Yet another suggestion for filtering objectionable content

Hi all,

This is about the much discussed (and often fiery) topic of how to make sure the users don’t associate the network mainly with objectionable content. My suggestion is to look at how Reddit Masstagger manages to deal with a lot of content flagging while keeping the user in control, but with a few additional features.


URI hash lists

A collection of hashes of URIs for use in filters.


URI hash list subscription

A process that allows users to subscribe or unsubscribe from automatically checking a public hash list for updates. The URI hash lists from these subscriptions would then be copied to the user’s computer and assigned an action based on the user’s preferences for that subscription.


What would happen

When the browser views a site, it would hash any URIs present and check them against the user’s list of hashes. If any matches are found, the browser then looks up the hash’s associated rule and takes action. This rule could be anything from displaying a green tick (to indicate trustworthiness) to completely disabling and blacking out the link.


Snuff example

Say the UK government wants to stop its citizens from viewing snuff. They accept that technology has advanced to the point where it’s impossible to enforce a law against it, but they still have an obligation to try.

They store as much snuff as they can on a URI hash list, then make this list publicly available. As they are storing hashed links rather than direct links, it can’t be said that they are promoting snuff.

Every computer sold with Safe preinstalled in the UK must now by law be subscribed to this hash list. The subscription must be configured such that every link produces an error message, and every embedded video is censored.

This provides the government with a satisfactory level of parental control, while still letting the user alter the subscription if they wish. Users from other countries could also opt in to this list to protect themselves from viewing snuff. Everyone wins.


:​)
Thoughts? I could write more hypothetical use cases? I haven’t thought this out in detail, but since Masstagger is so good, I felt it worth posting.

4 Likes

I moved this topic to Apps because censorship is against one of the foundations of the SAFE network. So it would not be implemented at the SAFE network level. It would not be a feature of the SAFEnetwork.

As you summed up here a government would have the ability to censor people yet again if made a feature of the network that people cannot bypass. And your example would see the government include any anti-government pages in their hash list. So we are actually back to a worse censorship model than the current internet.

Now at the APP level, say a browser plugin, then people can take control of their own browsing experience and be truly free to choose between totally uncensored network or to choose their desired filtered experience.

You know the thing about really bad stuff is that even with the current internet you dont’t typically see it without looking for stuff. I am sure your example exists on the web but in 25 years of web surfing I have never seen one. And that applies to a lot more examples.

Our government years ago wanted to censor our internet at the ISP level “to protect the children” and was defeated at every turn of their argument that their only retort was to call anyone against the filter a child abuser or a supporter of such. Every argument they put up, like yours, to mandate filtering was shown to be outdated or impossible to implement or anti-social or unworkable or just playing wack-a-mole or a number of other things.

It simply does not work when implemented country wide or network wide and as such would never be implemented on the SAFEnetwork as part of the network. But the discussions did show their was a lot of merit for those who wished for a filtered experience to implement the filtering themselves.

5 Likes

As in, a feature to be implemented in the browser. It seems like an essential feature, but if the browser is designed to be minimalist then a plugin bundled by default.

It seems like you’re suggesting that I support government censorship. The user would be able to control the filter, and replace the error messages with warnings, or remove the lists completely.

You know the thing about really bad stuff is that even with the current internet you dont’t typically see it without looking for stuff. I am sure your example exists on the web but in 25 years of web surfing I have never seen one. And that applies to a lot more examples.

  • Without censorship, the Safe network will enable and attract content more objectionable than the clearnet, just like any darknet does. That’s unavoidable of course.
  • Without search engines with advanced filtering ready-to-go on network launch, it will be easier to stumble onto objectionable content. Imagine DuckDuckGo with SafeSearch turned off, but supercharged by the darknet factor.

But the discussions did show their was a lot of merit for those who wished for a filtered experience to implement the filtering themselves.

Without a standard method of filtering, this will be too complicated. What if a government is dissatisfied with the lack of supported options available and chooses to implement their own method? What if a user wants to implement a filter but it gets too complicated for them?

I think you will find in this forum a few places where there has been support for a browser addon to do filtering. It is really common sense for this to be done and we have example addons for other browsers where filtering is done according to lists that users can change. ADBlock is one such example. It could easily become the basis for a content filtering addon with lists maintained by those people who love to create lists of such things.

BTW Sorry if my post implied you were in favour of censorship, I was really just trying to use the example and refer to it.

If Australia is an example they will simply fail in this attempt. Now for more controlling governments will they will simply not be able to do it apart from the Chinese model of police on the streets stopping citizens and checking their phones have the government APP installed and running.

The Addon would be a standard way and the government could mandate it is installed and used. My objection was that it become a feature of the network itself. As a feature/addon to the browser then this will happen if history is a good indicator of human behaviour. It will help to filter out the adverts and other malware/pishing sites (as they are found) on the network.

I think you will find in this forum a few places where there has been support for a browser addon to do filtering.

As a feature/addon to the browser then this will happen if history is a good indicator of human behaviour.

My suggestion is for how this could work. As far as I know, no-one’s suggested URI hash lists or fully customisable list subscriptions. If there’s a better place to post the ideas then I’ll repost there.

I suspect the functionality should be built into the browser instead of an addon to increase performance. I picture it working something like this:

Instead of following a parsed link, the browser turns it into a 256-bit hash. The first 20 or so bits are used as a memory address to access a chunk of information generated by the list subscriptions to check that there are no filters associated with these first 20 bits, and assuming there aren’t, loads the link properly. If a possible filter is detected, then the full hash is looked up in the user’s subscription list and any filters are applied.

1 Like

I would imagine that to build it into the browser would turn people off simply because they cannot be sure that it is not filtering things they do not know. If you feel that it needs to then one needs to fork the safe browser and build it in - thus an APP.

With a addon then they can disable the addon to see if they are missing anything.

Also the fundamentals of the safenetwork is to have it censor resistance and so I doubt Maidsafe will even consider it being built into the browser. And for good reason as I said above, it can have hidden censorship as well.

But as an addon then it is fine. Addons give more flexibility in that different ones can be built by different people and each can still use the publicly available hash lists and people can have a choice of which filter addon best suits them. Once you make a one size fits all filter then you risk later revisions making it slow.

I don’t know if any of the adblockers out there use hash lists but i suspect they are or should be. They load up the text into a hash list when starting I suspect. Hash lists was one way of saving space in memory in the old days :slight_smile: you know when we walked for 20 miles in the snow to get to school :joy:

Can I make a suggestion for the hash list that there is a text version directly related to the hash list and can be simply verified. Not all people will want to blindly trust a hash list that can easily have other URIs being filtered out that is not obvious. The list would also need to be broken up into categories so that the user can select what types of material they want filtered.

3 Likes

Can I make a suggestion for the hash list that there is a text version directly related to the hash list and can be simply verified.

This should be easy enough to do. The browser (addon) would have a function to convert URI lists into hash lists, so it would just require the list-maker to release the URI list as well. But one of the purposes of making it work with hash lists is to allow ‘closed-source’ lists where the source list can’t be released for whatever reason. As the list alone can’t restrict the user or hide any information without the user letting it, I don’t see that as an issue when necessary.

The list would also need to be broken up into categories so that the user can select what types of material they want filtered.

I’m thinking the hash list subscriber would handle that. So, for example, the government might want to block both malware sites and copyrighted music, but they might want to block malware sites moreso. The user wants to listen to copyrighted music, so they open the hash list subscriber and disable the pre-configured copyrighted music list. Instead of disabling the malware sites filter, they might just cautiously change it to show a warning instead of refusing to let the site load.

I personally think that a filter like this would encourage people to break censorship, because it would be so easy for the user to neuter a hash list.

I distinctly remember walking uphill both ways too. :wink:
EDIT: And don’t forget the Wolves! The horror. The horror.

1 Like

I would subscribe to a voluntary list, I’m a pi-hole-er myself. If there were an add-on, an open review list, hashed. I think it’s a good idea.

I would like a choice of lists.

I would subscribe to a lot of lists too I think. I wouldn’t personally have any set to block the links, just to give warnings — link blocking would only really be for children using their parents’ computers, library/school computers, and a box-ticking preset that governments can mandate for preinstalls of browsers.

I’m expecting the advert and malware lists to be made which I’d likely subscribe to the better ones.

I’m guessing that once we reach a point where the pre-releases can’t be killed (reach a certain level of autonomy), then a demo of a filtering system will have been implemented to avoid bad press. Could anyone link me to a relevant discussion about implementations?

Would there be any reason not to subscribe to all of them?

Well if there are many malware lists then why subscribe to all, just subscribe to the one that is both more reliable (no/less false positives) and comprehensive.

The existence of malware lists are a huge mistake. It is the best way to advertise harmful sites that would otherwise be unknown to the vast majority of users.

In a system like Safe the correct way is to use reputation systems where each user can choose the level of reputation required for the content to be visible.

filtering out/blocking system

filtering of search results.

They both have their purposes. The filtering blocking system, if used, stops another site forwarding the browser to a malware site. Many sites use forwarding when you land at a page that has been retired and the site owners forward the browser to the new page/site. Now if a site forwards you to a malware site then no amount of reputation systems that hide the links will stop the browser going to that site, but a filtering blocking system will stop you. Also it will stop a person from occidentally clicking on a link which would otherwise take them to the malware site.

As to giving the person a list of sites, then that is really censorship at its base level. Oh no we can let people know the names of bad sites, they might go to them, which is hogwash. If they want to trawl through a 10K to 1 million long list then they would more likely just do a search for those sites.

Remember malware sites are those that GIVE YOU malware on your machine. So if someone wants to use a list to go to them, then they are just stupid or highly uneducated aren’t they

2 Likes

While I certainly disagree on the first half, about the existence of malware lists, I do agree that the way to go is personalized reputation based whitelisting.

Users would pick trustworthy sources of reputation (I can see a business opportunity there if “pay the producer” gets implemented) and aggregate them into their own personalized score for the documents on the network. If they would publish these scores, they could also act as a reputation source for others.

1 Like

Because you might be able to get a more complete list from multiple sources. If you get too many false positives from a list, then you could just unsubscribe from that particular list.

Why does it matter if a malware site is known? For extreme disturbing content this makes sense, but not malware.

Anyway, that’s the point of the hash list. You could choose to close-source any list so that you can’t look up sites from it.

I also think a reputation-based system based on users that you trust and users that they trust and so on will be very important in the network, for preventing spam for example.

But this is very off-topic. I’m suggesting what I think is a good implementation of a filtering system that serves different purposes, and a reputation system surely belongs in another topic.