Dynamic filtering examples

Dynamic filtering can be used to block much more aggressively than what would normally happen when relying only on the default filter lists. With dynamic filtering, web pages are definitely more likely to break, but for many users this is acceptable, so long as the content of a web page can still be read.

Following are some examples of using dynamic filtering vs. not using dynamic filtering (i.e. relying solely on the static filter lists), with both scenarios using the default filter lists. The top row in each table shows the used bandwidth.

I didn't report below the comparative results without a blocker, that would be a lot of noise detracting from the main topic here, but I provide a summary of what would have happened without µBlock with default filter lists. (That is with click-to-play enabled for plugins -- it would be much worst without this.)

I used my HAR parser tool to extract the results.

Example 1 -- An article on TechCrunch

URL: http://techcrunch.com/2014/10/07/yahoo-lays-off-employees-in-india-reportedly-up-to-2000-affected/

The article could be read all fine with dynamic filtering. For many users it's often only what matters for most sites.

Without µBlock enabled at all, 61 different hostnames were hit, with the consumed bandwidth at 2,627,068 bytes.

No dynamic filtering	Dynamic filtering 3rd-party <script> 3rd-party <iframe>
1,144,039	418,578
0.gravatar.com
b.grvcdn.com
i.api.grvcdn.com
o.aolcdn.com
pthumbnails.5min.com
r-login.wordpress.com
rma-api.gravity.com
s0.wp.com	s0.wp.com
s2.wp.com	s2.wp.com
s.aolcdn.com
tctechcrunch2011.files. wordpress.com	tctechcrunch2011.files. wordpress.com
techcrunch.com	techcrunch.com
zor.livefyre.com

Example 2 -- An article on The New Yorker

URL: http://www.newyorker.com/magazine/2013/03/11/up-all-night-2?currentPage=all

The article could be read all fine with dynamic filtering.

Without µBlock enabled at all, 24 different hostnames were hit.

No dynamic filtering	Dynamic filtering 3rd-party <script> 3rd-party <iframe>
1,346,535	487,095
ajax.googleapis.com
api.parsely.com
assets.adobedtm.com
dff7tx5c2qbxc.cloudfront.net
plugin.mediavoice.com
subscribe.condenet.com	subscribe.condenet.com
use.typekit.net
www.googletagservices.com
www.newyorker.com	www.newyorker.com

Example 3 -- An article on Bloomberg

URL: http://www.bloomberg.com/news/2014-10-03/yahoo-said-close-to-investing-in-snapchat-at-10b-value.html

The article could be read all fine with dynamic filtering, though an interactive widget to display a graph of Yahoo's stock price over time didn't display at all.

Without µBlock enabled at all, 48 different hostnames were hit, with bandwidth consumption at 2,095,142 bytes.

No dynamic filtering	Dynamic filtering 3rd-party <script> 3rd-party <iframe>
1,615,283	141,590
a.disquscdn.com
bloomberg.disqus.com
cdn.gotraffic.net	cdn.gotraffic.net
cdn.taboola.com
disqus.com
fonts.gotraffic.net
images.taboola.com
location.bloomberg.com
login.bloomberg.com
netstorage.taboola.com
personalization.bloomberg.com
q.bloomberg.com
trc.taboola.com
www.bloomberg.com	www.bloomberg.com

Example 4 -- An article on Live Mint

URL: http://www.livemint.com/Industry/TM8tDvrv3OfeYjeXkEPXZI/Flipkart-hits-100-million-sales-target-in-10-hours.html

The article could be read all fine with dynamic filtering, though commenting through Disqus was not possible. Of course, if ever you care commenting, turning off whatever dynamic filters for a specific site is as simple as point-and-click.

Without µBlock enabled at all, 25 different hostnames were hit, with bandwidth consumption at 1,276,894 bytes.

No dynamic filtering	Dynamic filtering 3rd-party <script> 3rd-party <iframe>
736,197	281,111
a.disquscdn.com
cdn.taboola.com
disqus.com
h.ppjol.com
images.taboola.com
livemint07.disqus.com
netstorage.taboola.com
s.ppjol.net
trc.taboola.com
www.livemint.com	www.livemint.com

Example 5 -- An article on SecureList

URL: http://securelist.com/blog/research/66988/tyupkin-manipulating-atm-machines-with-malware/

The article could be read all fine with dynamic filtering. An embedded Youtube video was reduced to an hyperlink (which is a good thing IMO).

Without µBlock enabled at all, 18 different hostnames were hit, with bandwidth consumption at 1,969,789 bytes.

No dynamic filtering	Dynamic filtering 3rd-party <script> 3rd-party <iframe>
1,657,834	1,131,906
25zbkz3k00wn2tp5092n6di7b5k. wpengine.netdna-cdn.com	25zbkz3k00wn2tp5092n6di7b5k. wpengine.netdna-cdn.com
i.ytimg.com
kasperskycontenthub.com
s0.wp.com
securelist.com	securelist.com
s.ytimg.com
www.google.com
www.youtube.com

Example 6 -- An article on Wired

URL: http://www.wired.com/2014/10/feds-silk-road-hack-legal/

The article could be read all fine with dynamic filtering. The Disqus comments were not loaded though -- a nice side-effect IMO.

Without µBlock enabled at all, 46 different hostnames were hit, with bandwidth consumption at 3,459,474 bytes.

No dynamic filtering	Dynamic filtering 3rd-party <script> 3rd-party <iframe>
2,643,977	1,520,884
a.disquscdn.com
disqus.com
fonts.condenast.com
html1-f.scribdassets.com
html2-f.scribdassets.com
images.outbrain.com
odb.outbrain.com
plugin.mediavoice.com
s1-f.scribdassets.com
s2-f.scribdassets.com
widgets.outbrain.com
wired.disqus.com
www.scribd.com
www.wired.com	www.wired.com