One particular problem (or opportunity) with political summaries are ideological leanings. I'd be interested in summaries from different viewpoints, e.g. a social democratic reading of the WaPo Sanders article would be significantly less favorable.
I guess that might look like a browser extensions where individual users give their own short summaries of an article, you can see the various summaries of a given article, and you can see what other people are reading and have summarized. Sounds a lot like Twitter in my mind, but with the additional cross-referencing by article URL.
This is a great idea, thanks for sharing. Now that I think about it, most of our early adopters (people who read a lot) would be present on desktop and for such users a community driven browser extension would be a great MVP.
Thanks for your support and interest. This is could one of pg's "do things that don't scale" kind of oppurtunity.
The idea is that when you have concentration of users with somewhat similar reading habits (like the HN community), economies of scale kick in where the cost of analyzing each article drops to pretty much nothing/view and access to summaries is near instantaneous in 99.9% of the cases.
As a personal assistant type service it would be limited to a very small niche of very wealthy/no time users. I'm very excited by all the support this is getting and am going to get started on a MVP soon. You'll be one of the first people I'd reach out to :)
What exactly do you need to build that isn't already available?
Somebody else is going to start emailing me summaries for $100 a month while you're messing around with whatever it is you think you need to build first. :)
I will reach out to you shortly and we'll work something out. :)
What exactly do you need to build that isn't already available?
A couple I can think of:
1) A community of people with diverse interests who can analyze articles across topics. (I can't summarize an article on a quantum physics theory for example)
2) I reasonable number of consumers who can bring down the cost/consumer.
I have had some feedback and time to evolve the idea and have come to realize that there is a power law of online content consumption (something like 10% of the content gets 90 percent of the eyeballs).
So the way to kickstart this idea would be to identify the most popular content (ideally before it starts trending) and have it analyzed by a real human who could be paid hourly. Niche topics like high tech should be left to the community to summarize (wikipedia style).
I've been thinking to the same (core) idea recently, but was not planning to launch it on hackernews but on my friends. I'll be glad to talk with you about that :)
If you want to, please reach out at dumon.david at gmail dot com.
For someone just coming here, TL;DR version of my idea was: Some people are busy and have very little time for casual reading, having a fair idea of what an article is about and what to expect will enable them to spend their precious time reading quality pieces which will really enhance their knowledge (Kind of like having an Amazon reviews and human written summary for every popular article on the web).
The Summarize service on OS X does a decent job at summarizing content, and it integrates will any app that uses Text Services. On the article about Apple's recycling[1], the summary it produced was similar to the one you produced.[2] One advantage with the OS X Summary is if something looks interesting, the slider can be moved to reduce the aggressivity of the summary, thus yielding more details.
From WiseGuy:
Various media sites claimed apple recycled $40 million worth of gold from iphones, they were dead wrong.
What actually happened is that Apple is under statutury obligation to recycle a certain weight of e-waste depending on it market share or weight of electronics sold (depending on state laws). The e-waste doesn't have to be of the manufacturers own products.
Apple paid third party recyclers to recycle mostly CRT's and PC's (iphone have hardly any gold and are much more valuable refurbished, In fact, phones and tablets often don’t count toward the overall recycling requirements in many state laws.) and probably incured a loss rather than a 40 million windfall as claimed in news articles.
From OS X Summary:
The most egregious and inaccurate storyline goes something like this: Apple, out of the goodness of its heart or perhaps fueled by monetary incentives, took old iPhones and iPads that were brought back into its stores, took them apart, melted down the roughly 30 milligrams of gold in each phone, and ended up with 2,204 total pounds of gold.
...More commonly, however, it was reported that Apple recycled “90 million pounds of e-waste through its recycling programs,” with much of this e-waste being iPhones and other old Apple products.
...Linnell told me that, generally, Apple is considered to be a good faith participant in many states’ recycling programs, and said that the company has also set up recycling programs and contracted with e-waste recyclers in states where there are no e-waste laws.
Edit: I should add that anything that helps people get more out of their day is somthing that can be inserted in the time vs. money equation, so your idea is definitely on the right track.
I have tried several summarizing services but none does as good a job as a real human. Try to take some article that you haven't read and run it through an algorithmic summarizer, most of the time it wil be a very poor experience, specially so when compared to say a friend explaining it to you in short. Thats what wise guy is. I don't believe in the forseeable future AI will be able to beat humans at this but if it gets close it'll only further improve our unit economics.
Currently our unit economics on staff analysers look something like this WITHOUT any tech assistance:
Median wage in low-income english speaking country: $2/hour
Analysis time for a average length 1000 word article: 6 minutes
Actual cost/article: 20 cents
Average Views per summarry at scale: 20 (Extremply conservative estimate, the Apple summary got 1000 views in an less than an hour)
As you can see the average cost of Human summarizing/view is less than a cent. Such a service at such a low price point should be seen as valuable by a large number of people; or at the very least sustained by advertisment (though I would very much prefer a subscription or micropayment based business model)
I estimated from the number of people who clicked through the link in the post and further to the medium post on the headline links to. The number would be very close to 1000 at the very least, but most likely several thousand.
Is your Pocket reading list too long for you to keep up with?
You find articles/news to read from multiple sources (twitter, facebook, rss, media outlets, Hacker News, Reddit etc etc.) but can't find time to read them all?
Are you irritated by all the clickbait headlines and manipulative snippets and wish you had a fair idea of what you are getting into before reading a piece?
WiseGuy (placeholder) aims to solve your problems.
There is a small (maybe large?) consumer segment that loves to read stuff and is hungry to absorb as much knowledge as possible. I have a morning and before bed ritual of going through my favourite news/article sources (which includes HN) and just reading stuff, but of late I have started feeling overwhelmed by my "addiction". Clickbaits and manipulative snippets designed to make me visit the site only to later find the piece pointless are the most irritating. I'm sure many on HN would relate.
WiseGuy will be an app that I wish someone made for me, but no one has yet so I will solve my own problem + maybe millions of other people.
How it works is simple:
1. You find a piece that you would normally read.
2. Share it to the app just like you would "Save to Pocket".
3. A real human will read and craft a summary of the piece for you and tell you if you should invest time to read it yourself. Some articles will be insightful, include deep analysis that should be read in detail to be properly absorbed, while others could be summarised in 2 sentences to tell you what you need to know. WiseGuy will do it for you.
Article is based on a study by researchers on what is termed "value deflation" of solar energy. In a free market where the marginal cost of producing solar electricity is zero, the value of solar energy during the day (the panels most productive time) will deflate as more and more panels come online. This will happen because there will be a LOT more electricity available at say, noon, than there is demand for, leading to a huge drop in wholesale prices. Batteries can help but not completely solve this problem. Researchers calculate that the price needs to come down to $0.25 per watt by 2050 — down from around $3 per watt today. We are not on track for that given current pace of incremental improvements. Value deflation could be the difference between solar staying as niche technology in the grand scheme of things or it taking over the world.
Original Article: 2700 words (~20 minute reading time)
WiseGuy Summary: 148 Words (less than a minute)
WiseGuy Analysis: The article is very well researched and goes into deep details. It references to several other studies and analyses the situation, solutions and future in detail. It includes useful graphical data. Reading of the full article is highly recommended if the summary and the subject intrigues you. Rating: 4.5 Stars (maybe crowd sourced?)
Piece questions how is it that over months and months the average donation has stayed at 27 dollar. It gives out a few possible reasons and then proceeds to calculate what the latest average is. First, averages involving millions of numbers change slowly. Second, the campaign encourages a $27 donation explicitly (with a button on its fundraising page) and implicitly (by making it a point of pride). And, third, it actually does waver from that point. It finally concludes based on public data that the actual average should be 27.88, 88 cents more than claimed.
Original Article: 800 words (~5 minute reading time)
WiseGuy Summary: 96 Words (less than a 30 seconds)
WiseGuy Analysis: Clickbait headline, not much new information. WiseGuy summary is sufficient to explain the piece, feel free to skip it. Rating 2/5
----------------------------------------------
How is WiseGuy better than alternatives? Well for one it is powered by real humans. It is much better than NLP based summarizers which are not even close in quality, they are actually terrible. Also while there are sources where one can get curated human summarized content, they actually only summarize news and not any random article/blog that you want right now. WiseGuy is for whatever you want summarized, whenever you want it! You can send your entire Pocket Reading list with a 100 saved articles to Wiseguy and have all of them analysed on demand.
Business Model- It will work as an ad supported community where the analysis will be provided by members of the community who have actually read the piece. If possible ad revenue will be shared with top contributors based on quality.
There will also be a subscription based plan which will remove ads and will guarantee an analysis from our staff summarizers in case no one from the community take it up within a reasonable time. I have run numbers and the economics work.
Does this have the potential to be a unicorn? I would assume not but so did the founders of AirBnB think of their company.
I can't say how big the market for this service will be but I do know that it will solve the problems of a large number of people. A future where all consumers of online content will first run it by WiseGuy, or make it the primary source of discovering useful community rated content is not unimaginable at which point this could be a 10 billion dollar business.
One study I came across that made me realize this might be more than just a niche problem: https://www.washingtonpost.com/news/the-fix/wp/2014/03/19/am...
Besides I have some ideas of how such a basic service might evolve into something that solves other related problems as well.
How will I use 20,000 dollars? Firstly where I am from adjusted for Purchasing Power Parity, $20,000 dollars will be worth $200,000 so I will be able to do much more with it than a founder in Bay Area.
I plan to use it to pay my living expenses while I take a semester off at grad school, build an initial team, site expenses and paying staff summarizers to solve the chicken and egg problem of building a community. I'd work out of my apartment so there will be no office expenses.
Best Case WiseGuy would be able to save millions of man hours, help hundreds of thousands read quality stuff and avoid the rest, absorb more knowledge from the limited time they can spend on casual reading and pay thousands of people to basically.....read.
BTW if HN is going to help us get into Ycombinator Fellowship then it seems proper that it also helps us come up with a name (or keep the current one), so PLEASE suggest away!
Also I will be analysing some posts on HN front page every day to demonstrate how WiseGuy might work. Look out for them. Feel free to share articles you want analysed in this thread too, I am just one guy but I’ll do my best!
Don’t worry too much about being nice. I am going to give this a shot even if I don’t get through YCF so I appreciate any suggestions/criticism, especially from HN community as most of you would be the ideal target market and early adopters.
Are you irritated by all the clickbait headlines and manipulative snippets and wish you had a fair idea of what you are getting into before reading a piece?
Might I suggest using your product to come up with a better headline for your submission? :P
Its in most content publishers interest to have headlines which give out enough information to make one curious but not so much that it make clicking through unneccesary.
As a reader with limited time it is in your interest to get as much info in as few sentnces as possible (best case, just from the headline).
Headline: "Trump Calls megan kelly bimbo" okay so now I know enough and don't need to actually readd a 800 word article. No need to click through; as a reader this is what I want.
Headline: "You won't believe what donald trump just called megan kelly": now I'm just the right amount of curious, i'll click through, read the artcile and then wonder why couldn't I just get it in one line. But the publishers gets the adviews. Thats what he wants.
This is a situatuion where inerests of the consumer and business are mis-aligned that's what WiseGuy is trying to fix :)
It was in my interest to have a headline that'll make HN users clickthrough (but not an inaccurate one) and read, so I did exactly that.Hope you get me now and understand how WiseGuy is solving a real problem. :)
"The person reading it will like it!" is an age-old justification for spammy advertisements for products that cannot market on intrinsic merit. It does not automatically make it not-spammy or non-baity.
TL;DR version of my idea: Some people are busy and have very little time for casual reading, having a fair idea of what an article is about and what to expect will enable them to spend their precious time reading quality pieces which will really enhance their knowledge (Kind of like having an Amazon reviews and human written summary for every article on the web)
The signal to noise ratio on HN could be massively improved if the site implemented a simple filter which disallowed resubmission of duplicate URLS. I browse the site by the "New" tab and, when a popular news story breaks, it's not unusual to see the same link posted half a dozen times, or more.
Personally, I'd find a filter which removed all submissions with the word "Awesome" in the title, or any sites which used those annoying JavaScript overlays on top of what you're trying to read, a great help, too!
As regards your service, good luck with it, but I don't think it's for me. It sounds like it could be a lot of work for you to keep it up and I'd be a bit dubious as to just how much reading and comprehension time your summarisers could devote to [especially] complex or nuanced articles.
I'm also reluctant to put too much faith in someone else's interpretation as to whether an article is good, unbiased or [to quote your example] "well-researched", etc. Those things are all "in the eye of the beholder" and one man's well-researched article might be another man's blatant propaganda piece. I'd rather form my own opinions, by going to the source, than play Chinese Whispers.
The point of this service is not to replace actual self-reading; but rather - Give people some idea of what the article is actually about so they can decide if they can/should invest time to read it.
Most people don't have time to do all the reading they want to do and having more info than just a clickbait headline will help them allocate their time in a more productive way.
Example this is how Vox.com is showing new articles on its "new" page:
"Big Marijuana is coming — and even legalization supporters are worried" (An activist explains why he left the movement."
The headline and snippet gives you no information what the article is about. You have to actually read 2500 words over 50 paragraphs to know what it is about, in the end you might like it, or be dissapointed, its a gamble.
But by reading a well written analysis you can have a fair idea of what you are getting into before you invest 15-20 minutes of your time into it.
Do this for 10 articles a day and 300 a month, the time you save adds up quickly.
For someone just coming here, TL;DR version of my idea was: Some people are busy and have very little time for casual reading, having a fair idea of what an article is about and what to expect will enable them to spend their precious time reading quality pieces which will really enhance their knowledge (Kind of like having an Amazon reviews and human written summary for every popular article on the web).
I guess that might look like a browser extensions where individual users give their own short summaries of an article, you can see the various summaries of a given article, and you can see what other people are reading and have summarized. Sounds a lot like Twitter in my mind, but with the additional cross-referencing by article URL.