Show HN: Pol.is – a new commenting system powered by machine learning and D3 | Hacker News Reader

Show HN: Pol.is – a new commenting system powered by machine learning and D3

pol.is

126 pointscolinmegill41 comments

I really like the landing page. Mostly I watch these intro screencasts prepared to wince and jump forward, but yours is quick to the point and non-annoying.

The product seems minimal in a nice way, but I'm very curious about the direction it could go. Like, having discussions somehow based on the calculated groups.

Based on my own "bored at the bus stop" type experiences, I'm guessing mobile users can tend to get pretty seriously involved with series of yes/no questions with some social component, and yes, I'm thinking about Tinder.

Thanks mbrock, we were just yesterday considering adding ukulele and glockenspiel to the video to make it completely inaudible.

Yes - you are looking at it the same way as we are. We've built a core that could be outfitted in a number of different ways to fit a variety of circumstances. Many of the features early users wanted were appropriate for users across verticals, and we built those first. Now we're onto things like making it 'one click' for professors who use course software, etc. There is no end in sight for that stuff, so user growth will at least in some part determine how much we invest in each vertical over the next couple months. cc Steve Blank

Didn't realize how sensitive to costs I was. I read, "try Polis free," but as I scrolled down it wasn't difficult to notice the image with $15/hour and it almost made me close my browser until I watched the video and then looked at the image again more closely before I realized it was about minimum wage. Based on my experience, I'd recommend using a different question and/or image. Just a thought. :)

Hahahaha! I suppose I've been looking at that too long to have thought of that. Yes, great idea. That would be quite a fee.

I, too, assumed that the $15/hour had to do with the cost of Polis.

I used to work for what I guess you could say is a "competitor" in this arena, and what you've done here is terrific; simple and to the point and the visualizations of group voting is easy to digest. I didn't even have to sign up to participate in the question in this thread about how to pronounce "pol.is".

Really well done. Good luck!

Thanks narsk!! Get in touch through the app if you'd like to share any more insights. As a young startup we're always looking to broaden our perspective.

Genuinely intrigued. However, $100/month seems like a pretty hefty price tag for an individual.

You would generally get a bigger sign up if there were multiple price points with different levels of service.

Really nice! I really appreciate the focus on good design and clear presentation of data. Something seeming this "simple" is not actually simple! Well done. I spend most of my life considering online tools for consensus building, and this is one of the more innovative and well executed I have seen.

I would love to see modules for this kind of non-binary polling in our open source decision-making tool[0]. I have never been a fan of straight polling because it misses some critical aspects of the discussion by pre-supposing the options, and most issues are not best approached with a conflict-based method of "A vs B, one wins". In fact the real answer is often a synthesis or evolution beyond where anyone started. Many of our users have requested polls to do temperature checks or decide quickly between options, but I have resisted because I don't want to flatten discussions into only a limited set of options that might leave out important aspects. I much prefer your approach!

[0] http://www.loomio.org

!! First, Amen. Second, just seeing loomio for the first time - could you create a pol.is account and message us? Would be great to share ideas.

Yes! I would love to get in touch and talk more. I will contact you directly. Best of luck with pol.is! The world really needs better ways to make decisions and we're going to need all kinds of complementary tools.

Is Loomio also embeddable and accessible through an API?

Not at the moment, but it's something we're working toward! We will have an API later this year, and embeddable Loomio decisions is on our roadmap.[0]

Of course as an open source project, if you had a strong use case you are more than welcome to jump in and build what you need it to do! [1]

[0] http://www.loomio.org/roadmap [1] http://github.com/loomio/loomio

Thanks. I do actually have a couple of potential use cases. Got to have a look at the github folder later today. Would be dfntly great to stay in contact with you.

I love the design of the demo stuff, but I really hate that I have to click-and-drag to go to the next bullet / click one of those little dot things. If I was swiping naturally with my finger on a touch-enabled device, sure, but I'm just going to hunch that most customers wouldn't be viewing analytics like that.

Are we forgetting what a good 'ol >next and <previous can do?

Those two buttons are sitting on a preproduction branch as you write this :) great call, agreed

You made my night.

It will be hard to top the love we've gotten here in terms of making our night. This makes our month.

But if you happen to have a handful of professors who could use pol.is in class this fall...

What an interesting and unique approach to online discourse! In general terms (not asking to reveal the secret sauce), how do you define the opinion groups? Latent Semantic Analysis?

Someday in the future we'll post a blog entry on all the math :)

Pronounced "poLEE" like the French word for polite ("polis" would be plural, male or mixed-gender)?

Or more like "police"?

Or Poe-less? See http://rationalwiki.org/wiki/Poe%27s_Law

Rather different interpretations (but all meaningful in the context of online discussion, interestingly!).

I thought to pòlis, in Greek.

Edit: https://pol.is/5hfrym

In the office we say the latter. The meaning behind the term isn't posted on the site but is appropriate to answering your question: http://en.wikipedia.org/wiki/Polis

In short, Aristotle thought there was a max size to the self governing city state, so we played off of that as the goal was to create a comment system that could handle tens of thousands of opinions and stay orderly (think virtual town hall / lots of people on the bus to work responding to a mayor's question to the whole city). Email, discussion threads, comments that scroll down, etc, top out pretty well beneath that, and simple upvoting and downvoting doesn't capture groups.

While we started out in politics, we're exploring a lot of different avenues right now with our current users, including education, replacing comments on blogs, etc.

In Greek, it actually sounds more like the "lis" in "lisp" (short i):

http://vocaroo.com/i/s07xxGjtkQB1

StavrosK, I spent a moment saying both and we have been saying it your way.

We've been saying: po as in police, lis as in lisp

I have heard that a valid pronunciation of the Greek is also:

po as in hot (closer to the 'a' sound that 'o' can make) lis as in lisp

You'll have to let us know if that's right!

The "o" in "police" is long, you need a short one. The English pronunciation is what you would transliterate as "poulis" (can't write IPA on the phone, but it's a short o followed by a oo sound). In Greek, the o is as in "hot" in duration, but not in sound (not close to the "a' sound).

There's only one way to say it in Greek (the one I posted above), I suspect that the person who instructed you to pronounce it as in "hot" meant the duration. I can't think of an example of an English word with a pure "o" sound at the moment, I'm afraid. You're going to have to listen to my pronunciation above to get it right :-P "O" as in "hot" is fine, as long as you don't pronounce it too much on the "a" side.

What I like about this is that it could be integrated into consensus-based decision making tools and processes. It would make it much much easier for a group to see what's happening overall and even in a hierarchical structure such as a corporation it would be handy to have.

It looks really nice too. Not the most exciting thing to be staring at but by god it's pretty. Like Game of Life.

We thought about replacing the dots with little animated people that run to their position...

Awesome comment - replying inline:

...could be integrated...

pol.is does have an API

...consensus based decision making tools and processes...

One of the most compelling pieces of feedback we got in two years of R&D was "...could be used for upstream analysis of potential downstream policy impacts..." which was exactly what we were thinking when we built it. When I worked briefly in D.C. ... well imagine someone in charge of forestry policy who has never worked with the people in the field. How could he write a proper survey? Open ended text is vastly superior for gathering their feedback in response to potential policy changes. We're trying to make that seamless.

...much much easier for a group to see what's happening...

It's probably a bit heady but our driving purpose is to help organizations become more conscious of themselves, be they governments or corporations, and you speak to that point. That's why we show the visualization to all participants.

I only had a cursory look at the demo so excuse me if this is explained somewhere on the site and I missed it... would it work if all the questions were written in a different language from English? (I am not sure if there is some "sentiment analysis" applied to the questions, for examples, and if this is possible only for English due to libraries etc.).

It works for any language, though we haven't localized the UI yet. Here's a recent conversation that ran in Taiwan.

https://pol.is/8bfzc9

(We feel ok about sharing this since the owner posted it to Twitter)

This is interesting and the way I picture it, it must be very complicated.

Let's say you save the url and the selector, and scrap it periodically over time, how do you handle HTML change over time ? or does it simply stop tracking the page in that case ?

Sorry- mistakenly posted on a different thread.

Really nice work. Well done! Is it possible to display user names via the API?

Yes. We're presently focusing on taking in users from 'your' context. 'Display' in this case would mean 'see who was in what group in CSV format when the conversation ends'. Conversations can be thousands of people, so it won't work to show the names in the visualization.

Thx Colin. Sounds great for analytical purposes as well as concensus building research! Am dfntly looking forward to give it a try.

Does this have any algorithmic defense against astroturfing?

No, at the moment. This is less important for conversations that are more 'cloistered', like those that might happen internally, in person at a conference, in a town hall, or in classrooms.

On the open internet, it's obviously more complicated, we have a number of things in our pipeline e.g., taking in cred / user context from other communities through our API

This is pretty cool. One minor complaint: when I vote, the animation of my dot moving seems painfully slow.

Now let groups date: Boom, Match.com acquisition.

I keep reading "pol.is" in DCI Matt Burke's voice, from Taggart.

Reply on news.ycombinator.com