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?
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.
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
I bounced without trying it because I couldn't figure out what it was going to actually give me. I would have been more likely to sign up or drag the button to my toolbar if you gave me a sample of what value I would get out of it.
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!
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.
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.
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.
Hi Arek,
This is exactly what we think as well. Without an engaged community of quality contributors we won't go anywhere and this is our only priority.
I love your ideas, thank you for sharing them. One problem we face with convincing bloggers is the mental barrier of them thinking that people are just going to read the tl;dr and not the whole content. We believe that the reverse is actually more likely and are starting to aggregate data to back this claim up.
If you are up to discussing intelligent gamification with us, I'd be delighted to have a chat with you. Contact me on twitter @stanmarion or by email stan@<tldr>io
Thanks for the kind words
Gamification, if done right, can be more powerful than money, HN, SO, are a little addictive, we all know that... but not sure how it is translated to money. Qell, if money was everything, not sure if we had Wikipedia, right?
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 :(
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 :)
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.
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
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.
Yep, that is definitely planned. Especially a karma-like system to reward users who work. Since you're helping people by providing tldrs, we are thinking about the count of people who read tldrs you created, meaning the number of people you helped.
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.
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.
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.
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!
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.
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.
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!
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!
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.
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.
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!
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.
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.
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.
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.
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?
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.
yes you're absolutely right. We think summarizing content is useful for one person because it help sink knowledge in and really understand what one is reading. The fact that it can help thousands of people save time and browse more efficiently on the Internet is the cherry on top.
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.
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 !
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.
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).
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!
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?
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.
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.
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.
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.
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
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!
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)
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.
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!
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.
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 :)
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.
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.
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 ?
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?