Show HN: Summaries of interesting content, written by people

by louischatriot • 14 years ago

131 103 14 years ago

sync• 14 years ago

Lovely drag/drop bookmarklet tutorial w/ animated arrow and helpful prompts and everything.

Though, really, I would like to try the service out without having to drag/drop a bookmarklet and then going off to another website and then clicking tl;dr. Can't I just put in a URL to try it out? Or provide a demo URL for us to try out?

louischatriot17• 14 years ago

Thanks for the feedback on the onboarding :)

It is not yet possible to try the service without the bookmarklet, but here is an example of a summary we created for the video of Anders Hejlsberg on Typescript that is on the frontpage so that you can see what it looks like : http://api.tldr.io/tldrs/506aee4ec8f814a2370004fa

wodow• 14 years ago

For me, the bookmarklet tutorial pointed the arrow at the address bar and then congratulated me when I overwrote the page's URL with javascript by dropping the button. (Firefox 15.0.1 on Linux). But I do like the idea!

louischatriot17• 14 years ago

Yep, sorry about that! It is not possible (with Chrome nor Firefox) to detect if the link was correctly dropped on the bookmarks bar or above.

Thanks for the feedback!

no_news_is• 14 years ago

This goes in an iframe on the page as an overlay.

http://tldr.io/iframe/iframe.html?url=http%3A%2F%2Fblogmaver...

jey2• 14 years ago

Yeah, I was expecting something on the homepage like a top 10 list of popular tl;drs that I could try out without doing anything else first.

czzarr25• 14 years ago

hey, sorry about that, for now you can get them by following us on twitter!

louischatriot17• 14 years ago

We are considering such a feature but don't want it to disrupt the onboarding flow, so it is a tough ongoing debate right now :)

vhf• 14 years ago

>whithout [...] going off to another website and then clicking tl;dr.

I had the same feeling, but then I clicked tl;dr on the page where they show you how to add the bookmarklet !

stfu• 14 years ago

Exactly. Especially if a site wants to get a permanent spot in my browser, it must demonstrate its value upfront. And without some good examples this is not going to happen.

czzarr25• 14 years ago

sorry about that, we will make sure to work on showing the value more upfront

louischatriot17• 14 years ago

Cofounder here. We put such test URLs at the end of the onboarding process but it is true they may appear too late! Thanks for the feedback!

jmsduran• 14 years ago

Yes I agree. Once I found that I had to install a bookmarklet into my browser in order to try it, I just exited the site altogether.

A demo without having to login/install anything extra would be nice.

matthieurouif• 14 years ago

I love the few summaries I could read, it is to the point. If you can find writers, it could become a must have

Charlesmigli3• 14 years ago

Cofounder here. Thx that's cool. Building the community is the true challenge you're right.

tarr11• 14 years ago

This is a great idea and a solid execution.

It should automatically stick the tldr on the page if I installed the bookmark, and a tldr exists.

Charlesmigli3• 14 years ago

Hey cofounder here. Thanks for your feedback. What you mention is a great feature that will be available when we'll release browser extensions.

alecdibble• 14 years ago

That's a good idea! Have you thought about making Chrome/Firefox extensions as well to interface with your service? I feel like the extensions would be more integrated and provide a better user experience. I would use this service if an extension notified me that the article had been "tldr'd. Having to remember to click it is inconvenient and I will probably forget.

Overall though, it's a really good idea and I am interested to see how you flesh it out.

Charlesmigli3• 14 years ago

Hey, cofounder here. Thanks for the feedback, you are totally right the bookmarklet is a first prototype to gauge interest for the product. We will package it as browser extensions for the reasons you mentioned and other awesome ideas that we have in stock.

alexcabrera• 14 years ago

+1 for Chrome extension. I don't have a bookmark bar visible by default (too much chrome for Chrome).

jamespitts• 14 years ago

Pretty cool.

You might want to put a list of the day's top tl;drs on your landing page, or def. under "want some". When you roll over the title of the story, you get a floating tl;dr.

This would encourage skeptical users to do what you want them to do (drag that link up to the toolbar).

czzarr25• 14 years ago

Thanks! 3rd cofounder here. This is something we are indeed working on, but we put off the launch for too long already and really wanted to get feedback. You're right though that discoverability isn't great as it is. Try it on most links here on the HN frontpage though, we're doing them for you today!

asdfprou• 14 years ago

Onboarding process was awesome and effective. I went to the site meaning to just check it out but the process was almost too easy and now I have the tldr bookmarklet just sitting there on my toolbar.

Also I love the hover feature on the bookmarklet when you first see it.

louischatriot17• 14 years ago

Cofounder here, thank you for your comment on the onboarding process! We spent a lot of time on it :)

rkda• 14 years ago

Isn't this what gis.to does?

louischatriot17• 14 years ago

Indeed the philosophy is close to gis.to: having humans, and not computers, write summaries. There are some key differences though, mainly the bookmarklet which enables you to create a summary of the webpage you're on without having to leave it. Also, I am under the impression that gis.to is not active anymore, the latest summary dates back to January, correct me if I'm wrong ?

wiradikusuma• 14 years ago

This is great! Love the cute effect when dragging! I think the content of TL;DR should be in Wiki a'la StackOverflow, so people can refine it together. And probably showing a couple, not just one, as TL;DR can be very opinionated. And/or show if the TL;DR is written by the content owner to give "authoritative" feeling. Anyway I understand if it's too much then it won't be "TL;DR" anymore, so you must find the balance.

czzarr25• 14 years ago

Hi, thanks for the feedback we're glad you like it! To answer your questions: - all tldrs are editable by users, so that the content can be refined by the community. - we feel that tl;drs should convey the message and tone of the author of the underlying resource and as such shouldn't be opinionated. - we would love to help content owners author, spread and control tl;drs of their content. If a content owner here is interested in talking with us about such opportunities, please contact us on twitter or by email

ArekDymalski3• 14 years ago

Got "We detected that you are using Internet Explorer to browse this site" on Android Chrome. Will checkk laterr.

czzarr25• 14 years ago

Hi, this is strange as we followed the html5 boilerplate way of detecting IE. We will look into the matter, thanks for the feedback. As a sidenote, our site isn't optimized for mobile (to put it mildly) so I recommend trying it on a computer.

carlesfe2• 14 years ago

This is a fantastic idea. I'm going to give it a try for some weeks and see how it develops. Thanks.

By the way, a slightly more elaborated version would be much appreciated. I'm thinking of

- A button/notification which changes color when there is a TL;DR available

- Something sidebar-like

- Better community integration

louischatriot17• 14 years ago

Cofounder here, thanks for your feedback! This is indeed an early version, and to answer your points more precisely:

- This is indeed planned, and as you can see in the comments you're not the only one thinking that! It cannot be done with a bookmarklet but we will release Chrome and Firefox extensions that will have this feature.

- I don't get what you mean by "something sidebar-like", are you talking of a pannel on the top of your screen automatically displaying the summary if it exists?

- Yep, building an active community around tldr.io will likely be key to its success. We are thinking hard on how to do it, and if you have any ideas or suggestion they are more than welcome

carlesfe2• 14 years ago

Forget about the sidebar. I thought it was a good idea , but it isn't.

Regarding the community, It wasn't in the line of "building a community as a key to success" but rather give profiles some more strength, use something karma-like to reward users who work, the ability to score TL;DRs and allow multiple per page, etc.

mmahemoff• 14 years ago

I really like the bookmarklet intro, but my concern with a bookmark is it's static. If it was an extension, it could change to show if there's already a tl;dr present.

The web is a big place, so realistically there will almost never be a tl;dr waiting, for at least the first year of this service, which means the bookmarklet is effectively a "create tl;dr" button.

louischatriot17• 14 years ago

That is true, and we are planning on making such an extension.

You are also right about the fact that the bookmarklet is currently a "create tl;dr" button, since we have so little of the web covered. For now we can provide you with tldrs this way: - You can follow @tldrio on twitter or like the "tldr" page on facebook - We are planning on summarizing all (at least most) of the articles from some well known blogs such as AVC and bothsidesofthetable - We are thinking of a newsletter with the latest hot tldrs - If you have any other ideas please share them !

peloton• 14 years ago

I actually preferred the bookmarklet first approach (vs extension). I'd much rather test a bookmarklet first and then graduate to a browser extension.

czzarr25• 14 years ago

thanks for the feedback, this is one of the reason (on top of cross-browser compatibility) for which we chose to create a bookmarklet first

c4urself• 14 years ago

Nice site, really nice UI with the arrow pointing to the bookmark bar. A change in the color of the tl;dr; bookmark when there's a bookmark available would really go a long way! Great job!

louischatriot17• 14 years ago

Thanks for the feedback! What you describe would be a big improvement but is not feasible with a bookmarklet. We are planning on releasing a browser extension that will enable just this.

makmanalp• 14 years ago

Great UI! One cool idea could be to have automated summarizer snippets already on the page, and let the user edit those, or suggest those to the user. This would result in so much more content.

louischatriot17• 14 years ago

Thanks! We did consider to show an automated summary of any webpage that was not already summarized by a human, but our tests showed that our users prefer to summarize from scratch than from a computer-generated summary that's usually of very low quality.

rahooligan• 14 years ago

This is really cool & solves a problem for me. I installed the bookmarklet and am looking forward to using it. I see that you are relying on the community to summarize pages & don't have a large enough data set built up yet.

What would be awesome is if you could somehow automate summarization using NLP. Check out condensr.com. They are able to summarize restaurant reviews...might not be at the same quality level as your current summaries but it came out of research from MIT & could be useful to you.

czzarr25• 14 years ago

Hi, thanks a lot for the feedback! we're glad that we're solving a problem for you! You're right that we have little content for now, but we don't think NLP is the answer.

Humans are so much better than algorithms at summarizing content and we really want to focus on quality rather than quantity for now. We want to become a platform for the content that people deem worthy of being summarized and are hopeful that the overall quality will therefore be astounding. As such we are focusing on building a community of excited contributors.

I'll definitely checkout condensr.com to see if it can help at all though!

martinpannier• 14 years ago

Interesting idea. Would love to know how you guys will tackle the tl;dr generation issue—how to incentivize people to contribute, etc.

This is definitely a very interesting space. We ARE moving towards bite-sized pieces of information anyways, and you guys seem to be riding on this trend.

Best of luck, and looking forward to how this will turn out!

louischatriot17• 14 years ago

We do too :) Thanks !

citricsquid• 14 years ago

Slight tangent, there was a kid (maybe 15?) a while back doing an automated content summariser for mobile and he was all over the tech sites for being incredible after getting an investment, but I can't for the life of me remember the name and I want to see his project again. Does anyone recall his name or the company? All I can remember is a chinese (maybe russian) firm invested in him and he was young.

edit: found it! "Summly", looks like their site is down :(

n1c2• 14 years ago

Seems to just be down because of an upgrade: https://twitter.com/Summly

Hates_• 14 years ago

13 weeks is a long time for an upgrade.

louischatriot17• 14 years ago

Yep, the kid is named Nick d'Aloisio. We think his take on the problem is interesting but we believe algorithmic summarization will not be able to produce summaries that are good enough.

He says he is currently working in stealth mode and will launch a product soon, and you can be sure we will be among the first to test it to see if our hypothesis is correct :)

eli• 14 years ago

If you want to play with automated summarization yourself, check out the libots library. It's neat to play with, but the quality of the results varies pretty widely.

n1c2• 14 years ago

This stings a bit because I've been [slowly] building basically the same thing!

Any particular reason you guys went with Express/Mongo?

czzarr25• 14 years ago

well we could talk about it! No particular reason, we're not developers/designers by trade, we have a generalist engineering background and we wanted to do a startup so we just went with what was buzzing at the time and learned everything from almost scratch. (could definitely be that it wasn't the smartest technical choices but since we've been handling the HN spike on our 20$/month linode VPS pretty handsomely so far we must be doing something right)

JumpCrisscross• 14 years ago

Send them your CV and GitHub!

bgrohman• 14 years ago

As a user, how do I know that the summary I'm reading is an accurate reflection of the article? Do you have any plans to validate the tl;dr entries? Have you thought about a rating system to allow users to rate the tl;dr entries?

czzarr25• 14 years ago

For now, you have to trust us ;) But yeah we obviously have things (votes, endorsements, verified badges and the likes) planned to ensure the tl;drs reflect the underlying article accurately. It is a first world problem for now though.

Doches• 14 years ago

I'd have read the description of the tool, thought 'eh, cool', and clicked away if it weren't for the absurdly cute instructions for installing the bookmarklet. Probably one of the best, tongue-in-cheek pieces of tutorial I've seen in a while.

Your tutorial made me smile! How rare is that?

czzarr25• 14 years ago

hehe glad we could make you smile! We put a lot of work into it. We thought it would be hard to convince people to install it otherwise as it is sort of a technical abstract conceptual thing for now so we added humor :)

RivieraKid• 14 years ago

This is great, I regularly wish something like this existed. Almost every interesting article / blog has a terrible signal to noise ratio.

zerostar072• 14 years ago

Good luck with the site. I had a similar idea, also implemented as a bookmarklet (here: http://news.ycombinator.com/item?id=2909309 which later mutated to http://noteplz.com/). Maybe the signup requirement is an overkill, wiki-style voting might work just as well.

czzarr25• 14 years ago

I don't know what wiki-style voting is but I'll look it up. Thanks for the feedback!

zerostar072• 14 years ago

I meant allow everyone to contribute and have an up/down voting system that also does not require login (based on cookies / ip maybe? dunno)

engtech• 14 years ago

It may be better implemented as a firefox/chrome extension?

Given that the majority of sites will not likely have a tl;dr I am unlikely to click the bookmarklet every time I open a web page.

czzarr25• 14 years ago

indeed, wait for it, it's coming soon!

ArekDymalski3• 14 years ago

Great. I prefer extension, as I have the bookmarks bar permanently hidden.

johndavi2• 14 years ago

We took a whack at this problem (agree it is big issue worth solving) last year with Streamliner, http://www.streamliner.co.

(Hint: ignore the video-specific text and paste in a blog post or article URL.)

Couple of examples: http://www.streamliner.co/s/S2C72QTZ/booster-camera-video-of... (video) and http://www.streamliner.co/r/eErbi/always-be-skeptical-of-nut... (text)

We rolled out with video initially because of text-stripping issues and the extreme low likelihood of users installing a bookmarklet/extension. Building a community of engaged trustworthy summarizers is also very hard.

Good luck!

czzarr25• 14 years ago

thanks!interesting take on the problem. Would love to talk to you about your experience if you're up for it!

johndavi2• 14 years ago

Sure thing, be glad to. Ping me whenever at john@<myusername>.com

secoif2• 14 years ago

Planning on creating API for this? I can see content aggregators being very interested in integrating.

czzarr25• 14 years ago

There is one, we're just not advertising it yet. If you're interested, please to get in touch with us. meta@<tldr>.io

switch332• 14 years ago

As an interesting idea; why not host a contest for summarizing wiki pages to promote growth and usage of the system? Also, I think besides your algorithmic ranking system, can't you also incorporate crowd-sourced ranking into the equation? Don't sell this business before it really grows. I really think that shortened topics with good information can be a lot more beneficial and even grow to be the new facebook or google sort of deal.

czzarr25• 14 years ago

Hi, this is a great idea and we're definitely going to give it some serious thought. If you have some experience with these kind of things, would you be interested in talking about it a little on Skype or by mail?

switch332• 14 years ago

As another idea you can also look at how Duolingo has developed since it is mainly crowd-sourced. As for the Wikipedia summation contest I would say do it by number of tl;dr's as well as quality and have some kind of cash award for users that post the most with best quality.

I'm no expert on any of this; just a college student with a interest in a lot of things. But growth is what will make or break this as a viable service. Duolingo is successful because it is an incentive to learn another language.

The incentive to summarize something that is difficult is there; but if people are not willing to read the whole thing in the first place then they probably aren't willing to summarize it. The problem is to solve this by providing at least some form of incentive or general interest in it. That requires some key thinking.

nashequilibrium• 14 years ago

I actually built something similar for my rss feeds and some bookmarks using the python "Clips" library but i felt that it needed to be a feature in a broader product, so i decided against trying to build a business with it. The difference with mine is that the algo automatically summarized the content for you.

louischatriot17• 14 years ago

That's one way to approach this problem, and there are a couple companies trying it, most notably Summly. We believe that software-made summaries are not good enough, which is why we down the human-written summaries road!

thebigshane• 14 years ago

Ok, now I wait until someone writes another extension/bookmarklet that automatically inserts the tl;drs (if any) into the HN front page.

czzarr25• 14 years ago

bear with us, we're working on that!

_xhok• 14 years ago

I think the idea's really neat, but at the same time I'm concerned about the effect that things like this have on the attitude of our culture. A large part of critical reading is being able to take the time to read an article/passage in-depth and judge its ideas for yourself.

From a business standpoint, are there maybe certain types of content that just aren't viable to "summarize?" Would it be better to tailor your service towards certain other types of content? Just my thoughts.

czzarr25• 14 years ago

Our goal is to help people find and select what they really want to spend time reading. We feel that a summary is the perfect step between a title and the actual content and that it actually incentivizes people to read the whole thing once they're hooked by the summary.

The people that won't read the full article after reading the summary probably would have just bounced or skimmed the article extremely quickly.

You're right that there are probably certain types of contents that lend themselves to be summarized better than others. We're still figuring this one out by observing what our users and experimenting with what we summarize.

olliesaunders• 14 years ago

This doesn’t actually refute your concerns but we’re also in a time with more amateur authors than ever with a lot of them unable or disinterested in making their points succinctly.

Angelo8000• 14 years ago

Yeah, my sentiment exactly.. I don't want to have to go to another website. I'm lazy.

stevewilhelm• 14 years ago

Pinterest is super easy. tl;dr sounds like work.

louischatriot17• 14 years ago

I don't think the two can be compared, since we're really focused on summarizing content. But you're right: summarizing content is a lot of work, it takes a couple minutes to create the summary of an article of average length. That is why summaries have so much value, and also why this problem has yet to be cracked.

rooshdi• 14 years ago

I was starting to work on almost this exact same tldr concept recently but decided to shift to another project instead. It's great to see another group of people tackling the signal to noise issue. It's definitely a growing problem that needs to be solved. Great start thus far and I'm excited to see where you take this!

czzarr25• 14 years ago

thank you!

erenemre• 14 years ago

Best. Bookmarklet. Tutorial. Ever.

Thank you.

I, too, needed something like this forever. Every time I see a long article, I complain about the lack of a tl;dr.

Unfortunately, I am not so optimistic about humans creating the content. You guys have a really, really hard job. Creating the right tools is easy. Convincing people to create tl;dr's is hard.

I'm really curious about how you are planning to solve this problem.

I really do hope that you'll succeed though : ]

Edit: Also, for me, this tool is great for articles that are new. Assuming you don't have enough contributors at the moment, after seeing a few articles that don't have tl;dr, I probably use the tool less and less everyday. Until I forget about it. I'm sure I am not the only one.

louischatriot17• 14 years ago

Thanks we worked a lot on the bookmarklet tutorial :)

You're completely right, the real challenge starts now, we need to find a way to get humans to contribute, because right now you have too much probability of getting a "summary not created yet" message when clicking on the bookmarklet. We already have some ideas we want to explore, if you do have some please tell us it would help a lot!

eranation2• 14 years ago

Any ideas how to make money out of it? I thought of a similar idea after developing an addiction to reading every article on HN in fear I'm missing something that will change the way I do things (sadly, every other day, I find something that justifies it, this one could be an example if it works out)

But my thought was that I will gladly pay someone 10$ a month, to scan through HN instead of me, and send me summaries of things that I'm interested in (e.g. web frameworks, JavaScript, Scala, CSS3/HTML5 etc)

Here is a feature that Wikipedia can't have for ethical reasons, and you might be able to get away with:

- user clicks on bookmarklet (hopefully a chrome / ff extension soon)

- no tl;dr

- user has a button saying "ask someone to summarize it"

- now you either hire some good shorthand writer, or crowd source it and cut a percentage of the transaction

another way is to say - the first 3 hours of an article's peek (e.g. you do it actively for each front page article in HN for a start) you charge 1$ to see the tl;dr version of it , after 3 hours, it's free for all.

I think a lot of people here would love not having to read everything just to find the point or understand if this is really relevant

I think you have a huge market here, saving time is a very important service nowadays, especially when there is so much good content wrapped with a lot of needless, but mandatory decorative language.

czzarr25• 14 years ago

Hi, these are really great ideas, thank you.

We've thought about it more in terms of volume of consumption for now but are going to look into the users' patterns to see what fits them best. We also thought about paid mobile apps, or plugins for content creators if we can prove them it drives more traffic to their content.

ArekDymalski3• 14 years ago

I think you should first focus on building big-enough community of people who'll create high quality, non-biased, not-spammy summaries. Without it you won't have anything to monetize. Eranation's ideas were interesting (in the area of financial incentives you could experiment with Mechanical Turk, maybe GitTip, MinuteBox etc. as well as BitCoin payments). However you also should check for other, non-financial motivators for creators. I guess that many bloggers and other people trying to build their online brand would be interested in it for clicks, social shares, Whuffies, Klout score impact etc. The another option worth testing could be some kind of intelligent gamification of the process. In that case browser extension would be handy as it could provide all the necessary tools like author profile, sharing buttons and so long. Good luck - I really hope to see your button everywhere soon, as it's great idea.

nancyhua• 14 years ago

Tldr. Where's the version of this that jacks straight into my brain matrix-style?

louischatriot17• 14 years ago

Ha ha that would be great indeed. Maybe the next evolution of the Google glass ?

bbayer• 14 years ago

which summarization engine did you use? Also how accurate is your boiler plate removal algorithm? Can you point me open tools that you use?

czzarr25• 14 years ago

the human brain :)

lhnz• 14 years ago

Very smart move on sending emails to people who clicked for summaries and didn't get them. (But in which they are later created.)

czzarr25• 14 years ago

we thought you'd like that :)

Show HN: Summaries of interesting content, written by people

Related Stories