Fallacies plugin development

[poblish]

Re: http://www.facebook.com/pauliewaulie/posts/189564251172698

Here's the initial plan:

1. Roll into BannedList plugin first (can refactor later, though there's a lot of common ground, and I think we have the same audience).
2. Use right-click, select, submit - just like current BL plugin (try it out if you haven't already!) - and store as new GAE entity.
3. Store URL key just like 2009 plan (Andrew: find that sheet!).
4. Entries initially stored "unapproved" and invisible. At first, approval will be a completely manual job by "one of us". (We should then try to send an email to user, but probably no big deal if that fails.)
5. Getting results for page: would need to do asynchronous lookup - in just the same way as BL plugin - to get all approved keys for page.
6. Challenge, then, is how to mark up the page in a clean way. Will have to be supremely subtle (someone: design mockups!). Coloured backgrounds probably won't cut it.
7. We can provide stats in due course: Weekly Top 10; All-time by writer (using http://journalisted.com/ info).

Questions:

• Get sponsorship or serious interest before doing any work!
• Use http://yourlogicalfallacyis.com/ 's icons?
• Team up with them?
• BL plugin icon already shows the BL score. Can we show Fallacies count here too?

[poblish]

URL key save:

1. Get user selection as plain text string [across multiple text nodes, so strip and normalise]
2. Constrain to 2000 chars
3. Save

URL key load (approach A):

1. Need access to plain text representation of DOM, stripped and normalised.
2. Search for loaded input... (how? Node link is broken!)
3. Need to get Node of match start (NA), and Node of match end (NZ).
4. Use existing BL highlighting code to highlight matching sections in NA and NZ.
5. Identify all text nodes between NA and NZ and highlight all.

URL key load (approach B):

1. Load input.
2. Iterate all text nodes, searching for any text match (stripped, normalised) with the start of the input.
3. If a match, so far so good. Store matched section and Node.
4. While we have unmatched input characters, continue to (5), else clear matched Node list and go to (6)
5. Move to next visible text node and check next bit of input matches the start of the node's (stripped, normalised) text. If so, store matched section and Node.
6. If all text matched, highlight all matched Node sections.
7. Else, log error on server and abort.

[pauliewaulie]

Here's how I'd suggest we do it.

1. Two levels of usage to start with - 'contributors' and users. Anyone can see the fallacies that users highlight.
2. As a contributor, I could look at a page, highlight and rightclick on text and tick a 'fallacy' button that indicates what the fallacy concerned is.
3. As a user, I can look at a page that has been reviewed by a contributor and see the fallacies highlighted by a contributor.
4. We demo this by finding a bunch of volunteer reviewers and finding a particular high-profile bit of text (a speech, a controversial article etc) and 'fisking it'
5. Rolling it out, we could build an online 'exam' that people have to pass to prove they know the fallacies - then when theyre' through, they get contributor access

All a bit airy-fairy for now, but wdyt?

[poblish]

Fine with almost all of that, except for (5)... I'm not sure anyone I know is qualified to write an up-front test for the fallacies, though I bet they could weed out the vast majority of bogus submissions as an administrator.

Also, if the test were any good, it'd be a pretty big hurdle for users to get over - we might end up with 5 users submitting good stuff, rather than 100 users submitting OK stuff. If the test were too easy we might end up with a lot of rubbish or spam. Either way, we'd probably have to add a user management system with logins so that we can track the 'passedTest' status for each user.

I'd definitely prefer to make it as easy as possible for users, at a cost of a bit more work for admins. Perhaps we can reconsider this if / when: (i) there's too much work for admins to do, (ii) we come up with a test and do the work to support it.

[poblish]

Have started work on this, and (1) - (3) [and part of (4)] are mostly done.

[poblish]

Hi Paul, could you have a think about how the fallacies should be displayed onscreen: some kind of highlighting, and definitely one of that site's icons, I should think. Would also be good to make available the display name of the submitter the date, and their comments - all of this in an elegant and uncluttered way.

Will have a go too!

[poblish]

Try-out version just committed! : 3eca916e2e6c77ecebfb734c628f7a89507f7a14

[pauliewaulie]

Andrew - for some reason, I can't find the new version to download - where is it? Remind me, does it work in Firefox?

Ta

PE

On 12 July 2012 00:20, Andrew Regan < reply@reply.github.com

wrote:

Try-out version just committed! : 3eca916e2e6c77ecebfb734c628f7a89507f7a14

---

Reply to this email directly or view it on GitHub: https://github.com/poblish/bannedList-Plugin/issues/30#issuecomment-6922858

Paul Evans Tel: 07973 714206 (mobile) paul@memeserver.co.uk http://www.memeserver.co.uk Twitter: @paul0evans1 http://www.twitter.com/paul0evans1 Skype: paul0evans1

LinkedIn: http://uk.linkedin.com/in/paulliamevans

+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ Social media for influencers: Free regular briefings here: http://www.memeserver.co.uk/briefing

Ever asked yourself 'why is politics still done like this?' http://www.politicalinnovation.org**

[pauliewaulie]

Andrew - I've loaded it up and looked at it. Boring things first, the 'name' field when I submit a logical fallacy is pre-populated with 'Andrew Regan' - I can change it easily but you may want to fix that ;-)

Secondly, when I hightlight a fallacy, and then reload the page, I can't see it hightlighted by the plugin. Is there some approval process that you run? Maybe first-time commenters need to register and set up an account? And maybe every tagged fallacy needs to be linked to from a separate website? This would be good because we could set up an RSS feed from it so that others could see them - help grow take-up?

[poblish]

Thanks for having a go. Manual approval at the moment: I don't even have a report page at the moment, but this'll be made a lot easier in due course, and, yep, we could easily expose the stats via RSS etc. The other thing is that I've deliberately restricted this to Telegraph, Guardian, and Blogspot pages at first (just imagine if we had to do a lookup for every single page you ever loaded!), so even though I've approved your two, they won't show up until I lift the restriction. Have another go with one of the above sites as I'd like you to see one of these things for real .

BTW, do you feel confident enough with Github to raise new issues for what you'd like to see and what you think we need? Otherwise I'm liable to start writing these things down on PostIt notes or storing them in an email draft.

Sure, can fix the name thing - wonder if it can be pre-populated...