Today, news and information comes at a furious pace. In the wake of it, it’s become an art and a science akin to monitor this deluge of information, to see when something is about to break, and to know if you can trust it.
Storyful has become a leading expert Real-time Discovery—that is, the lightning-fast-paced work of monitoring and verifying the real-time web. Their 200-person global team helps news organizations and brands stay on top of current events as they unfurl.
“We discover and verify the content from social media using our own technology and open source technology [editor’s note: including feedly!], monitoring the social web in real time,” explained Derek Bowler, Storyful senior journalist and special projects lead, who also helps lead the company’s internal work flows, processes, and tools.
Storyful’s ability to work together across timezones and continents is central to the value that they create. They have global offices in Ireland, Hong Kong, Australia, and New York, and each team works together in real time. “Collaboration is at the core of Storyful,” says Bowler.
Organize what you are monitoring into feedly Collection.
Storyful creates a feedly Collection for every story they monitor like 2016 Decision, funny videos, cat videos, ISIS, and more. It’s an easy way for them to follow multiple sources on the same topic in one place. And when they seem a Collection updating with many new articles, it often means that a new story might be breaking.
Create a diverse mix of sources with your Collections
When Storyful creates a topic to monitor, they carefully hand pick sources that include as many known YouTube accounts from that particular location, Facebook feeds from active posters, key Twitter accounts, and any relevant sub-reddits. They ensure that they have at least one feed from each channel, often many more.
“That’s a one-stop shop because a lot of things we see happening in social media are encompassed in those channels,” says Bowler. “We knew a year ago that if we were monitoring those four major social platforms effectively, we were not able to monitor the topic effectively. The best thing about feedly is that it allows you to bring it all to one place.”
When Storyful editors start to see some feeds updating with increasing velocity, they know that something big is breaking.
Create an archive
One way Storyful uses feedly is a bit unconventional: They use it as a YouTube archive that is easy for them to search through. They have over a thousand YouTube videos that they monitor. By connecting the YouTube feed to their feedly, it becomes easy for them see what is breaking, but also use search terms to find a relevant video.
Connect feedly with other open source tools
“There are a lot of open source tools that you can combine with feedly to create a really powerful discovery tool for discovery desks to minimize their workflow,” Bowler says. “I no longer see feedly as an RSS reader.”
In particular, Storyful likes to use:
FB-RSS – This tool creates feeds from Facebook pages.
IFTTT + Slack – Storyful relies on Slack for their team communication. So, they create Google Alerts that they import into feedly. And from feedly, they use IFTTT to push breaking articles into their Slack.
What do you use to monitor every day news?
Are their tools, tips, or tricks that you or your organization use to be the first to know something? Share them with us!
Over the last few weeks, we have begun publishing a content marketers series to help content creators thrive. This week’s installment on content distribution applies to anyone looking to reach more people with what they create online.
If a tree falls in a forest and no one hears it, does it make a sound? If you create a piece of content, but no one reads it, does it exist?
Despite investing time, money, and sweat into creating the content, driving readers to your content can be just as difficult. Whether you are a content marketer, a blogger, or a big publisher, this has becoming increasingly difficult in an accelerating world of online content and biased social feeds. In our “State of Content Marketing” report, one in five marketers reported distribution as a top challenge.
We went to three companies with thriving content marketing strategies—Buffer, Help Scout, and InVision—and asked them about their distribution strategies.
Here are seven ways to distribute your content.
1. Social
Despite recent conversations that it has become increasingly important to earn attention on social media because of the changing algorithms that encourage brands and publishers to use paid social ads, brands are continuing to use social as a primary channel for content distribution. “Social traffic is definitely down for us,” says Kevan Lee, content crafter at Buffer. “At the same time, it’s our second biggest referrer source. Still we get about 100k or so a month from social. That one is definitely significant.”
2. SEO
Sure, SEO is a long play that can take months to deliver benefits. But over time it can become the gift that keeps on giving. Forward-thinking blogs still report that SEO is key to their audience traffic. For some, this is a deliberate strategy that they have invest in over time. For others, it has been a product of a bigger commitment to quality content.
Buffer has seen tons of traffic originating from search. “Having written the longer form content and writing it focused on specific topics has been a good strategy—though I’m not sure if I’d call it a strategy,” says Lee. “I’m not sure if we set out to do it necessarily but it ended up working out that way.”
3. Social Ads
In recent months, the feed algorithms at major social networks have continued to morph, making it harder for business content to stand out without the help of paid social ads. It takes some experimenting with your content and audience—and some mulah—but social ads is another way to increase the reach of your content.
InVision spends about $4000 per month on content promotion various social media channels, including Linkedin, Facebook, Twitter, and other industry-specific outlets like Dribbble, according to head of content at InVision, Clair Byrd. “It’s not a lot of money,” she says, “but it’s enough to understand what content works best and it helps us create more things that work best.
4. Email
Growing an email list for content is a great way to ensure that the people who are interested in your content are getting it sent straight to their email inbox.
Buffer, for instance, has a list of 45k which it uses for distribution.
InVision says that email is still its most powerful channel and does one overall content email a week—a piece of content that performs strongly. “They’re free, and they tell us a lot about what’s working,” says Byrd.
InVision also send out a stand alone email for key content initiatives, depending on how bit the impact is for the company.
“We run content like people run product,” says Byrd. “Everything is campaign based, and everything we can tear apart. So if we think that a release is going to be Tier 1, we will support the content just like the Tier 1 product.
5. Partnerships
Syndication partnerships with other blogs or publications are another way to engage a large audience that goes beyond your user population. “One of our main distribution channels that runs almost automatically is our partnership with the Huffington Post,” says Gregory Ciotti, content strategist at Help Scout. “We set up agreements with them and Business Insider. They’ll handpick something they want to run. We just require that they use real canonical tags to protect our search. They offer us a small byline that links back to HelpScout. So all of those are happening automatically. People will overestimate how much traffic is sends back, but either way, it is helpful. Syndication is definitely fantastic.”
6. Content Submission Sites
There are many sites for communities to post interesting and relevant content. Dribbble, Quibb, and Reddit are just a few examples. For some content distribution strategies, it may make sense to participate in the conversations at these sites and to submit content in a way that helps the community.
The key, of course, is to respect the site for what it is—a community—and to avoid spamming by truly becoming a part of it and taking part in the conversations.
7. Sponsored Content
Many brands are increasingly using sponsored content services, or native ad platforms, like Outbrain and Taboola. These services try to reach more people placing your content within online publications that reach a relevant audience or post about similar things. In addition to distribution, it can provide a method of testing the efficacy or your content among a known persona or help you explore what persona reacts well to your content.
These are our experts’ six recommendations. What do you use to reach more people with your content?
Some people are calling it Content Shock or Content Clutter. We like to call it the Content Renaissance.
Whatever you call it, many of us have been talking about the same idea: Content is coming to us at an ever increasing rate.
The result is that ideas, trends, and memes have become a swiftly moving current that will easily overwhelm us or leave us behind if we let it.
The Content Renaissance is creating an explosion of content.
We live in an unprecedented age in which we are creating more content than ever before and in which we have more access to information and ideas than ever.
Ninety percent of the world’s data has been generated in the last two years, according to one study. Google publishes 20 petabytes of information every day, according to Promodo in 2013. To put it in perspective, there have been 5000 petabytes of information created from the dawn of civilization to 2003. In 2014, WordPress reported that it was publishing 17 posts every second—or 1.5 million posts per day. In that year alone, 72 million websites were created.
And the production of content is still growing. Ninety-two percent of marketers are creating more digital content now than they did two years ago, and 83 percent expect this number to continue to rise, according to Accenture.
If this isn’t overwhelming enough, social networks are accelerating this flow of information still faster. According to Domo, every minute:
Facebook users share almost 2.5 million posts
Twitter users tweet almost 300,000 times
Instagrammers upload almost 220,000 photos
This influx of content carries plenty of upsides for individuals and publishers alike. It’s easier than ever to share your gospel. Creatives thrive in this new medium, and businesses are creating engaging new experiences.
However as consumers of this content and as merchants of ideas, the firehose is overwhelming as it is satisfying.
Enter Trend Acceleration
The result of the Content Renaissance is that content has become the currency of change. Ideas are being exchanged, embraced, and evolved at an ever increasing rate.
As a result, while trends used to come and go over the span of years, these days discussions are moving thought leadership at a cadence of months or weeks.
This deeply impacts the way we do business and the way we operate in this world.
For marketers, in particular, it means that the ability to engage consumers with fresh and relevant points of view is a swiftly moving window. The current of new ideas is so fast, that it is easier than ever to attach your brand to something considered more passe.
For PR people, it means a bigger challenge in cutting through the noise with something sharply unique.
For corporations, it means creating products that serve customers in a quickly changing competitive landscape. Small businesses have to move just as quickly with often times fewer resources.
So how do you stay ahead of it?
The key is to identify the right signals and then implement the right workflow to monitor those signals.
1.Recognize the need to monitor for yourself and your company, and dedicate time or people resources to the task.
Monitoring is an art, science, and a process, and above all, it takes time and attention. Recognize that with your daily work or within your business and be deliberate about setting aside time for yourself or people resources within your business to do it well. Have you set aside some time for yourself to crunch through these trends? Do you have the right people on the job? Do you have the right tools in place?
2. Identify the right signals — create the right mix of places and people to follow.
Many algorithms today try to guess what is important to us by predicting future interests based on past behavior. Staying on the cusp, however, requires a proactive, but efficient, way of identifying what the right signals are. A signal is that place from which new ideas and new trends are broadcast.
Identify the most important pieces of news or thinking on the new trend and use those to develop a mix based on:
Key influencers and thought leaders – Who are the main voices in the articles? Who are the main people they cite? Who are other thinkers in that realm?
Key producers in Twitter, YouTube, Facebook, and Reddit – Real-time channels can indicate the earliest rumblings of a trend. Find the right people in each of these social networks and add them to your list. Storyful, a company that specializes in newsgathering from social networks, recommends finding relevant content from each of these social channels to get a full picture.
Specific Geographies to keep a pulse on – Is your main industry magazines in New York? Is it a new irrigation system being tried in Venice? Is it human trafficking legislation in San Diego? Consider some local sources in your content mix.
Power key words – Are there buzzwords that show up within the trend you are monitoring? If you are following the tea party movement in America, should you be following any mentions of “Sarah Palin,” “immigration,” “gun ownership”?
Here are some great Collections that other people have created to follow the latest trends in a given industry:
Sarah Levine and the most important publications in Maps.
3. Centralize these signals into the fewest number of places that allow you to follow these people, publications, and trends most efficiently.
Much of today’s content is on different platforms and requires you to go to it. Flip this around and have this content come to you by creating the right workflow. Focusing in on the workflow will allow you to crunch through the onslaught of content quickly and methodically. We’re a bit biased, but naturally we recommend feedly as a place for you to house all the blogs, publications, Google Alerts, YouTube feeds, Facebook feeds, and Twitter feeds you are monitoring.
With feedly, you can create a separate Collection for each specific trend, vertical, product, or industry that you are monitoring. With feedly Pro, you can even use power search to search for specific terms within a Collection, organized by media type or date.
4. Archive the data points.
Create a methodology for saving and organizing the precious content pieces you find online that speak to the topic you are following. Many people like to use Evernote or Pocket to save articles. We highly recommend using tags within feedly to save and organize the great pieces of content you want to share with others or return to later.
5. More eyes, better vision: Crowdsource monitoring
The more people who are collaborating together the better a pulse you will have on trends and the richer your conversations will be around them. Create an internal system through which all employees can contribute important content on new trends and key teams can ingest this information in an organized fashion.
6. Broadcast your interest and join the conversation
The more you can show your interest in a particular conversation the easier it will be to have conversations with the right influencers. Consider taking the content you are monitoring and broadcasting your interest in the vertical via social channels.
Here are some starter thoughts for how to stay on top of the fast pace of trends.
What are ways that you and your company have found effective in finding the right signals or organizing them into workflows that are easy to follow?
feedly is your newsfeed for work. It allows you to become more powerful in reading, sharing, monitoring, and collaborating on content. Learn more about feedly and going Pro or Team here.
A few years ago Marc Andreessen famously proclaimed “Software is eating the world.” I have to think he was speaking figuratively, but with the rapid ascent of machine learning one has to wonder. Science fiction writers and movie producers have often envisioned a dystopian future, but perhaps more troubling are real life warnings from Stephen Hawking and Elon Musk. Most people take a positive view, though, and it’s hard to argue against the potential benefits of now seemingly inevitable applications such as self driving cars. Reasonable people can take either side of the argument, but it is impossible to argue that progress will stop: like most technology, it may have unforeseen benefits and consequences but its advance is inexorable.
Is this guy going to enslave the human race or make us banana pancakes for breakfast every morning?
So whether you want to learn more about the state of machine learning, how it works, or simply wish to be better versed when greeting our future robot overlords, read on!
Terminology
There are a few different terms that get bandied about in articles about this topic: machine learning, artificial, deep learning, etc. They all mean something slightly different, but I generally use deep learning since it makes me look the smartest. Here in Silicon Valley, looking smart is almost as important as having “hacker” in your job title. So now that that’s settled, what is deep learning?
First and foremost, it’s an excellent marketing term! Computer scientists are generally terrible at naming things, so when we get something right, people perk up (see also: cloud computing).
As a CEO, what sounds better “deep learning” or “back office ERP systems”? As a journalist would you rather write about “deep learning” or “software automation”? As a student, would you rather study “deep learning” or “artificial multi-layer neural networks”? You get the point.
And by the way, the last example was a trick question, deep learning is an umbrella term for various types of multi-layer artificial neural networks. But whereas the latter term evokes feelings of a post Thanksgiving, eyes glazed over torpor, the former gets you a front page article in the New York Times.
Style and Substance
Usain may win on swagga, but even he can’t match deep learning’s performance.
While deep learning is a great marketing term, it’s also delivering amazing results. For example, in the field of computer vision, there is an annual competition where top universities and research labs all try to outperform each other. Deep learning techniques were first used by the University of Toronto in 2012. How’d they do? Well, as you probably guessed they won. But they not only won, they were 70 percent better than anything else. As a comparison, when Usain Bolt famously blew away the field in the 100m dash, he won by an incredible two tenths of a second. That’s only 2% better than second place.
True Artificial Intelligence
But it’s not just the results that make deep learning groundbreaking. Deep learning approaches things in a fundamentally different manner.
Previous state-of-the-art algorithms heavily relied on “feature engineering.” Essentially, researchers developed an incredibly deep understanding of the problem at hand and teased out the key characteristics need to solve it. Then the algorithm would simply learn how to best balance or “optimize” these characteristics.
On the other hand, deep learning simply says, “Back off, humans! Just give me all the data and let me do my jam.” The algorithms actually manage to do the feature extraction part of the process as well as the “optimization” part all on its own.
Think about this for a moment. Usually the clever parts of algorithms are discovered by people and computers are used to mechanically calculate things many times faster than we possibly could. But deep learning directly consumes raw data and independently reasons about how to best frame the problem and solve it.
This is a real step forward towards “true” AI. Consider where the intelligence lies in each approach. Sure, the previous optimization algorithms may be clever, but the real magic in the process was in humans reasoning about the data and extracting the right properties. With deep learning, this entire process is done without human intervention. This is much closer to what most would consider human thought.
Computers Playing Games
This transfer of responsibility makes it much easier and quicker to address many different problems. In software, one of the most important steps in solving any problem is modeling it correctly. Sometimes, this is really easy. Take chess as an example:
While we humans see this as a single image, computers need a numerical representation. Here it’s easy. There are 64 squares on a chessboard, and six types of pieces. We can use a list of 64 numbers to represent a particular board set up. Each number will simply identify the piece in that square.
This is elegant in its simplicity. And indeed, chess has been a popular target for computers. Many students (myself included) program a very simple chess playing algorithm as part of an introductory computer science course. At the other end of the spectrum, researchers have built optimized algorithms capable of beating the best players in the world without deep learning techniques. This is bread and butter for pre deep learning algorithms, a pretty simple modeling “space” and a huge search “space” where computers can win by simply evaluating more scenarios than humans can.
Even our fondest childhood memories aren’t safe from deep learning.
Now consider a very different game, Super Mario World. Take a look at the screen capture. How would we go about modeling this? While I get the warm fuzzies thinking about all the hours I played this game as a kid, thinking about how to model this for a computer gives me a headache. Mario and his friends can be anywhere on the screen, how do we even represent where Mario is? He’s not a single point or shape, he’s this blobby thing.
Maybe we’d have to preprocess an image to extract out Mario, turtles, blocks, etc and assign them some area on the screen. Not impossible, but it’s definitely not as clear as playing chess. With deep learning, we don’t have to worry so much about input representation. Instead at each frame, we can simply feed in the image into the algorithm at every frame. We also define the possible actions (left, right, jump, etc.) and a goal (win the stage). This is much clearer, and indeed someone has done something very similar.
This isn’t to say that pre deep learning algorithms couldn’t perform well on Super Mario World or that it would be a harder problem for them than chess. It’s just to illustrate the different way problems are approached by the different techniques, and how using deep learning makes seemingly complex challenges easier to solve.
How it Will Affect You
There is no question that whoever you are, your life will be, well, deeply impacted by deep learning. The relative ease of applying these algorithms and their wide applicability make this a certainty.
This is already happening in the virtual world with products such as google translate and image search. But it’s clear deep learning will soon encroach on the physical world as well. The aforementioned self driving car technology will almost certainly utilize deep learning.
Ok, Ok. If you’re a precog, you probably don’t have to worry about deep learning.
In the workplace, no profession will be left untouched. As a marketer, deep learning may be able fully utilize big data to optimize how to best spend your budget. As an educator, deep learning could tailor customized assignments for each student (mind bendingly, by learning how students best learn).
The truly revolutionary aspect of deep learning is that since these algorithms reason, they are not limited to replacing repetitive, manual labor oriented tasks; they can replace jobs we typically think must be performed by a highly experienced, educated human being.
How it Will Affect Us
So what about feedly? Machine learning and content centric apps have a long history together, probably starting way back with Greg Linden’s Findory site in 2004. The results have been decidedly mixed. The quandary is that machine learning approaches need user data to improve, but in today’s world users demand great experiences right off the bat.
This is a hard problem, one that’s still not clear how to solve. Indeed, machine learning is probably not the complete answer here: a hybrid approach combining social, machine learning, and explicitly stated user preferences may prove to be the best approach.
Further downstream in the user experience, things look a lot better. A main area of focus in deep learning is “natural language processing” (NLP), essentially improving computer’s understanding of human text and speech. Great strides forward have been made in recent years, and more seem to be on the way.
These techniques could be used to understand the content of an article, and the types of articles a particular user likes. More interestingly, they could also be used to learn the depth of a user’s knowledge. The system should be able to learn that I have only a surface understanding of politics and recommend introductory articles while also knowing I have a more in depth understanding of machine learning and recommend more advanced material. This should create a much richer user experience.
To be frank, when I first started educating myself about machine learning it was mostly out of geeky curiosity. The algorithms were fascinating, but I wasn’t that convinced they would ever be good enough to use in my day job. My opinion has been completely turned around. The flexibility and expressiveness of these models is astounding, and would seem to be capable of modeling aspects of the most complex machine ever created, the human brain.
To date we haven’t done much machine learning work at feedly (recently we’ve been busy building out our team collaboration offering), but we are fully convinced of its potential and plan to invest more resources in the future.
Don’t Stop Here
Edwin, our CEO, loves to say “Never Stop Learning.” In that vein, here are some more resources about this topic:
Humans Need Not Apply – A terrific youtube video done by CGP Grey (in feedly) that dives deeper into the potential societal impacts deep learning may have, and what we must start thinking about in order to best adapt.
A feedly search (pro and team users only) is full of great content on recent deep learning happenings.
Andrej Karpathy’s blog (in feedly) is a great resource to learn more about the actual algorithms (some coding skills required). He’s teaching a class right now so it’s not so active, he still seems to be tweeting though.
WildML (in feedly) is another great blog that is more technical and in depth.
Finally if you happen to remember any math at all from high school or college and want to learn more about the nuts and bolts of machine learning, Andrew Ng’s Coursera course is starting up again. He is an amazing lecturer and makes the material very accessible. If you can still calculate a post dinner tip without pulling out the calculator app on your phone, you should be fine.
Congratulations, you made it to the end! Now when deep learning comes up at dinner parties, you should be able to blow people’s minds with a few incisive comments. If you flop on your face, feel free to leave a comment to let us know why. Comments are also a great way to determine if your recently built AI bot can pass the Turing Test. I’ll answer all comers and we hope to dive deeper into this topic in future blog posts.
feedly is your newsfeed for work. It allows you to become more powerful in reading, sharing, monitoring, and collaborating on content. Learn more about feedly and going Pro or Team here.
A few years ago Marc Andreessen famously proclaimed “Software is eating the world.” I have to think he was speaking figuratively, but with the rapid ascent of machine learning one has to wonder. Science fiction writers and movie producers have often envisioned a dystopian future, but perhaps more troubling are real life warnings from Stephen Hawking and Elon Musk. Most people take a positive view, though, and it’s hard to argue against the potential benefits of now seemingly inevitable applications such as self driving cars. Reasonable people can take either side of the argument, but it is impossible to argue that progress will stop: like most technology, it may have unforeseen benefits and consequences but its advance is inexorable.
Is this guy going to enslave the human race or make us banana pancakes for breakfast every morning?
So whether you want to learn more about the state of machine learning, how it works, or simply wish to be better versed when greeting our future robot overlords, read on!
Terminology
There are a few different terms that get bandied about in articles about this topic: machine learning, artificial, deep learning, etc. They all mean something slightly different, but I generally use deep learning since it makes me look the smartest. Here in Silicon Valley, looking smart is almost as important as having “hacker” in your job title. So now that that’s settled, what is deep learning?
First and foremost, it’s an excellent marketing term! Computer scientists are generally terrible at naming things, so when we get something right, people perk up (see also: cloud computing).
As a CEO, what sounds better “deep learning” or “back office ERP systems”? As a journalist would you rather write about “deep learning” or “software automation”? As a student, would you rather study “deep learning” or “artificial multi-layer neural networks”? You get the point.
And by the way, the last example was a trick question, deep learning is an umbrella term for various types of multi-layer artificial neural networks. But whereas the latter term evokes feelings of a post Thanksgiving, eyes glazed over torpor, the former gets you a front page article in the New York Times.
Style and Substance
Usain may win on swagga, but even he can’t match deep learning’s performance.
While deep learning is a great marketing term, it’s also delivering amazing results. For example, in the field of computer vision, there is an annual competition where top universities and research labs all try to outperform each other. Deep learning techniques were first used by the University of Toronto in 2012. How’d they do? Well, as you probably guessed they won. But they not only won, they were 70 percent better than anything else. As a comparison, when Usain Bolt famously blew away the field in the 100m dash, he won by an incredible two tenths of a second. That’s only 2% better than second place.
True Artificial Intelligence
But it’s not just the results that make deep learning groundbreaking. Deep learning approaches things in a fundamentally different manner.
Previous state-of-the-art algorithms heavily relied on “feature engineering.” Essentially, researchers developed an incredibly deep understanding of the problem at hand and teased out the key characteristics need to solve it. Then the algorithm would simply learn how to best balance or “optimize” these characteristics.
On the other hand, deep learning simply says, “Back off, humans! Just give me all the data and let me do my jam.” The algorithms actually manage to do the feature extraction part of the process as well as the “optimization” part all on its own.
Think about this for a moment. Usually the clever parts of algorithms are discovered by people and computers are used to mechanically calculate things many times faster than we possibly could. But deep learning directly consumes raw data and independently reasons about how to best frame the problem and solve it.
This is a real step forward towards “true” AI. Consider where the intelligence lies in each approach. Sure, the previous optimization algorithms may be clever, but the real magic in the process was in humans reasoning about the data and extracting the right properties. With deep learning, this entire process is done without human intervention. This is much closer to what most would consider human thought.
Computers Playing Games
This transfer of responsibility makes it much easier and quicker to address many different problems. In software, one of the most important steps in solving any problem is modeling it correctly. Sometimes, this is really easy. Take chess as an example:
While we humans see this as a single image, computers need a numerical representation. Here it’s easy. There are 64 squares on a chessboard, and six types of pieces. We can use a list of 64 numbers to represent a particular board set up. Each number will simply identify the piece in that square.
This is elegant in its simplicity. And indeed, chess has been a popular target for computers. Many students (myself included) program a very simple chess playing algorithm as part of an introductory computer science course. At the other end of the spectrum, researchers have built optimized algorithms capable of beating the best players in the world without deep learning techniques. This is bread and butter for pre deep learning algorithms, a pretty simple modeling “space” and a huge search “space” where computers can win by simply evaluating more scenarios than humans can.
Even our fondest childhood memories aren’t safe from deep learning.
Now consider a very different game, Super Mario World. Take a look at the screen capture. How would we go about modeling this? While I get the warm fuzzies thinking about all the hours I played this game as a kid, thinking about how to model this for a computer gives me a headache. Mario and his friends can be anywhere on the screen, how do we even represent where Mario is? He’s not a single point or shape, he’s this blobby thing.
Maybe we’d have to preprocess an image to extract out Mario, turtles, blocks, etc and assign them some area on the screen. Not impossible, but it’s definitely not as clear as playing chess. With deep learning, we don’t have to worry so much about input representation. Instead at each frame, we can simply feed in the image into the algorithm at every frame. We also define the possible actions (left, right, jump, etc.) and a goal (win the stage). This is much clearer, and indeed someone has done something very similar.
This isn’t to say that pre deep learning algorithms couldn’t perform well on Super Mario World or that it would be a harder problem for them than chess. It’s just to illustrate the different way problems are approached by the different techniques, and how using deep learning makes seemingly complex challenges easier to solve.
How it Will Affect You
There is no question that whoever you are, your life will be, well, deeply impacted by deep learning. The relative ease of applying these algorithms and their wide applicability make this a certainty.
This is already happening in the virtual world with products such as google translate and image search. But it’s clear deep learning will soon encroach on the physical world as well. The aforementioned self driving car technology will almost certainly utilize deep learning.
Ok, Ok. If you’re a precog, you probably don’t have to worry about deep learning.
In the workplace, no profession will be left untouched. As a marketer, deep learning may be able fully utilize big data to optimize how to best spend your budget. As an educator, deep learning could tailor customized assignments for each student (mind bendingly, by learning how students best learn).
The truly revolutionary aspect of deep learning is that since these algorithms reason, they are not limited to replacing repetitive, manual labor oriented tasks; they can replace jobs we typically think must be performed by a highly experienced, educated human being.
How it Will Affect Us
So what about feedly? Machine learning and content centric apps have a long history together, probably starting way back with Greg Linden’s Findory site in 2004. The results have been decidedly mixed. The quandary is that machine learning approaches need user data to improve, but in today’s world users demand great experiences right off the bat.
This is a hard problem, one that’s still not clear how to solve. Indeed, machine learning is probably not the complete answer here: a hybrid approach combining social, machine learning, and explicitly stated user preferences may prove to be the best approach.
Further downstream in the user experience, things look a lot better. A main area of focus in deep learning is “natural language processing” (NLP), essentially improving computer’s understanding of human text and speech. Great strides forward have been made in recent years, and more seem to be on the way.
These techniques could be used to understand the content of an article, and the types of articles a particular user likes. More interestingly, they could also be used to learn the depth of a user’s knowledge. The system should be able to learn that I have only a surface understanding of politics and recommend introductory articles while also knowing I have a more in depth understanding of machine learning and recommend more advanced material. This should create a much richer user experience.
To be frank, when I first started educating myself about machine learning it was mostly out of geeky curiosity. The algorithms were fascinating, but I wasn’t that convinced they would ever be good enough to use in my day job. My opinion has been completely turned around. The flexibility and expressiveness of these models is astounding, and would seem to be capable of modeling aspects of the most complex machine ever created, the human brain.
To date we haven’t done much machine learning work at feedly (recently we’ve been busy building out our team collaboration offering), but we are fully convinced of its potential and plan to invest more resources in the future.
Don’t Stop Here
Edwin, our CEO, loves to say “Never Stop Learning.” In that vein, here are some more resources about this topic:
Humans Need Not Apply – A terrific youtube video done by CGP Grey (in feedly) that dives deeper into the potential societal impacts deep learning may have, and what we must start thinking about in order to best adapt.
A feedly search (pro and team users only) is full of great content on recent deep learning happenings.
Andrej Karpathy’s blog (in feedly) is a great resource to learn more about the actual algorithms (some coding skills required). He’s teaching a class right now so it’s not so active, he still seems to be tweeting though.
WildML (in feedly) is another great blog that is more technical and in depth.
Finally if you happen to remember any math at all from high school or college and want to learn more about the nuts and bolts of machine learning, Andrew Ng’s Coursera course is starting up again. He is an amazing lecturer and makes the material very accessible. If you can still calculate a post dinner tip without pulling out the calculator app on your phone, you should be fine.
Congratulations, you made it to the end! Now when deep learning comes up at dinner parties, you should be able to blow people’s minds with a few incisive comments. If you flop on your face, feel free to leave a comment to let us know why. Comments are also a great way to determine if your recently built AI bot can pass the Turing Test. I’ll answer all comers and we hope to dive deeper into this topic in future blog posts.
feedly is your newsfeed for work. It allows you to become more powerful in reading, sharing, monitoring, and collaborating on content. Learn more about feedly and going Pro or Team here.
Welcome to the third installment in our content marketing series, this time focusing on whether to fund new content or to leverage other people’s content through curation.
It has now become an age-old questions for content marketers and social media managers: Is it more effective and a better use of resources to create your own content in-house? Or is it better to leverage the flood of relevant content out there on the web and curate content? Or should we do a mix?
Original content producers have long advocated for how customized content is more effective and more on brand. Others who curate prefer to use someone else’s voice or think they can post more content via curation, making it a more effective strategy.
Spoiler: There is no right answer. But we asked three admirable companies to tell us how they choose to break the two down:
Kevan Lee from Buffer: All Original
Everything we do on the blog is original. I think on social media we might be somewhat in the minority in that we tend to share our own stuff than we do others’. And so in terms of curation, I guess we curate individually to get ideas to inform our original content. I’m not quite sure if the curation is as evident to someone on the outside or to a community member.
We used to have a feature within Buffer called Content Suggestions, which was a curated list of content that Courtney would curate on a daily basis. So curation fit really well within that product element.
In terms of marketing we’ll share lists of stuff we find really useful, but for the most part we are 9:1, sharing our stuff compared to others.
Gregory Ciotti from Help Scout: A trial with curated content that worked
Right now we write our own content, though we did try a series with curated stuff.
It was basically a curated email on everything we’ve been reading on that particular topic. We didn’t want to manage a separate email list, though people actually suggested a separate email list.
We did had a whole separate set up for creating it. It needs to be written in a way that really hones in on a specific angle of a topic. Maybe do something about remote team culture and narrow it down to 15 links or so you’d be proud to suggest for something.
And then have an original write up. I’ve only done one, but I wrote a meaningful summary under each. These kind of curation pieces still demand a lot of thought.
People really appreciated it. There were zero complaints and it was really successful.
Clair Byrd from InVision: All contributed
My speciality is contribution content strategies [when users create content]. It’s really easy to understand what to make when people tell you.
Everyone will tell you to listen to your audience. Spend time with people and understand their pain and what teams are coming up against most frequently, and then align your business goals with the teams.
The harder question is: What kind of content do you make?
My recommendation is to have the budget test all forms and make things that you can rip apart into new mediums.
If someone is building a brand focus on contributed media then also do more traditional PR activities and pull them back and see what works the best.
I also do headline tests and landing page tests.
Then there is also an intangible. I think that content marketers have the most room to use their gut out of any kind of content markets. You can’t get to what gets to people’s hearts unless you use your own. If I have a really strong inclination after seven hours of looking at Twitter, I’m going to do that.
Data-driven decisions are really important, but you wouldn’t be doing your job if you didn’t use your instinct.
As far as curated content, I think that curated content itself is completely and utterly useless unless you’re testing. If you want to understand if your audience is interested in a UX topic, curated content could be a good way to see if there is interest in that type of area.
But beyond research, I have ethical problems with curation, and I would rather spend my time working on a contributor network than working on curation.
That said, I don’t think this is true for all content organizations. It depends a lot on your people. My audience wants unique stuff. They want unique opinions.
So my answer is just to test your own audience and then figure it out for yourself. There’s no end-all-be-all number.
feedly is your newsfeed for work. It allows you to become more powerful in reading, sharing, monitoring, and collaborating on content. Learn more feedly and going Pro or Team here.
Feedly is the best way to ingest the content you need for work by putting your favorite feeds in an organized newsfeed. Over the past few weeks we have rethought the way you can clean up and reorganize your feedly. I worked with the feedly team to design two different concepts and would love to hear your feedback to help us build the best organized experience possible!
In a recent survey with 5,000 participants, many of you showed us that you like to reorganize your feedly for two main reasons:
Spring cleaning Every once in a while you need to clean up your feedly to make sure you only follow the feeds that interest you. This involves removing inactive feeds (the ones that have not published in months), removing the feeds you don’t read anymore, and promoting articles to “must-read” publications, so you don’t miss a story.
Reorganization There are other times when you feel like reorganizing parts or all of your feedly. Maybe you have new interests or you want to split a topic into a few more specific topics, such as splitting your Marketing Collection into SEO and Digital Marketing Collections. All of this involves renaming Collections, moving them around and moving feeds from one Collection to another.
After a few weeks of design work with these two use cases in mind, we came up with two design directions:
Concept 1: Organize At a Glance
The main idea behind this concept is that everything is available in one page with just the crucial information you need to optimize your feedly. Your collections are listed on the right and the selected Collection’s feeds appear at the center of the page. This enables you to move from Collection to Collection without switching context.
View important information at a glance With this first concept we are showing you the essential information you need when you need it, no less, no more. For instance, when looking at a Collection you will see the feeds that are Must Read and those which are inactive. It’s just enough information for you to take action with no clutter.
Because your Collections are listed on the right side, you can easily navigate from one to the other rapidly.
Take the main actions in one click Most actions are one click away or one drag away. Hit the cross or the star icon to remove a feed from a Collection or mark a feed as must read, respectively (see below for examples). Use drag and drop gestures to move a feed to the Collection it should belong to and re-order your Collections.
_ Making a feed must read
_ Moving a feeds to a different Collection
_ Reordering a Collection
Concept 2: Organize with Deep Site Information
This second concept takes advantage of data tables and the feedly slider. The main page displays all of your Collections. After you select a Collection we use the feedly slider to show all of the feeds it contains.
This concept focuses on showing you as much information as possible in a consistent way so you can easily decide what action to take on each item.
See all the data you need
Both the Collection list page and the feed list slider are tables displaying all the information you need to quickly undestand where you should take action. Quickly see which Collections have the most inactive feeds and which feeds last posted a long time ago.
_ Last posted data on feed list
Use a consistent popup to edit your feeds Whether you want to edit the name of a feeds, mark it as read, remove it from a Collection, move to a different one or add it to multiple Collections, a consistent dropdown menu will be there to accomplish all these tasks across the application.
_ Editing a feed
_ Reordering a Collection
Both of these concepts are available on InVision (here and there). There are a few things you can interact with so you can get a feel for them. Have a look and let us know about what works and what doesn’t. Feel free to leave comments here or within the InVision prototypes.
We are looking forward to listening to your feedback! Antoine and the feedly Team.
Feedly is the best way to ingest the content you need for work by putting your favorite feeds in an organized newsfeed. Over the past few weeks we have rethought the way you can clean up and reorganize your feedly. I worked with the feedly team to design two different concepts and would love to hear your feedback to help us build the best organized experience possible!
In a recent survey with 5,000 participants, many of you showed us that you like to reorganize your feedly for two main reasons:
Spring cleaning Every once in a while you need to clean up your feedly to make sure you only follow the feeds that interest you. This involves removing inactive feeds (the ones that have not published in months), removing the feeds you don’t read anymore, and promoting articles to “must-read” publications, so you don’t miss a story.
Reorganization There are other times when you feel like reorganizing parts or all of your feedly. Maybe you have new interests or you want to split a topic into a few more specific topics, such as splitting your Marketing Collection into SEO and Digital Marketing Collections. All of this involves renaming Collections, moving them around and moving feeds from one Collection to another.
After a few weeks of design work with these two use cases in mind, we came up with two design directions:
Concept 1: Organize At a Glance
The main idea behind this concept is that everything is available in one page with just the crucial information you need to optimize your feedly. Your collections are listed on the right and the selected Collection’s feeds appear at the center of the page. This enables you to move from Collection to Collection without switching context.
View important information at a glance With this first concept we are showing you the essential information you need when you need it, no less, no more. For instance, when looking at a Collection you will see the feeds that are Must Read and those which are inactive. It’s just enough information for you to take action with no clutter.
Because your Collections are listed on the right side, you can easily navigate from one to the other rapidly.
Take the main actions in one click Most actions are one click away or one drag away. Hit the cross or the star icon to remove a feed from a Collection or mark a feed as must read, respectively (see below for examples). Use drag and drop gestures to move a feed to the Collection it should belong to and re-order your Collections.
_ Making a feed must read
_ Moving a feeds to a different Collection
_ Reordering a Collection
Concept 2: Organize with Deep Site Information
This second concept takes advantage of data tables and the feedly slider. The main page displays all of your Collections. After you select a Collection we use the feedly slider to show all of the feeds it contains.
This concept focuses on showing you as much information as possible in a consistent way so you can easily decide what action to take on each item.
See all the data you need
Both the Collection list page and the feed list slider are tables displaying all the information you need to quickly undestand where you should take action. Quickly see which Collections have the most inactive feeds and which feeds last posted a long time ago.
_ Last posted data on feed list
Use a consistent popup to edit your feeds Whether you want to edit the name of a feeds, mark it as read, remove it from a Collection, move to a different one or add it to multiple Collections, a consistent dropdown menu will be there to accomplish all these tasks across the application.
_ Editing a feed
_ Reordering a Collection
Both of these concepts are available on InVision (here and there). There are a few things you can interact with so you can get a feel for them. Have a look and let us know about what works and what doesn’t. Feel free to leave comments here or within the InVision prototypes.
We are looking forward to listening to your feedback! Antoine and the feedly Team.
Today is the second installment in our content marketing series, this time focusing on how to execute on awesome content. This is because we believe content is a currency. It is the marketplace for new ideas and, increasingly so, a core engine in providing value to businesses and to customers. Among the many people who use feedly, content marketers are some people who know this truth best.
Whether you’re at a startup or big company, time is the most coveted resource for many content marketers. In our feedly content marketing survey, 49 percent respondents said lack of budget was the biggest problem, 39 percent of respondents said lack of headcount, and 36 percent said it was the volume of content
So we posted the question to two content marketing pros: Gregory Ciotti from Help Scout, and Kevan Lee from Buffer.
Some overall insights from our panelists:
Being dedicated to content has enabled them to produce at a more efficient rate than when they were divided among marketing activities.
If you have the resources, it helps to have someone dedicated to content strategy and separate people for content production.
In their own words:
Gregory Ciotti, Help Scout: 8 to 32 Hours
This is one of the toughest things to answer. You might sit down and write something for 40 minutes, but the process was five years in the making. The average time that it takes me is about an eight-hour work day, but you need to put in the research and the thinking into it, too. There’s probably a 16-hour disparity because I’m counting the time it takes to interview people and learn more. But for the writing part, a really solid post probably takes at least eight hours.
Kevan Lee, Buffer: 4 to 15 Hours
The time changes based on how much experience and how many blog posts you’ve written in that style.
My time got shorter and shorter the more I did it. I ended up getting it down between four and six hours per post. But that was after writing four to five posts a week. Within a few months, you’ve written over a hundred posts, so writing at that volume helps you cut down the time as well.
As we’ve grown the team, we’ve switched to encouraging the writers to be able to do two posts a week. So that would be like 30 hours of the week—so 15 hours per blogpost. It’s a way to ramp it up and get into the pace that you can get to once you’ve written at a high volume. It definitely depends, especially as you get the opportunity to practice.
feedly is your newsfeed for work. It allows you to become more powerful in reading, sharing, monitoring, and collaborating on content. Learn more feedly and going Pro or Team here.
A driving belief for us at feedly is that content is a currency. That is, content is crucial to the way we work today. It is the marketplace for new ideas and, increasingly so, a core engine in providing value to businesses and to customers. Among the many people who use feedly, content marketers are some people who know this truth best. So today we are launching a new series for content marketers that provide tips and tricks on how they can perfect their art. Here is the first installment.
Whether you’re a veteran writer or new to the game, one essential question for every content marketer will guide the rest of your content strategy and performance: What kind of content should you and your company produce?
Not only is it a crucial question—it is a core challenge for many content marketers. In our recent feedly Content Marketers’ Report, about a fourth of respondents said that understanding what kind of content to produce, personalizing the content, and localizing the content are some of the top challenges that they face.
The answer, of course, depends on your company, your marketing goals, and your customers.
So we asked three leading thinkers about how they answer this question.
Help Scout: Just ask the customers
Gregory Ciotti, Helpscout
According to Gregory Ciotti, head of content at Help Scout, his team focuses on three things:
Talking with customers.
Creating personas based on their conversations with their customers.
Content customized to whether the customer is new to Helpscout, considering using Helpscout, and about to buy Helpscout.
Greg shared some of his favorite practices in his own words:
01 Talking with customers
I love the question what do you wish you knew then that you know now? I like to ask that of people.
And they’ll tell you. I struggle with this, I’m currently struggling with this, and you take it from there and see if you can find someone else who has addressed that topic before to see if you can talk to someone else who has figured that out.
I never really had a great fresh opinion on support by just browsing what’s already out there. It’s almost always better when someone just tells you, hey, I’m having a really hard time writing support updates for my team. What do those look like. What kind of supporting from my team Help Scout should I share? What kind of reporting numbers should I share? Maybe I won’t know at the time, but I could go ask someone else. I could go approach someone else and figure that out and then go from there.
02 Creating Customized Personas
I’ve always said write for an audience of one. I actually use specific people. There are certain support managers I always have in mind. I follow along with what they’re doing. So it starts with a persona, but I really think that you should pick specific people. I really want to see what a support manager and at least 30+ people on their team currently struggling with. What are the difficulties there? And when you have a single person in mind, a lot of times it’s a simple thing to just go and ask them if you have questions.
03 Content customized to whether the customer is new to Help Scout, considering using Help Scout, and about to buy Help Scout.
I think people complicate things using customer journeys. You’d almost be better off if you’re somewhat new to break it down to a reverse pyramid. Essentially, you ask yourself, what’s the high level stuff? What’s right about the middle of the range, and what’s the right material that could work at the bottom?
For us, at bottom of our little pyramid, is product marketing materials. Such as choosing a help desk, white papers, and those sorts of things.
The middle is best built around the key personas that you have. To give an example, support managers are really a key person for Help Scout. They’re often the kind of person who gets buy in for the team and company to use Help Scout. So the middle of the funnel, we try to really create deep dives for support topics and for support managers.
It’s key in that middle section to be very honest in what you’re able to create or and what material is better to get from someone else. So we’ve opened up a guest author program. This is because I’m just not going to be able to tell you advice that a director of community could tell you. Or a head of support of some specific kind of company could tell you.
The top layer is where we try to keep it square in the style of something only from a team of our size or bigger could be able to write. It’s not really just going for anything. We mix in support topics that are more like holding conversations with customers—something more of a customer service representative would use. But we also have stuff that’s like from the team. People talk about our publishing strategy, we’ll talk about how we do onboarding. We talk about how we build product. We talk about, really, anything that relates to a company around our size.
Buffer: Headline monitoring and keyword trends for fresh ideas
Finding headlines that resonate with their product on Twitter or on feedly.
Keyword tracking on social
Writing topics that create thought leadership in the social media space
Experimenting with new forms and tracking performance
Kevan Lee, Content Crafter at Buffer, shared some of his secrets. In this own words:
01 Finding headlines that resonate with their product on Twitter or on feedly.
In the past it was a lot of intuition-based ideas. It was the stuff that tended to resonate with us as we’d scroll through Twitter or our feedly feeds. Headlines that caught our eye or topics that caught our attention or stuff we thought about and thought we’d love to write about from our perspective and see if we could put it to use for our audience. So that was kind of a big chunk of our ideas back in the day and continues to be moving forward, too.
02 Keyword tracking on social
We’ve also kind of transitioned into a more disciplined approach where we’re thinking of topics that have more of a specific goal to them. So do we want to rank for certain keywords that we think have a lot of traffic or are useful for the audience we want to serve. Can we write about a topic that ties into Buffer pretty well? So, like how to manage multiple social media profiles or different things like that.
03 Writing topics that create thought leadership in the social media space
Our hope and our goal was to focus the buffer content so that it might be well tuned to the audience for whom Buffer might build the product. So kind of creating some brand awareness or topic niche awareness there and just trying to do our best to stand out as a thought leader in social media in the social media space.
Our assumption was that taking a similar approach with well-researched, in-depth content would help set us apart in that way. And it’s been a fun journey toward that.
04 Experimenting with new forms and tracking performance
We typically create semi-long-form pieces. It’s typically like 2000-2500-word posts. We do about four times a week on the blog. That’s kind of our bread and butter.
But we’ve also tried ebooks and marketing resource kits and things like that. I’ve done some webinars in the past. I’ve been doing some Slideshares currently. Getting into Medium. Kind of a long list of random stuff.
Invision: Empowering customers to create the content
Clair Byrd, Invision
Invision creates tools for designers to create prototypes and to collaborate with other teammates. When it comes to determining the type of content they produce, they take a specific approach: Letting go of the reins and letting their customers provide the voice.
How do they do this? According to head of content Clair Byrd, they:
Create a contributor network for their users and share their platform.
Empower their users by allowing them to write what they feel passionate about.
They integrate their users’ content for prospective customers.
In head of content Clair Byrd’s own words:
01 Creating a contributor network for their users and share their platform.
We have our publishing platform, which has become a big deal for our users. We have a really highly specialized audience of people who write on the platform. They get to be the rock stars. We are highly visible in the design community and thereby with lots of people.
So how do you think of content ideas? The answer for me is, we don’t have to.
02 Empowering users by allowing them to write what they feel passionate about.
It’s really empowerment that is our hook.
We give people the ability to write whatever they want, and that empowers people. It positively impacts our production of when they want to do it and what they want to do.
We reach out to people we really like. Or sometimes people reach out to us, and we basically let anyone write for us. But there are a couple of content programs reserved for a specific caliber of writer.
03 Integrating users’ content for prospective customers.
We hook the users’ content into the sales funnel for renewals and upgrades. If we are on the verge of an upgrade or enterprise deal, we use content to bring them into the brand.
feedly is your newsfeed for work. It allows you to become more powerful in reading, sharing, monitoring, and collaborating on content. Learn more about feedly and going Pro or Team here.
A few years ago Marc Andreessen famously proclaimed “Software is eating the world.” I have to think he was speaking figuratively, but with the rapid ascent of machine learning one has to wonder. Science fiction writers and movie producers have often envisioned a dystopian future, but perhaps more troubling are real life warnings from 
