuz]

Maybe that’s why She canceled her fireworks show.

[Taxman]

Don’t try to teach a pig to sing;
It wastes your time and annoys the pig!

[Taxman]

LOL!

That is good!

[uncitizen]

Bet you've been waiting years to use that one!😜

[x]

Maybe they used the algorithm with what they thought were swing states and didn't bother with states they thought were certain to go for one candidate or the other.

It would be a waste to run a complicated computer program to figure out who'd carry Massachusetts or California. Maybe they felt the same way about Wisconsin and Michigan.

That doesn't explain Pennsylvania or North Carolina, though. Maybe they were worried about those states, or maybe the algorithm was a total bust.