Historic reports against old policies versus current reports against current policies

[JimKillock]

Fiilter policies from ISPs occassionally change. @Mike-ORG is looking at the current ISP policies to see if there is anychange in advance of the new round of classifications by volunteers. However this raises a question or two about consistency over time of our blocking results.

When a category (for instance) is removed from blocking we may get the following kinds of scenario:

1. A report is filed and classified as "blocked according to ISP policy", for example in 2019 under the old policy
2. That report and classification is logged and reported in 2019 statistics
3. A new 2021 policy takes the the site out of those to be blocked, and the site is unblocked
4. The statistics may then reflect the current position, or change the 2019 results when the site is manually reclassified

Do we need to adjust our policies?

[dantheta]

The network descriptions (and policy statements) are stored in the CMS, which retains revision information. If we do edit them to bring them up to date, we still have the old versions stored (although I'm not sure yet if the older revisions can be fetched through the CMS api).

If we want to make the effective-from-to date range explicit, we could add a temporal range to the ISP policy records, and/or record versions that model a superceded-by relationship between policy records for the same ISP. A report review would be associated with the version of the policy description that was in effect at the time of the review.

A related issue is the storage of the URL's current block status. The system stores the latest result at the top level, and has historical test results stored separately (sometimes offline, where we have aged records off the system). For the example above, the original ISP report will be updated as "unblocked" when a block is no longer detected, effectively changing the 2019 results (if they were to be regenerated using the current data).

I think our best approach for the moment is to keep the 2019 results in their document frozen, and work on 2021 using the latest data and policies. If the review in 2021 updates some of the historical reviews so that they are no longer considered incorrect, we've still got good quality data for comparing 2019 and 2021 from their document forms (essentially snapshotting).

[Mike-ORG]

I've made a document that lists each ISP's current policy as well as the policy summary we currently list on Blocked.org.uk.
https://openrights.maadix.org/nextcloud/index.php/f/45134

I've noted which are up to date (most) and those we may want to amend. Sky was the only one that had major differences between our summary and the current policy.

[Mike-ORG]

Regarding Sky's filter policy, their website currently states:
http://storage.sky.com/prod/helpcentre/manuals/skybroadband/SkyBroadbandBuddy-CategoriesandPlatforms.pdf

PG

Adult, Blogs & Personal Sites, Chat & Forums, Dating, Explicit Content, Gambling, Government and Politics, Health, News, Online Games, Online Shopping, Social Media, VPN and Proxies.

13
Adult, Dating, Explicit Content, Gambling, VPN and Proxies

18
Explicit Content

The policy summary on Blocked.org.uk is:
Sky filters are on without users choosing them, but can be switched off. 40% of Sky customers have filters on as a result. The 13 rating blocks: Dating, Drugs and Criminal Skills, Anonymizers, Filesharing and Hacking, Pornography and Adult, Suicide and Self-Harm, Weapons, Violence, Gore and Hate, Cyber Bullying.

So the difference is:
Blocked.org.uk summary only lists the “13” rating, and that appears to have changed substantially since last update. The current 13 rating is shorter, no longer listing Drugs, Criminal Skills, Anonymizers, File sharing, Hacking and several other categories.

@JimKillock Is it possible that Suicide and Self-Harm, Weapons, Violence, Gore and Hate, Cyber Bullying are now encompassed by “explicit content”? Should we amend the policy summary?

Additionally, the following platforms are blocked currently on the 13 rating:

Twitter, Tinder, Skout, Bumble, Badoo, Zoosk

Should we list these platform blocks in the policy summary?

[JimKillock]

Hi Mike, I doubt that 'explicit' covers these but will check with Sky directly.

[Mike-ORG]

Thanks Jim.