tech

How algorithms (secretly) run the world

13 Comments

The requested article has expired, and is no longer available. Any related articles, and user comments are shown below.

© 2017 AFP

©2019 GPlusMedia Inc.

13 Comments
Login to comment

I often feel like I live in one big algorithm designed several decades ago.

0 ( +3 / -3 )

Algorithms can't predict Black Swans (Taleb).

1 ( +2 / -1 )

Ultron, Skynet, the Matrix. Is it just around the corner, or is it already here?

3 ( +4 / -1 )

The biggest algorithmic product is in the White House.

-3 ( +2 / -5 )

To add insult to injury all these models have given a host of nerdy names, all coated with the incomprehensible jargon pencil chewing language of excessive meaningless. After all did the regression with quantitative and qualitative variables predict the next President of the USA would be Donald Trump, or Great Britain would leave the European Union, Nope not by the hair on my chinny chin chin did it.

http://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_reg_sect057.htm

To cap it all, Banking and Government institutions, climate change organizations are prepared to provide the whole caboodle with a ever growing mountain of cash, completely oblivious to what it all actually means.

2 ( +4 / -2 )

I agree with itsonlyrocknroll, and I'm also reminded of the movie INTERSTELLAR, where (SPOILER ALERT) Cooper asks the school principal about the waist and inseam size of his trousers, reminding the principal that if his mere trousers require two numbers to be properly measured, how can you measure a human with a single number?

0 ( +1 / -1 )

Algorithms..chat bots..very primitive AI.

Coming soon, or very nearly here:

Big Data..powerful software and machines..advanced programming..self-learning AI.

Trouble for Homo sapiens?

1 ( +1 / -0 )

Unfortunately We humans have become too reliant on technology which includes algorithms. GPS is one of the biggest issues. On many occasions the GPS in your car will take you to your destination using the longest possible route. If you just used old school techniques and asked someone on the street for directions you would get to your destination a lot faster.

1 ( +2 / -1 )

Anyone involved in regular trading on stock markets know how prevalent and pervasive algorithms are in manipulating share prices to large trading house's advantage. Retail investors are entirely at their mercy.

0 ( +0 / -0 )

I first heard 70 years ago - figures don't lie, but liars sure can figure. Seems like nothing has changed.

1 ( +1 / -0 )

To add insult to injury all these models have given a host of nerdy names, all coated with the incomprehensible jargon pencil chewing language of excessive meaningless.

They're not "nerdy names", and the fact that you don't understand the "incomprehensible" jargon of "excessive meaningless" doesn't actually make it incomprehensible or meaningless. They are all very descriptive and meaningful terms for someone actually in the machine learning field. If you're not in the field, of course it sounds meaningless because they're field-specific terms. You probably also don't know what a "quantum harmonic oscillator" or "1,3,7-Trimethylpurine-2,6-dione" is, but that doesn't make the terms meaningless.

After all did the regression with quantitative and qualitative variables predict the next President of the USA would be Donald Trump, or Great Britain would leave the European Union, Nope not by the hair on my chinny chin chin did it.

Of course not, that's a terrible application of machine learning as it's almost 100% based on human sentiment, and humans are unpredictable. Trump winning should be proof enough of that.

Not to mention that results can vary greatly depending on how both the data being fed into the algorithm and the algorithm itself are structured, as well as the type of problem that's trying to get solved. It isn't an exact science, there are a lot of variables that will affect the accuracy of any algorithm on any given data set, and for many problems it's simply the wrong tool to use.

The point is you can't use one terrible example, consider it representative of the entire field, and proceed to call it all meaningless just because you don't understand it.

1 ( +1 / -0 )

So much for good old fashioned hard work and dedicated research!

0 ( +0 / -0 )

Hi Otacon512, Risk analysts use a number of models, predicting different variant forecasts to hedge extreme market swings. These would be in the form of derivative trade automation. Colleagues used regression with quantitative and qualitative variables or the baseline theory, to predict how the UK population would turnout and model region by region patterns and comparisons. The Trump win?, and the extenuating circumstances? i.e the Putin effect? Maybe it all boils down to which newspaper one reads? or Believes?

Forecasting the Vote: A Theoretical Comparison of Election Markets

http://www.columbia.edu/~sk75/KouSobel.pdf

Snowberg Wolfers Zitzewitz economic forecasting-NBER

https://people.hss.caltech.edu/~snowberg/papers/Snowberg%20Wolfers%20Zitzewitz%20economic%20forecasting-NBER.pdf

0 ( +0 / -0 )

Login to leave a comment

Facebook users

Use your Facebook account to login or register with JapanToday. By doing so, you will also receive an email inviting you to receive our news alerts.

Facebook Connect

Login with your JapanToday account

User registration

Articles, Offers & Useful Resources

A mix of what's trending on our other sites