I've been looking through hundreds of job ads for about a month now. This is honestly one of the best. They are putting emphasis on your passionate and fit, rather listing random technology stacks and bs requirements.
Thank you. We worked hard on our last two or three job advertisements. This represents the culmination of probably 8 months of conversation and experimentation.
Yes! There is a need for more people in journalism who understand the principles of data relationships and are comfortable with modern technology. Whoever takes this job or one at another newspaper, here's a subset of my desiderata for modern journalists:
Every article should have machine readable metadata including:
* The author
* The geographic and political point(s) or region(s) discussed in the article
* URLs of related content (serious news outlets need to fight their desire to keep the user from leaving the site)
* A boolean that indicates if the article uses anonymous sources (tools could then filter out such articles when desired)
* Implicit version control showing each published revision of an article and who made the change
> * Implicit version control showing each published revision of an article and who made the change
That last one made me laugh. With traditional print, you'd have a newer version of an article (say the afternoon edition) highlight the changes or the newspaper would publish a correction in a separate section of the paper.
You'd think with modern technology there would be a better way of presenting that type of information but it seems like the real response from the publishing industry has been to simply perform ninja edits and hope nobody notices.
Yes, I've noticed the same thing and it has annoyed me. Newspapers are historical artifacts. They have a duty to history to be accountable about such changes.
Actually, I'd love to see Google require this (and possibly, additions), for qualifications for Google News listings.
Most especially: reputation tracking of reporters, authors, editors, and publishers, most especially on accuracy (inclusive of corrections, which reduce but don't eliminate error penalties).
> Actually, I'd love to see Google require this[...]: reputation tracking of reporters
I'm pretty sure I do not want my search engine to assign some "reputation" values. Instead, I'd prefer people to get educated about their news sources and make their own choices.
SEO-based "reputation" is just a misnomer; asking for actual people being rated opens the door to easy silencing of critics.
The problem is not the media, but our consumer attitude to news: Only a few track down the source of a news and read the actual ticker message or scientific paper. Even fewer understand the background of a media outlet, its tone/agenda and business relations.
In your Utopos the "truths and facts" would not be your own, it would be those of whoever controls the reputation database. You can watch a system like this unfold in China[1] - and because it is China, we all agree without a second thought that this is a tool for oppression...
Any reputation or recommendation system is subject to gaming. The questions are: what fundamentally does it promote, what are its internal incentive structures, and is it fundamentally trustable.
You don't get away from any of those questions (or concerns) no matter what you replace it with. Aiming at truth itself strikes me as vastly preferable to the extant model.
Understanding that truth itself can be imprecise, and building considerations of this (say: through fuzzing or randomising SERP) into the system also helps.
What significant amounts of research, ranging from highly formal to informal, have show, is that the lack of accountability by reptuation has lead to numerous actors hijacking our epistemic systems. Lauren Weinstein, Pew, Snopes, ProPublica, and others, have reported on this at length.
Would you pay for something like this? Because if not, I see zero reason for publishers to implement this with their already limited amount of developer labor.
Yes. I already do pay for subscriptions to various newspapers/periodicals. I see plenty of reason for publishers to rise above what other publishers are doing and provide more value.
These are all desirable, but IMO also really miss the point.
Jobs like this are about the input of a story, not the output. Ensuring that a health story is backed up by thorough analysis of government data. Or a story created entirely from scratch because a bot noticed an uptick in some particular dataset.
That has the potential to be really powerful journalism. Metadata isn't unimportant, but it's not more important than the story itself.
(IIRC, the NYT already has APIs that fulfill almost all of your requests, incidentally)
I don't disagree that what I describe may not be what NYT is looking for. I'm excited enough about their commitment to devote programmer resources to news that I am eager to provide that feedback.
> (IIRC, the NYT already has APIs that fulfill almost all of your requests, incidentally)
Really? I poked around and just now registered for an API key. I see no documentation more detailed than "Data is returned in JSON". I'll keep looking.
Hm, the site has changed since I saw it last, seemingly to only include automated API documentation. That's a shame.
IIRC it's the Article Search API - it has a field named "fl" that controls what fields are returned, but it now the documentation doesn't seem to specify what fields it can return.
> Jobs like this are about the input of a story, not the output.
> Metadata isn't unimportant, but it's not more important than the story itself.
I see redefining what a modern reader should expect of serious journalism as a forcing function that would prompt the writer/editor to provide that data. It would affect how they report and provide better input.
I agree, but I wouldn't expect to see news orgs provide this data until readers actually do demand it. Most are (slowly) going out of business, so they're not about to spend very expensive developer resources on features few people are asking for.
Unfortunately, I think you are correct. No newspaper is likely to invest in raising the bar and I'm in a small minority that wants these features. I fear that all providers of serious journalism will keep chasing advertising revenue instead of serving readers.
@untog's got this one right. There are other teams that focus on improving the CMS, website and apps — but this job posting is specifically geared towards research, data analysis, reporting, tools for search/viz/compare, databases for journalists to rely on, and so on, and so forth. As he says, the input.
I think I understand the distinction in what you're looking for in this opening. I'm still excited about the idea of journalism taking it's output more seriously. Until news organizations get that, how will they know what the input should be?
I'm serious about wanting an attribute for unnamed sources. I would love for NYT and The Economist to have a preference setting where I can tell it "don't show me articles which use unnamed sources".
You can still have a masterpiece newspaper article without metadata dripping all over it, but no amount of metadata is going to fix shoddy journalism. Fortunately, the metadata stuff isn't rocket science.
What is rocket science, however, is computational journalism. I think the NYT is onto something here. What happens when you pump up great journalists with the ability to mine data, create visualizations, and explore complex relationships that are intractable without computational assistance?
First question. What languages are in use? Because I know LAMP, but this might be on Microsoft or MEAN. In which case I'll have found and cleaned up 3 projects and a resume for no reason. All the information needs to be presented upfront, I don't want to have to go looking for it.
I believe it's somewhat flexible — as this post is for a new, small team (within Interactive News). So there's not necessarily a big existing codebase to maintain, and they're going to be working on some greenfield reporting tools.
That said, we use a lot of JavaScript, a fair amount of R, a medium amount of Ruby, some Python, and occasionally Go. Stats skills are a plus, as are database skills, as are design chops.
Hi! We're a mix of Python, Ruby, Node and the occasional bit of Go. We decided not to add that to the job description because we don't make our hiring decisions based on command of a particular language. If you have concerns, send me an email: jeremy.bowers@nytimes.com.
I want to join the NYTimes and build a "chronicle" feature. I hate how news stories develop overtime, but there is nothing to "chronicle" the story from beginning to end.
Linking to related stories is a temporary fix, but a standard timeline would be better. It should even motivate people to contribute to "follow-ups" on stories. Report what happened to the affected parties, where are they now, etc...
This is in the interest of keeping people informed and engaged. Instead of endlessly chasing "breaking news"
Take the "Oroville Dam" story for example. I would love to be able to subscribe to a timeline of updates as more unfold. A year from now, will there be a story about how the dam was fixed? How government money was spent? etc...
Or whatever happened to those girls who were locked in that guy's basement? Whatever happened to them? Where are they now?
I recently purchased the domain cnnisfake.news to expose all the times the media releases a fake story. This thread and posts have inspired me to get it working!!
I had a similar idea, but wanted to build it outside of a single news organisation. Basically a news aggregator, but it would detect when different articles are about the same topic and try to tell if the newer article added anything to the story. This aggregation system would then track public interest in a story over time (based on clicks) along with media interest (based on new stories)
I also wanted to add in user, publication and reporter 'leanings' by having a mechanism where users could say that they thought a story was left/right/neutral. The act of voting would count as a push having the affect of slightly moving the publication and reporter in one direction and the voter in another. I would then use an ELO type metric so a user who was very far in one direction wouldn't have the same impact as a user more in the center.
That's actually awesome. There is a lot of stigma around "duplicate" posts. But I think in the context of showing different angles of the same story, that would be cool.
I've been toying with this very idea for quite some time. I don't like that I have to seek out a complete story instead of being able to see links that form a narrative in a single place. I also wish I could easily see different sources of the same information to compare the perspective.
Sorry about that — I think I posted it a few minutes before the hiring thread went up (or at least, a couple hours before I noticed it).
I wasn't really expecting this listing to reach the front page — but I can understand that it did.
There's been so much energy / handwringing / chatter in recent months about our increasingly post-factual era, governments' relationships with journalism, and possible technical remedies, that I thought this new team (which has some very sharp people on it) might strike a chord here.
Heads up that the mailbot is throwing out responses saying there's no jobs available. Because this was posted today, I'm sure if that means there are no longer any jobs available or the responder just hasn't been updated.
> Thank you for your interest in Interactive News. We do not have open positions at this time but we would like to stay in contact with you as our hiring needs evolve.
To hazard a response, meeting a deadline creatively is knowing how best to trim, tweak and reimagine your concept to get in done in time.
One of the most fun and different aspects about writing code in a newsroom is the extremely different timescales that are involved compared to normal programming.
If you think it sounds like fun to sit down in front of a blank page of HTML with a fresh government dataset, RStudio, a copy of D3, a cup of coffee and 36 hours to see how good of an exploration and explanation you can come up with, this might be the job for you.
You know the saw about software development; if the features are fixed, flex the deadline. We're the opposite: We can't move a general election or we might have to respond to breaking news. We value your ability to make decisions about how to get a piece of software working quickly but without making it impossible to maintain. It requires creativity, and we think of it more of an art than a science.
59 comments
[ 3.4 ms ] story [ 143 ms ] thread* actively advertising open source contributions
* not requiring endless whiteboarding sessions
* wanting to review past projects as a large part of the interview
Their open source blog is also quite interesting: https://open.blogs.nytimes.com
Bravo NYTimes. Proving again that you get it.
Every article should have machine readable metadata including:
* The author
* The geographic and political point(s) or region(s) discussed in the article
* URLs of related content (serious news outlets need to fight their desire to keep the user from leaving the site)
* A boolean that indicates if the article uses anonymous sources (tools could then filter out such articles when desired)
* Implicit version control showing each published revision of an article and who made the change
That last one made me laugh. With traditional print, you'd have a newer version of an article (say the afternoon edition) highlight the changes or the newspaper would publish a correction in a separate section of the paper.
You'd think with modern technology there would be a better way of presenting that type of information but it seems like the real response from the publishing industry has been to simply perform ninja edits and hope nobody notices.
http://microformats.org/wiki/hnews
Support this. ALL MEDIA, please.
Actually, I'd love to see Google require this (and possibly, additions), for qualifications for Google News listings.
Most especially: reputation tracking of reporters, authors, editors, and publishers, most especially on accuracy (inclusive of corrections, which reduce but don't eliminate error penalties).
I'm pretty sure I do not want my search engine to assign some "reputation" values. Instead, I'd prefer people to get educated about their news sources and make their own choices.
I'd like to see those have some foundation in truth and fact rather than SEO inflation.
The problem is not the media, but our consumer attitude to news: Only a few track down the source of a news and read the actual ticker message or scientific paper. Even fewer understand the background of a media outlet, its tone/agenda and business relations.
In your Utopos the "truths and facts" would not be your own, it would be those of whoever controls the reputation database. You can watch a system like this unfold in China[1] - and because it is China, we all agree without a second thought that this is a tool for oppression...
[1] https://chinacopyrightandmedia.wordpress.com/2014/06/14/plan...
https://en.m.wikipedia.org/wiki/Criteria_of_truth
Any reputation or recommendation system is subject to gaming. The questions are: what fundamentally does it promote, what are its internal incentive structures, and is it fundamentally trustable.
You don't get away from any of those questions (or concerns) no matter what you replace it with. Aiming at truth itself strikes me as vastly preferable to the extant model.
Understanding that truth itself can be imprecise, and building considerations of this (say: through fuzzing or randomising SERP) into the system also helps.
What significant amounts of research, ranging from highly formal to informal, have show, is that the lack of accountability by reptuation has lead to numerous actors hijacking our epistemic systems. Lauren Weinstein, Pew, Snopes, ProPublica, and others, have reported on this at length.
https://www.reddit.com/r/dredmorbius/comments/5wg0hp/when_ep...
Jobs like this are about the input of a story, not the output. Ensuring that a health story is backed up by thorough analysis of government data. Or a story created entirely from scratch because a bot noticed an uptick in some particular dataset.
That has the potential to be really powerful journalism. Metadata isn't unimportant, but it's not more important than the story itself.
(IIRC, the NYT already has APIs that fulfill almost all of your requests, incidentally)
Really? I poked around and just now registered for an API key. I see no documentation more detailed than "Data is returned in JSON". I'll keep looking.
IIRC it's the Article Search API - it has a field named "fl" that controls what fields are returned, but it now the documentation doesn't seem to specify what fields it can return.
> Metadata isn't unimportant, but it's not more important than the story itself.
I see redefining what a modern reader should expect of serious journalism as a forcing function that would prompt the writer/editor to provide that data. It would affect how they report and provide better input.
I'm serious about wanting an attribute for unnamed sources. I would love for NYT and The Economist to have a preference setting where I can tell it "don't show me articles which use unnamed sources".
You can still have a masterpiece newspaper article without metadata dripping all over it, but no amount of metadata is going to fix shoddy journalism. Fortunately, the metadata stuff isn't rocket science.
What is rocket science, however, is computational journalism. I think the NYT is onto something here. What happens when you pump up great journalists with the ability to mine data, create visualizations, and explore complex relationships that are intractable without computational assistance?
http://www.nytco.com/careers/newsroom-summer-internships/
That said, we use a lot of JavaScript, a fair amount of R, a medium amount of Ruby, some Python, and occasionally Go. Stats skills are a plus, as are database skills, as are design chops.
Linking to related stories is a temporary fix, but a standard timeline would be better. It should even motivate people to contribute to "follow-ups" on stories. Report what happened to the affected parties, where are they now, etc...
This is in the interest of keeping people informed and engaged. Instead of endlessly chasing "breaking news"
Take the "Oroville Dam" story for example. I would love to be able to subscribe to a timeline of updates as more unfold. A year from now, will there be a story about how the dam was fixed? How government money was spent? etc...
Or whatever happened to those girls who were locked in that guy's basement? Whatever happened to them? Where are they now?
I recently purchased the domain cnnisfake.news to expose all the times the media releases a fake story. This thread and posts have inspired me to get it working!!
I also wanted to add in user, publication and reporter 'leanings' by having a mechanism where users could say that they thought a story was left/right/neutral. The act of voting would count as a push having the affect of slightly moving the publication and reporter in one direction and the voter in another. I would then use an ELO type metric so a user who was very far in one direction wouldn't have the same impact as a user more in the center.
That's a really good idea. Maybe we can all collab and make this.
http://digiday.com/media/two-years-vox-com-reconsiders-card-...
https://blog.medium.com/welcome-to-series-a-new-type-of-stor...
I wasn't really expecting this listing to reach the front page — but I can understand that it did.
There's been so much energy / handwringing / chatter in recent months about our increasingly post-factual era, governments' relationships with journalism, and possible technical remedies, that I thought this new team (which has some very sharp people on it) might strike a chord here.
> Thank you for your interest in Interactive News. We do not have open positions at this time but we would like to stay in contact with you as our hiring needs evolve.
Edit: Bowers says that he's fixed it now.
One of the most fun and different aspects about writing code in a newsroom is the extremely different timescales that are involved compared to normal programming.
If you think it sounds like fun to sit down in front of a blank page of HTML with a fresh government dataset, RStudio, a copy of D3, a cup of coffee and 36 hours to see how good of an exploration and explanation you can come up with, this might be the job for you.