Angry Blocker

Deep learning's answer to propaganda

Ad blockers block ads. The Angry Blocker blocks angry web pages. It is a Google Chrome Extension and Firefox Add-on. You can download, read why we created it, how it works, available websites, suggest a new website, support us, find out about our neural networks and view some media statistics.

Free Download

Angry Blocker is now available in the Chrome Webstore and Firefox add-ons website. Click the link and then "Add to Chrome" or "Add to Firefox".


Angry Blocker in action on an angry web page

Why Angry Blocker?

In this post-truth modern world with endless alternative facts it's very easy to get over emotional, form false judgements, and lose control of our decisions. When we allow emotions to blindly force us to do something we become vulnerable to propaganda. In 2016, both ends of the political spectrum used emotionally charged headlines to influence public opinion on Brexit and Donald Trump. Such news is dangerous when it leads us to make choices we wouldn't have made in our normal emotional state.

Angry Blocker is our first attempt to address this issue. Angry Blocker protects you from negative emotions and helps you to be aware of the type of information you consume online. When you visit selected websites like CNN, FoxNews, BBC, etc. Angry Blocker's proprietary deep neural network estimates the emotional content of the page which you are about to read and notifies you about very sad and angry pages. Then it's up to you to decide whether to read those pages or not. But once notified you are aware and therefore protected from misleading overly emotional reactions and invasive propaganda.

Just like a healthy food diet, we believe that internet users should also follow a healthy information diet. Angry and sad content should be consumed with precaution. Too much sugar and fat can lead to health problems, but at present we've found ourselves in the world with too much anger and hate. Angry Blocker will help to restore a balance, because the world is still a good place.

How It Works

Whenever you visit an angry or sad news page, we will hide the page from you. Instead you will see a nice picture from nature and a message like "This page is too angry for you" with links to "Go back", "See it anyway" or view some better news on our website. Like this:

Angry Blocker showing mountains Angry Blocker showing a beach

You will also observe the icon in the top right corner of your browser changing from a question mark question mark icon to the predominant emotion expressed in the page such as love icon for angry. You can click the icon to view the full analysis, showing the percent of each emotion:

Angry Blocker graph showing emotions

There is also an options page which allows you to specify how much anger you want to be protected from:

Angry Blocker options

Available Websites

Angry Blocker only works on selected media websites: abcnews.go.com, aljazeera.com, bbc.co.uk, bbc.com, bloomberg.com, cbs.com, cbsnews.com, cnn.com, dailycaller.com, dailymail.co.uk, economist.com, eonline.com, espn.com, facebook.com, foxnews.com, ft.com, guardian.co.uk, huffingtonpost.com, independent.co.uk, latimes.com, mikebloomberg.com, nbcnews.com, newsmax.com, newyorker.com, npr.org, nytimes.com, pbs.org, politico.com, rt.com, reuters.com, news.sky.com, skynews.com.au, telegraph.co.uk, theatlantic.com, thedailybeast.com, thesun.co.uk, thetimes.co.uk, time.com, usatoday.com, washingtonpost.com, wnd.com, wsj.com, yahoo.com.

Suggest a website

If you think there is a particularly angry website which needs blocking, please tell us about it:

Your name:

Your email address:

Website to block:

Comments:

Support Angry Blocker

The best way you can support is to tell your friends about Angry Blocker. Because if the idea behind it becomes popular enough, it could impact upon the way articles are written. You can also send us feedback on what you like and what could be improved.

Deep Neural Network

Behind the scenes, the Angry Blocker extension extracts meta tags from the web page you are viewing and sends it to our server. We use our emotional model, a multi-layer recurrent neural network, to analyse the text and estimate the percentages of the following emotions: love, haha, wow, sadness, anger, the same five emotions which Facebook currently uses. It does this by first turning words into vectors and then running a multi-layer recurrent neural network.

Statistics

So far we have analysed the emotions in 1205586 web pages. This is a lot of data and we can do some interesting analysis. For example, you can use this table below to see which media companies produce the most sad articles according to our emotional estimates:
CompanyCountryRegionNum articles% love% haha% wow% sad% angryEmotion
AccuWeather12718338373shock
aljazeeraQatar162822014153120sadness
chroncom621291029259delight
deutschewellenewsGermany1642118202119neutral
bbcnewsUK193042619192114compassion
skynewsUK160422119202119neutral
CBSNewsUSA314382317192021neutral
ABCNewsUSA306402814201919love
NBCNewsUSA231702418191821polarity
nytimesUSANew York282173315161719polarity

Contact Us | Share: