Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Guardian's Typerighter

github.com

Post reply on HN
Print view View original

11–20 of 22 posts

Re: The Guardian's Typerighter

#11

Post by ghusto »

Speaking only on the technical aspects, I think this is a cool (technical) idea, but over-engineered:

> Both the Checker and Rule Manager services are built in Scala with the Play framework. Data in the Rule Manager is stored in a Postgres database, queried via ScalikeJDBC. > > Google credentials are fetched from SSM using AWS Credentials or Instance Role.

For what's essentially a regexp checker? Good lord. Sounds like someone had fun though.

Speaking on the sociological aspects, I like it less. How many squiggly red underlines do you think it will take before the writers fall in line and switch their creativity off? There's something to be said for technology staying out of the creative fields.

Speaking on the irony aspects (emphasis mine):

> The goal of the Typerighter project is to address the above challenges with a universally available document proofing service, via an _easy-to-use and minimal interface_, that is accurate, transparent, and responsive to change.

Re: The Guardian's Typerighter

#12

Post by paulddraper »

ghusto wrote: post #11

Speaking only on the technical aspects, I think this is a cool (technical) idea, but over-engineered: > Both the Checker and Rule Manager services are built in Scala with the Play framework. Data in the Rule Manager is stored in a Postgres database, queried via ScalikeJDBC. > > Google credentials are fetched from SSM using AWS Credentials or Instance Role. For what's essentially a regexp checker? Good lord. Sounds li…

I fail to see the problem

Re: The Guardian's Typerighter

#13

Post by b3lm0nt »

ghusto wrote: post #11

Speaking only on the technical aspects, I think this is a cool (technical) idea, but over-engineered: > Both the Checker and Rule Manager services are built in Scala with the Play framework. Data in the Rule Manager is stored in a Postgres database, queried via ScalikeJDBC. > > Google credentials are fetched from SSM using AWS Credentials or Instance Role. For what's essentially a regexp checker? Good lord. Sounds li…

Brings to mind the recent overview that Ars Technica did of their infrastructure stack: https://arstechnica.com/information-technology/2023/07/ars-o...

Approx. 12 different AWS services and 30+ instances...

Re: The Guardian's Typerighter

#14

Post by rhystmills »

weinzierl wrote: post #9

Thanks Guardian, for making this open source! I really would love to see rules as well. Open source would be best, but if this is not possible even a proprietary snapshot would be super interesting to see.

Typerighter uses an extended ruleset but the core is based on the Guardian's publically available style guide: https://www.theguardian.com/guardian-observer-style-guide-a

Re: The Guardian's Typerighter

#15

Post by vr46 »

sph wrote: post #10

Ironic. I was reading their opinion article by the Ukrainian ex-defense minister just this morning, and caught a very basic spelling mistake any spell checker would have found. https://www.theguardian.com/commentisfree/2023/sep/08/ukrain... It's still there. "morehas". Classic Grauniad. Let's see if any Guardian editor reads HN.

It was correct when I just looked at it, and you still read it, though. Eventual correction works and is less important than getting the news out.

Any-way - the fact that they have a system that seeks to improve the copy and direct it towards their style guide is something to be celebrated when most of the news about, erm, the news, is how AI is replacing actual writers.

Re: The Guardian's Typerighter

#16

Post by 0cf8612b2e1e »

ghusto wrote: post #11

Speaking only on the technical aspects, I think this is a cool (technical) idea, but over-engineered: > Both the Checker and Rule Manager services are built in Scala with the Play framework. Data in the Rule Manager is stored in a Postgres database, queried via ScalikeJDBC. > > Google credentials are fetched from SSM using AWS Credentials or Instance Role. For what's essentially a regexp checker? Good lord. Sounds li…

> How many squiggly red underlines do you think it will take before the writers fall in line and switch their creativity off?

That’s definitely a hard line to walk. On the other hand, not everyone is David Sedaris, and most writing should cut down on the cuteness and just give me the facts.

Re: The Guardian's Typerighter

#17

Post by sph »

vr46 wrote: post #15
sph wrote: post #10

Ironic. I was reading their opinion article by the Ukrainian ex-defense minister just this morning, and caught a very basic spelling mistake any spell checker would have found. https://www.theguardian.com/commentisfree/2023/sep/08/ukrain... It's still there. "morehas". Classic Grauniad. Let's see if any Guardian editor reads HN.

It was correct when I just looked at it, and you still read it, though. Eventual correction works and is less important than getting the news out. Any-way - the fact that they have a system that seeks to improve the copy and direct it towards their style guide is something to be celebrated when most of the news about, erm, the news, is how AI is replacing actual writers.

They fixed it some time between my comment and yours.

Maybe I'm spoiled with our programming tools, but I would have imagined a major newspaper to have an automated style and spell checking system that runs every time one presses Save on their CMS, and this is proof they don't. I am surprised why no one at The Guardian has set this up before — it's not rocket science, but their core business.

Re: The Guardian's Typerighter

#18

Post by nxobject »

bborud wrote: post #8

This is a bit tangential, but I think there are ways to increase the value of this project. And please, don't take this the wrong way. I'm trying to make a constructive suggestion (to the maintainers). Why not present it as a library with a sensible API and a directory of example uses. Including an example of a single binary, minimalistic server that has a very simple, minimal interface allowing you to submit documen…

I don't think it's tangential at all – tools are meant to be used by people, and thinking about how to make tools low-friction and easy-to-use is intrinsically interesting, regardless or not of whether it's meaningfully actionable :) But only the authors would know what's feasible for them. I can only imagine that working at a legacy newspaper is a resource-strapped place for software engineering.

Re: The Guardian's Typerighter

#19

Post by nenaoki »

bborud wrote: post #8

This is a bit tangential, but I think there are ways to increase the value of this project. And please, don't take this the wrong way. I'm trying to make a constructive suggestion (to the maintainers). Why not present it as a library with a sensible API and a directory of example uses. Including an example of a single binary, minimalistic server that has a very simple, minimal interface allowing you to submit documen…

Lately I'm less convinced everything ought be dolloped out on spoons; having the source available for even just a few willing to delve down into the weeds is already a boon, imho.

Not that I would campaign specifically against such improvements, but that I think having raw, uncut gems laying around can be fine.

Re: The Guardian's Typerighter

#20

Post by vr46 »

sph wrote: post #17
vr46 wrote: post #15

Earlier quoted context omitted.

It was correct when I just looked at it, and you still read it, though. Eventual correction works and is less important than getting the news out. Any-way - the fact that they have a system that seeks to improve the copy and direct it towards their style guide is something to be celebrated when most of the news about, erm, the news, is how AI is replacing actual writers.

They fixed it some time between my comment and yours. Maybe I'm spoiled with our programming tools, but I would have imagined a major newspaper to have an automated style and spell checking system that runs every time one presses Save on their CMS, and this is proof they don't. I am surprised why no one at The Guardian has set this up before — it's not rocket science, but their core business.

This is just about priorities and theirs are different to yours - firstly, we don't know that they don't have a system that catches and warns of possible errors, we only know that articles may be published without addressing them. Secondly, the fact that it was corrected post-publish suggests that errors are picked up, but timely publication is more important.

The core business is delivering news, and typos halting publication is not a trade-off they want to take, they'd rather have the subs or copy editors batch process or review all the corrections required - maybe they use Pull Requests? :D

Post reply on HN