Hey everyone, this is a weekend project I built - it lets you search through US Presidential candidates’ speeches for topics and see what both candidates have said side-by-side (on a largish screen).
My thinking is that with all the divisive misinformation making their rounds these days, hearing what both candidates have said first-hand on the same topic presented in a side-by-side format, could help inform people of all perspectives in an unbiased way.
Some thoughts:
If there's a way to dilineate between education in a title versus a topic ("Secretary of Education", "patriotic education"), might be good.
I'm finding it weirdly hard to go from "the word is here" to "this is the context in which it was said". Some thoughts:
If there's a way to always have the search term centered in the block of text, ala that speed reading program.
Less aggressive highlighting. Maybe just change the font color, instead of highlighting and spacing.
Maybe change the text color of the sentence that it's in (as well), to help me quickly find the start and end of the most-relevant sentence?
Thank you for the UI feedback. Are you referring to the text in the main cards, or in the modal when you click on “Read in context”?
Re: differentiating between titles vs topics, that would be a hard thing to do with the current setup I have, which is for doing keyword-based searches.
Would you be interested in expanding this to other candidates (like those in other races, running for Vice President, or nominated by other parties)? Would it add significantly more work?
I’d love to. If you’re able to find a repository of transcripts for the speeches, I can add them to the index. May be users can then pick which two candidates to see side-by-side.
I'm not trying to be snarky here, but are you using 'ducking' as a verb for using duckduckgo? I ask because that makes sense in context, but I naturally assume that word to be a more polite version of 'fucking' because that's how I'm used to seeing it being used. Is ducking something people say?
So, the problem with this stuff is that the content is one thing, but just like the semantic web the semantics are quite another thing. What does proposal X or claim Y mean?
For example, "We will stop the radical indoctrination of our students and restore patriotic education to our schools." What does "racial indoctrination" refer to? What does "patriotic education" refer to? I have a strong idea, but that's just my opinion.
Politicians try to stir as many emotions as possible without actually saying much. That way they don't have to follow through with anything but still get people motivated to support them.
It wasn't that long ago that Texas nixed critical thinking from K-12 education. The reason given was, and I wish I was making this up, that teaching children critical thinking might cause kids to question their parents and their church.
> How about we just teach kids to think for themselves?
Have you tried? This is something that the majority of people will not do no matter how much you tell them they should. They don't want to, and many can't.
Are you basing your hypothesis on the adult population that has gone through our existing pro-conformity, pro-follow-instructions educational system...?
That's a good point and would be cool to have that. I built this primarily to act as a starting point for keyword searches. Hopefully the quotes from the candidates are intriguing enough to prompt further independent research.
Do you know any datasets that provide good unbiased definitions of these terms that candidates use? Let me know if you do, I can index them and add to the UI.
I'd like to propose an idea - I'm not saying it's a good idea, just an idea.
Rather than incorporating "datasets that provide good unbiased definitions of these terms that candidates use", allow users (ideally quality controlled in some magical way) to annotate specific words and phrases, and attach comments to them (like: "We will stop the radical indoctrination of our students and restore patriotic education to our schools." What does "racial indoctrination" refer to? What does "patriotic education" refer to?) This would then make the site not just a useful informational tool, but also add a social media meta-conversation type of layer on top - in the case of the above comments, revealing to other users the vacuousness of political rhetoric, which a lot of people seem to take at face value, as a sincere and honest representation of true (as opposed to truthy) intent. Politics is largely disingenuous theatre imho, I think it would be healthy for this to be pointed out to people more often.
Obviously, there are all sorts of problems with this, one of which is that cynical people like me would use it for shitposting and that sort of thing. To that, I would suggest forcing a user to choose some sort of a ~perspective from which they are posting, and to be honest and take things seriously (and enforce this with an iron fist) - so in my case, I would often choose "Hyperbolic Cynicism" and treat the whole thing like the mockery that I perceive it to be (here I mean: both candidates, and the entire system). And other users could have the option of filtering out idiot comments like mine (or, choose to only see comments of this kind, or whatever the individual prefers).
And then in addition to speeches, you could also host various news articles, and let people do the same thing. Yes, there is noteworthy overlap with Reddit and many other sites, but there's also plenty of leeway for significant differentiation.
Of course, this would significantly change (and add risk to) the nature of your website, so perhaps you could do a completely different, renamed/unaffiliated version of it and tries this angle out.
Anyways, just an idea. Overall, I think it's a good idea and really nicely done, good job.
You're right, but its difficult for a website (or anything really) to encode the semantics without inviting criticism. These speeches, along with news broadcasts, make assumptions about their audience, and their content is carefully designed to appeal to that audience's biases and value system while appearing "neutral" or "factual" on the surface.
In this case, Trump's speech makes the assumption that his audience sees themselves as "patriots" and subscribe to the narrative that their way of life is under attack by people who "hate America" and want to indoctrinate people to do the same. This paints the other side as being against their value system and champions their values.
The problem is, even explaining that in those terms will cause people to take exception, so it's both difficult and dangerous for a website to do so and remain "factual" or "neutral".
Most people really hate it when you tell them things that run counter to their narrative, and have a hard time even seeing their narratives as biased one way or another.
That's a lot of discussion and analysis to add to a weekend project especially in today's media environment where stuff like QAnon sprouts new memes and conspiracies every day.
In this specific case Trump is grousing about non whitewashed version of American history that include education about slavery in the US, treatment of Indian/Indigenous tribes, etc etc. and generally anything that questions America/Capitalism/Traditional Gender Roles/etc. In short close to anything vaguely progressive happening in schools counts as "radical indoctrination" for some fragments of the right side of American politics.
There is the concept of an “interlinear”, text inserted in between lines of an existing text, usually translations but not always. The inserted text could be a “reading between the lines” of the original text. As in the novel “Interlinear to Cabeza de Vaca” by Haniel Long (I’ll mention that Henry Miller wrote an introduction to an edition of it, a recommendation for some).
I have wondered whether, for public speeches, someone could put together a wiki-like site so the public would be able to make comments (“interlinears”) on each single sentence of a given public speech. So one page of comments per sentence, or perhaps even a single phrase.
The structure, rigorously enforced, of focusing comments on the meaning and implications of a single sentence might keep the troll noise down or at least make moderating easier. The current rantfest in comments on political content is discouraging.
Is there any sense to the alignments across the row? or are they just all chronologically ordered or something. Ah just noticed the `Recent Fist`, `Oldest First`. So it's two streams side by side. Might help to avoid making equally sized rows in the two columns to avoid visually suggesting a pairing is going on.
Yeah, they’re two side-by-side streams. I can see how that can make it confusing. I initially had the rows take up only as much space as the words, but ran into problems further down with positioning the common Show More button where one stream would run longer than the other because I currently load a fixed number of rows per stream. I then would have to detect how many rows are needed to fill up each stream and load the appropriate number. Got lazy and decided to skip the complexity for a weekend project!
Do I understand correctly that they spend years preparing for a presidential race, but can't sit down with their teams and make a roadmap on what they plan to achieve in 4 years? So in 4 years we can actually see what they have achieved. Is that really that much to ask?
I imagine such detail wouldn't make it to the news. That probably exists but nobody cares because they'll lose interest before they get half way through reading it. Same way that (almost) nobody cares about side-by-side searching of their speeches. There are also voting records that you can look at to see what they really stand for, but again, nobody cares enough to bother.
This is a great UI, but the quality of the results will depend on the selection bias of the inputs. For example, if you used rally speeches for one side, vs prepared speeches for the other side, that would skew things quite a bit. Can you elaborate on the corpus which is backing this?
I've been working on something similar, but it's got less to say because we've constrained ourselves to just the presidential debates, since they're "fair". https://mytake.org/search?q=supreme%20court
I'm a big fan of "provably neutral" project to try to bring reality into politics, so I hope your project is successful!
It covers less than 50% of the candidates, so is not really complete, but is a good effort.
In any case, facts say more than speeches. Politicians just say that their public is expecting, so there are a lot of cherrypick in the concept of having webs like these.
Examples: Covid is at the same time "very worrying and letal" or "nothing to be afraid of, the people wearing masks are pussy" depending on the public
To have a webpage comparing the speech of each candidate against himselves under different publics would be much more interesting.
Pretty clean website! One concern I have is the comparison of speeches made relatively far apart. References to Covid in March are going to be pretty dated when compared to references made in October.
Thanks! The results are sorted by recency independently on both sides. So it should show the most recent comment a candidate has made on top by default, unless you choose to sort a column differently in the UI.
My hope was that providing a way to sort by recency will also help show candidates' position changes over time.
I searched for “crypto”, and it reports 34 results, but all of them seem to be creative ideas about which other words I might have misspelled, like “crystal”, “creator” or “scripture”.
That's typo correction kicking in. If a term (like crypto) has zero or very few occurrences in the candidates' speeches, the typo correction logic kicks-in and starts looking for additional results that are close misspellings of the word.
I'm thinking of indicating in the UI, when typo corrected results show up in the results vs the original search terms.
You should definitely at least indicate "no references found" and maybe respect common search engine flags like quotes, minus to exclude words specifically, etc.
If you search "Climate" for example, it's helpful to know that one candidate has a lot more to say than the other. Instead, you're filling the results with searches for "ultimate" (as a close misspelling?) and making it appear that they both speak about this equally.
Are there really lots of misspellings in these speech transcripts? Why not just turn off this misspelling feature if has a potential to mislead? Surely that's better than the much more unlikely chance of missing out a potentially relevant result that happens to have a typo.
These two old guys probably haven't said or even really thought the word 'crypto' so the spelling correction starts guessing nearest neighbors from their actual speeches.
Randomly clicked on "amendment", which shows noticeably different topics for biden vs trump. Trump mainly speaks about 'the second amendment', whereas biden tends to use it more colloquially.
Perhaps an n-gram search for multi-term searches and then settling on the top one?
My thinking is that with all the divisive misinformation making their rounds these days, hearing what both candidates have said first-hand on the same topic presented in a side-by-side format, could help inform people of all perspectives in an unbiased way.