We largely invest in consumer web services with a large number of engaged users where the users create the content. Services like this can become messy and hard to navigate. There is always a signal to noise issue.

I'm a big fan of curation in these services. Twitter has lists. Etsy has favorites. Tumblr has tag pages. These are all variations of curation in services that have a lot of noise in them.

Recently Kickstarter launched their own version of curation called Curated Pages. In the Kickstarter model, "Curated Pages are a way for organizations, institutions, and (soon) individuals to share projects they love on Kickstarter."

Here are some of my favorite Curated Pages:

The Magnum Foundation Emergency Fund – The Magnum Foundation Emergency Fund provides grants to photographers who are documenting social and political issues around the world. Now with Kickstarter, we can all help fund these important projects.

NYU's ITP Program – NYC's "media lab" and one of the most impactful and important pieces of NYC's tech community. They've curated a page of projects they like.

Creative Commons – The Creative Commons organization evangelizes for technology and legal frameworks that facilitate sharing and creativity.

The Sundance Institute – A curated page promoting Kickstarter projects from Sundance supported artists.

You can find all of the current curated pages at the bottom of the Kickstarter home page.

If you are interested in curating a page on Kickstarter, this feature will be made available to everyone soon.

If you are building a marketplace or a social platform, make sure to build curation into your model. It will make the service easier for everyone to navigate, particularly new users.


Comments (Archived):

  1. William Mougayar

    Wow. My favorite topic, Curation. I’m about to finish a blog post titled “Curation is overhyped”.Curation is becoming hyped and is starting to lose its real meaning. It’s used as a catch phrase for marketing now, because it seems to be the “hot” thing.Curation implies some level of expertise by the curator. The problem is that, if everybody is becoming a “curator” suddenly, regardless of their expertise level, you end-up with an abundance of poorly curated services. The Wikipedia curation works because there’s a rigorous process for that. If everybody is now a “curator”, we’ll lose the real benefits of curation.I’m for controlled curation by experts, not curation by the masses. Then it’s user-generated content, and it should be called as such.

    1. falicon

      I think going to the masses is the only way to get value out of the long tail…and I think it can work, but the problem is right now many people are confusing aggregation with curation…just collecting a bunch of stuff isn’t the same thing as specifically filtering out (or picking) a subset of information with a specific intent.I think it also depends a lot on who you are curating for…there’s no reason I shouldn’t be able to curate stuff for say my mom if I wanted to….no one else may get value out of it (so it prob. wouldn’t fall into your definition of curation) but that wouldn’t have been my motivating factor or intention anyway…I think of UGC as content created by a user…curation is content picked by a user (not generated)…and aggregation is just a system for collecting a bunch of content (so to me good curation involves simple tools for aggregation and then filtering, picking, and sharing the best from that aggregation).

      1. fredwilson

        yup. the curating for your mom is a great example

        1. FAKE GRIMLOCK


      2. William Mougayar

        I think Picking content is like sanitizing, selecting, triaging, editing. I maintain my position that Curation is currently hyped.

        1. Steven Rosenbaum

          Ah – ok, now I see. You maintain that curator = filter. I think you’re missing a key element. A curator is a Creator. A remixer. A maker. The aggregating, and organizing is useful, yes – but it is by adding content, and creating new experiences that the Curator emerges as an essential new force on the web.Curators emerge as the human layer – essential, and as of yet unformed. Fred’s Kickstarter example is a good one. And there will be many many more.I’d say not hyped at all, embryonic.looking forward to your post.

          1. William Mougayar

            I agree that curation is all of the above Steve, not just filtering. I didn’t imply just filtering. Curation is a before, a during and an after task. The maintenance is as important as the creation, organization, selection and filtering. And you might add promotion and user-engagement. As a curator, I want users to be engaged with the content I’m curating in order to improve on it or on my methods.

          2. William Mougayar

            I think it’s past embryonic. We’re full steam in the “peak of inflated expectations”, according to the hype cycle model.

          3. awaldstein

            At it’s core though William, curation expertise is simply a matter of popularity.You agree with that?

          4. William Mougayar

            Arnold, I think popularity can inform curation or a curator. Both are related, but it’s not “only” about popularity.

          5. FAKE GRIMLOCK


          6. William Mougayar


          7. FAKE GRIMLOCK


          8. fredwilson

            Awesome comment grimlock

        2. falicon

          I think it only feels hyped because you are connected to the world that cares so deeply about it…i would bet the average person on the street doesn’t know what it is…and if explained to them, I bet they would think it sounds great! (classic example in this realm I think is delicious – and there are still millions of Facebook users that have no idea something like delicious exists)

        3. reece

          i agree that curation is a hot topic these days, but that doesn’t mean only ‘experts’ should have license to curate.that’s the beauty of the internet, anyone can become an expert – whether a factual, proven expert, or just an expert at expressing oneself through their Tumblog – that opportunity exists for anyone on the web.

          1. fredwilson


          2. reece


      3. sigmaalgebra

        If you are picking for your mom, then ‘the best’, the ‘most popular’, etc. likely won’t do you much good.

      4. Mark Essel

        Groovy, also dig Charlie’s example. Now I want to curate a trip with friends 🙂

    2. Fernando Gutierrez

      Yes, but…Curators should know what they are talking about if they are suppossed to add some value. But the act of curating may turn you in some kind of expert. If you take the time to curate that means you care. And if you care you usually know something about it or are going to take the time to get informed.

    3. Steven Rosenbaum

      Curation by experts is old media. Because ‘expert’ was a term that was determined by the owners of the publishing platforms (TV, Radio, Print). You were an expert if CNN said so.Today – anyone can publish, so anyone can curate. It creates a democratic framework for content to be both created and curated.Yes, it means that there will be a large number of curators – but the expert is in the eye of the beholder. Folllow whomever you want – and unfollow if they’re curated and created content doesn’t meet your needs.Curation’s real meaning is emerging, and evolving. But the really good stuff is just around the bend. (@magnify)

      1. Dan Lewis

        I agree that the old definition of “expert” (a person deemed as such by a legacy publishing platform) is incorrect, and that we’ll see — and, in fact, have seen — the next generation of experts arise from proof of work, not from edict of king.But it doesn’t follow that those who run platforms, old (CNN, NYT) or new (Tumblr, Twitter) should simply outsource this to the masses, any moreso than an old platform would outsource creation to a random name picked out of the phone book. Once an expert curator emerges, the platforms would be well-suited to pick that person up for their joint benefit. To say that the platforms shouldn’t be kingmakers is only correct to a point.

        1. Steven Rosenbaum

          The thing about Curation is that it is contextual, so someone who curates great family physicians on the UWS of Manhattan is going to be useful, even important, to a neighborhood – but a rounding error to a platform provider like Twitter.Here social proof is far more meaningful than platform provided ‘experts’.The volume of information being produced is too vast, and the nichification of curators too granular for any uber-validation to be useful.The good news is – we’ll surely all find the curators that help us make sense of the noise.

          1. Dan Lewis

            Yep. I’m entirely agreeing with you. But what I’m suggesting is that, once that person emerges, someone should be hiring the curator or buying his list. In this case, New York magazine could for their Best Doctors edition. (Imagine if they did that city-wide!) Or someone could create a brand-new UWS publication/destination (akin to… but useful) with all sorts of curated lists that go well beyond the standard yellow pages.

          2. Steven Rosenbaum

            Ah, well – money. So, here’s the thing, people both make and curate content for different reasons. Clay Shirky’s book Cognitive Surplus makes a great argument for the fact that people participate for lots of reasons, money not being the one that is the most important.Conversation and Curation are uniquely human expressions – so i’m not sure that seeing good curation’s best result is being ‘picked up’ by paid media.It may be that lots of new curators get paid with an economic unit i call (for lack of a better word) joy. It feels good to share what you know – and payback is a ‘thank you’ or a comment.One this is clear, curation isn’t one thing – it’s the human input arriving to sort out a noisy web, and it will come in both amateur, pro-sumer, and pro flavors.

          3. FAKE GRIMLOCK


          4. fredwilson

            I agree that curation is like chocolate

          5. Robert Holtz

            @FAKEGRIMLOCK (can’t reply directly thanks to DISQUS limitations)…Grimlock, your point about chocolate is well and good but let me just point out… The reason we all know we like chocolate today is because of the first Aztec eons ago who took a bite of it and said to the guy next to him, “OMG, you gotta try this stuff.”

          6. FAKE GRIMLOCK


      2. fredwilson

        i agree steve

      3. William Mougayar

        Hey Steve, I know what you are saying, and I’m all for controlled curation. But too much curation output can create a poverty of attention. There has to be a happy medium between choice and quality. Otherwise, we’re back to noise being louder than signal.

        1. FAKE GRIMLOCK


          1. David Semeria

            Both the caps and the grammar are becoming tedious.Think about that.

          2. Robert Holtz

            Hey Dave, lay off of Grimlock. Will ya?He is much-loved here… caps and all. His remark was completely relevant to the discussion.Giant robot dinosaurs just don’t have much use for lower case. ;P

          3. fredwilson

            Grimlock is one of the best things to happen to this blog recently

          4. kidmercury

            in total agreement with you david. especially on the caps lock. though i’m going to drop some hate on disqus here, they def need a block feature.

          5. fredwilson

            I’m gonna disagree kid

          6. fredwilson

            Not to me

          7. ShanaC

            then you get a multipolar issue in which no one is able to decisively act to change tastes…

          8. FAKE GRIMLOCK


          9. ShanaC

            Peggy Guggenhiem and abstract expressionism

      4. reece

        well put, today’s web, YOU choose who the experts are.

        1. falicon

          Reecepacheco…I choose you! ;-)Sorry – couldn’t resist…and clearly it’s time for me to go get some sleep (I watched too much on Shelby today I think!) 😉

          1. reece

            hahaha… glad to hear it. 😉

          2. FAKE GRIMLOCK

            HURR HURR. ME LOLED.

      5. awaldstein

        Helpful language to think about this Steve.I’ve been looking for a concept for thinking about brands that are built at the intersection of their own content and curating other content around it. Topical design defining a community of interest.It seems that with increasing noise, curation can also become a float for original concept to be found.

        1. Dave W Baldwin

          Glad I read this reply today, before a second message I’ve promised someone.Thanks Arnold!

      6. Robert Holtz

        I was GOING to totally disagree with you until your very last sentence.In my view, curation has always been the better way to sift through content and media. What is REALLY changing are the ways and means by which we qualify and decide on which curators we will spend our limited attention spans.Curation’s REAL meaning IS emerging and evolving. And I agree with you… really good stuff is just around the bend. Your point is that the term “expert” is too absolute going forward. That deciding which curator is best is rapidly becoming entirely relative and personalized. There I mostly agree but I still think there will be MOST liked and MOST favorited, as Fred said, and THOSE curators CAN be called EXPERTS. And to my earlier point, they will have earned that title through a better process of qualification than whatever we’ve had heretofore.And THAT is why the good stuff is still ahead.

      7. Austin Clements

        I was about to comment something similar, but you explained it so well. In a time where there is entirely too much valuable relevant information to digest, individuals must rely on various trusted sources (publications, bloggers, friends, industry leaders). It would indeed be silly to assume that my trusted sources will be the same as almost anyone else’s.When I first saw this last month I thought Kickstarter is awesome for doing this. Finding interesting new projects will be even easier when users can look through the window of organizations they already identify with. I think the good folks over at Kickstarter can take it a step further by allowing the organization to actually raise funds for one of the projects they have selected. They can allow the ‘followers’ of the organization’s curation page to vote on which project gets the funding. That could lend itself to creative organizations hosting their own contests utilizing Kickstarter as the platform to facilitate the funding process.

    4. Dan Lewis

      If “everybody” is curating “everything,” I’d agree. But if “everybody” is curating things specific to each individual person, not so much. (Or in other words, I agree that “[c]uration implies some level of expertise by the curator.)The vast majority of Twitter lists are garbage for exactly that reason. But that doesn’t mean there aren’t very useful ones out there for exactly the contrapositive of that reason. So the signal:noise ratio of looking at lists themselves is very poor, and it’s kind of ridiculous to think the next step is curation of lists themselves. Never ends.Unfortunately, that’s typically the case, so in general, I agree with you. Take a look at the absolute mess the ecosystem formerly known as the blogosphere has become. First, there were blogs, and there weren’t so many on any given topic that you’d get too lost. Then, it grew; to curate, there were blogrolls. Then blogrolls became these unwieldy things which required curation themselves, which became big blog indexes, and later technorati/google blog search/ice rocket etc.All said and done, most of the curation tools disappeared, and with it the ability to find an audience for your blog. And then Twitter came along.It’s a failure of curation for the reasons you implied.



    5. fredwilson

      i don’t totally agree. curated pages that are highly liked or favorited areoften better than those created by so called experts

      1. Dan Lewis

        Curated pages that are highly liked or favorited are the ones created by the experts. The definition of “expert” is the problem there, not William’s point, really.

      2. William Mougayar

        There is more to curation than “liking and favoriting”. Liking and Favorites are great and they let the good stuff rise to the top, but it’s a “loose” characterization of curation. That’s my point. Maybe it’s semantics, but my other point is that when something starts to get used loosely, it loses its meaning.Liking and sharing are methods that surface the signal from noise. I wrote a post 6 months ago called “There is More To Content Manipulation Than Sharing and Liking”…

        1. Mark Essel

          Agree on overloaded terms losing their utility, said the same of relevance not long back.But part of curation is social filtering, and I believe we have a long way to go before that market is mature. There are petabytes of practical knowledge in human minds that has never made the transition to accesible digital networks. Mega corporations will rise and fall solving that potential gap.

          1. justin kazmark

            Reading this thread and reminded of an NYT story from a couple years ago exploring the over-adoption of the term. Thought you all might enjoy, so figured I’d curate it into the conversation:

        2. ShanaC

          Well yes. Though honestly at this point Ian not sure what counts as content.

        3. Robert Holtz

          That’s not what Fred said. Liking and favoriting is a way of gauging how curated content is resonating with the masses. But the curator themselves is playing a much more involved and intrinsic role; exposing connections, providing value-added insights, serving as a taste-maker, discovering emerging talent, providing a highly tailored experience to a given affinity group, providing relevance and context to a given social graph. No one said that liking or favoriting = curation. In my opinion, @kirklove put it best with the “gallery” metaphor. It really fits. Your warning is that if everyone is an expert, by extension then, no one is an expert. What Fred DID say is that, liking and favoriting is our filter against THAT because the public can very quickly communicate to others if the curator has done a good job or not. If something a curator has put out into the world is highly valued by a social graph you happen to care about, that is a very good indicator of value and it will push its own way to the foreground as the rest falls away into the quiet oblivion of irrelevance.

      3. sigmaalgebra

        Yes, but ‘liked’ by person A for their interest X is still better, much better. Now, how to write software to do that? For everything on the Internet, text or not?

      4. Robert Holtz

        Agreed, Fred. Curators that might come from “anywhere” whose efforts are highly-liked or “favorited” are in many ways better qualified as thought leaders than so-called experts whose qualifications may otherwise only be self-proclaimed. Curation, as we are describing it here, is not a free-for-all… its a meritocracy.

    6. kidmercury

      siding with william in this beef. or more correctly, i see the value in both sides, but think there is greater opportunity, in both value creation and profit potential, in curation by experts. often in many topics, non-experts are not qualified to curate.

    7. sigmaalgebra

      No, ‘curation’ is fully important. The problems are, (1) manual methods are too expensive and (2) the software out there sucks.”Curation implies some level of expertise by the curator.”: Not necessarily, especially if want to automate the curation. Then when doing automated curation well, there can be a crucial role for data from non-experts.

      1. William Mougayar

        I don’t totally disagree. Curators need automated tools to be productive at curation.But curation by non-experts leads to lower quality curation.

        1. sigmaalgebra

          I didn’t write at all clearly: I’m suggesting doing curation automatically with software. So, there are no human ‘curators’. The software needs data, but, curiously, it doesn’t really need its data from some unusual fraction of ‘experts’.To be more clear, we’d need a big pot of coffee, a bigger stack of paper, and maybe several new pens. I didn’t say it was easy.If you get near the EW Route 55 and the NS Taconic State Parkway, then I’ll supply the paper and pens, pay for the coffee, and settle for ice water.

          1. Steven Rosenbaum

            here’s a place where i think language matters. Curation is a human endeavor – not an algorithmic one.Computers can’t curate any more than they can write a song, or paint a picture, or carve a sculpture.Computers are tools, like brushes, carving knives, or guitars.Sites like TechMeme started out as algorithms, and added a human curation layer because code couldn’t create a meaningful collection with a human voice.Pandora has a ‘Chief Curator’ on staff.

          2. sigmaalgebra

            “Computers can’t curate any more than they can write a song, or paint a picture, or carve a sculpture.”Not really. It’s tough to say just what computers can and can’t do. And what they can do in the future stands to be way ahead of what they are doing now.Essentially everyone who has tried to do software for curation has failed. Right.So, they didn’t know how to do that. But, there is a way: I assure you, there is a way. Why? ‘Cause I know “how to do that”. I did that. The software’s ready for production.Of course, saying ‘software’ is misleading: The key is not ‘software’ but what the software does, that is, what other ideas and techniques it implements. Any such is math, understood or not, powerful or not. For more powerful techniques, understood, proceed mathematically. I’m not talking high school or college math here. And not all of it is on the shelves of the libraries. And for the math prerequisites, only a tiny fraction of university math profs have them all. The best of the math, when I first studied it, looked absurdly abstract, far from anything real. Alas, that stuff turned out to be one of the keys if only because we need a rather abstract approach to this thing.Next, a direct attack on ‘curation’ as humans would do it is both not promising (just as you said, “write a song, or paint a picture, or carve a sculpture”) and not necessary for good results. Instead, it’s enough to be able to have software that can work effectively with ‘meaning’ as humans mean it. And even there it is not necessary to take a direct attack, i.e., achieve ‘real artificial intelligence’.As usual in problem solving, a good solution doesn’t need all the assumptions one might imagine. If careful with both the assumptions and the techniques, then there are enough assumptions and data for good curation of all of the Internet content, text or not, now.I will agree that the solution is not “algorithmic”! Algorithms meant something for, say, sorting. E.g., heap sort does sort, is essentially ‘in-place, and achieves the (n)ln(n) Gleason bound both in average case and worst case. But for curation, algorithms are meaningless and, thus, nonsense. What is important is what the code, the software, the ‘algorithms’, implement, and that prior ‘stuff’ is, in this case, essentially forced to be some math, however, not very elementary math. So what matters is the math, not the algorithms. So, it’s not an algorithmic problem; instead, it’s a math problem.How to do that? Get a good ugrad major in pure math and then a good Master’s in carefully selected topics in pure and applied math, where some of the topics are rarely taught. Learn how to do reseach in math, e.g., get a corresponding Ph.D. Then get a lot of experience attacking real problems with that math. Curiously I didn’t mention ‘computer science’!

          3. sigmaalgebra

            Steven,We’re running out of the tree depth limits on Disqus!I’ve outlined some good news, ‘automation of curation’ as a solution to the problem of ‘curating’ the huge and exploding content on the Internet, and you are not actually thinking about what I wrote, staying with your prior understanding, resisting, and not learning.You have some problems in your response:First Problem. For your 1., 2., 3., of course not. So far any such is silly. E.g., I’m a big fan of music, between Vivaldi and Rachmaninoff. There music is ‘art’ as in the definition “communication, interpretation of human experience, emotion”. I’m not holding my breath waiting for computers to do well at that.So, I’m not talking about computers ‘creating’. Instead I’m talking about computers ‘curating’, and that’s quite different! I’m sure you can see the difference. Now that we’ve seen the difference, let’s move on.Second Problem. So, we can have humans ‘curating’ music, that is, the music critics. Or, there we get “the words without the music” (J. Heifetz). So, even human curating of music ain’t easy or very meaningful.Here we’re talking ‘curation’ as essentially collecting together. Even humans don’t do this very well, that is, very ‘meaningfully’: Last time I was at the National Gallery of Art on the Mall (thank you Mr. Mellon), I saw ‘curation’ moslty just by date, within that by geographical region, within that by artist. Gee, maybe a computer could do such ‘curation’! :-)!!Third Problem. On this thread, we are talking about the problem of the explosion of content on the Internet. There ‘curation’ means somehow ‘bringing together’ content according to various criteria.For that, I assure you, by far the best way to proceed is by some ‘base data’ from humans, really just a ‘representative cross section’ will do, that is, don’t really need experts ‘curating’, and some well designed software.

          4. Steven Rosenbaum

            So, I completely, totally, and passionately disagree with you.Period. Full stop.I agree that machines can aggregate. And I agree that machines can ascertain that some X number of people liked a thing.But curation is creative. It is taking diverse, often incongruous things and creating a unique, new creation out of sources and voices.By trying to hijack the concept and add it to the list of things that computers will do for us, you put at risk the HOPE for a coherent human filtered web.Let’s let computers fix spam first – i’ve been promised that for years. Then they can move on to creating content collections where you can’t ‘smell the metal’. (credit to Brian Alvey for that fine phrase).

          5. sigmaalgebra

            “I agree that machines can aggregate. And I agree that machines can ascertain that some X number of people liked a thing.”Right. But that’s not what I’m proposing. So, you are not describing what I’m proposing. Your description is nothing like the software I have ready for production.I don’t want to get deep into my software here, certainly without a lot NDAs, etc., but there are some simple cases you know well that contradict your claims.You are trying to assert that software can’t do something, and that’s thin ice to stand on.For your challenge of spam, yes, as a special case, my work solves that problem, too. For e-mail? Not really. For content on the public Internet? Yes, very much so.Invitation: If you get near the EW Route 55 and the NW Taconic State Parkway, on the SW corner is the Planet Diner, then I will buy you coffee, drink ice water, bring a big stack of paper and some pens, give you some intuitive explanations of some advanced math, outline some original math, explain the data needed from humans, and outline a solution to curation of all of the Internet, text or not.Yes, this work is some solid, practical progress in how to have computers work with ‘meaning’ as humans view it.Then with this progress with meaning, can direct focused content to focused, personal interests.Uh, the work is new, as in ‘research’ as in some advanced, applied math. That’s why I went to grad school in applied math.But it’s easy to get confused, misread what I’ve written, and, then, conclude that my claims are impossible. E.g., I didn’t say that my work could create music, only that it could curate music.I didn’t say that humans were not needed at all; I just said that they did not need to be experts or to curate.And I’m not talking popularity voting, Digg likes, or links. And I’m not talking singlular value decomposition, artificial intelligence, machine learning, speech recognition, heuristics, or image processing. And the key is not ‘algorithms’: My code implements some math, and the math is what is crucial. And you won’t see the math in high school or college. And some of the math you won’t see anywhere because it’s original (I’m qualified to do that).Again, in problem solving, have to be quite careful about what are assuming and what are getting and then have to work quite carefully with the hopefully meager assumptions. E.g., my work solves the spam problem on the public Internet but not in e-mail. E.g., my work curates still images but does not create them.It’s quite doable. But, not everyone will see how; in fact nearly no one will see how; and most people who think about it will get off the track into mud holes and claim can’t get there from here. That difficulty is part of the opportunity.It’s inevitable: If want to find some things that are really good, really good because they are really difficult and, thus, rare, then likely have to be willing to look carefully at some things that are quite rare, difficult, and different.I’m outlining a solution, and you are angry. Just what is it about a solution you do not like?

          6. PeterisP

            Pandora’s chief curator has not seen my tastes or created anything for me – the machines are curating my playlist for this evening. And they are succeeding. Better than many humans could.Machines will be able to do everything that humans can, given a bit of development – the “impossible to be creative” myth is considered completely baseless by all AI research. And even today, right now machines can do things that humans cannot – such as curating on a reasonable scale. Humans are poor curators – their curating doesn’t scale to be useful for an individual approach to every customer.In every sense that ‘computer is just a tool’, your own brain and reasoning is just a tool as well. If an overgrown visual cortex of a hairless primate can create an interesting group of artworks, then a silicon chip and some software, of course, can do the same.

          7. fredwilson

            I think it is a matter of taste. I don’t like Pandora in the least. I feellike it is Muzak

          8. FAKE GRIMLOCK


          9. William Mougayar

            You need both: human and algorithms in order to achieve the best results.Curation by humans only is tedious.Curation by computers only is subject to errors.The end-result is quality content that others can appreciate. You’re saving them time, giving them insights, or allowing them to discover areas they wouldn’t otherwise.

          10. sigmaalgebra

            For good automation of curation, do not really need humans ‘curating’.Instead, at least so far, just need some quite general data from some humans. Actually, you can see some of this with some really simple examples you already know well.

          11. Steven Rosenbaum

            sigmaalerbra: rather than mess with the word ‘curation’ (which I maintain = human organization) why don’t you share some examples of computer generated data that rises to the level of human creativity:1. A computer created song: __________2. A computer created piece of art: ____________3. Computer poetry or literature: ______________I understand that theoretically, there is a school of thought that this is doable. But I don’t agree. The nature of creativity is that it is a mix of intellect and instinct.This is an important debate, or the word curation will lose its meaning.

          12. PeterisP

            to Steven – Pandora does better curation of new music for my tastes than my human buddies and band mates can do.Curation is really the perfect word for the service they’re doing – I haven’t thought of it before today when this discussion sparked attention to the concept, but it feels like an exact match. And here you go – curation with no human bottleneck.

          13. William Mougayar

            Does Pandora curation or recommendations?

    8. Matt A. Myers

      Here, I’ll jump in with feeling;The goal of curation is to give a use or pleasure to the viewer. The expertise is that of knowledge or feeling. If someone has a similar feeling of enjoyment then you connect with them on the deepest level possible. This works in an art gallery or technical fields.Everyone’s an expert of feelings – it’s innate.Sharing and exploring are finding connections with others who feel the same or similar to us.The platforms that exist now (minus the ones who don’t get it) are allowing us to connect with our feelings now; not simply communication tools anymore – thank goodness.I now go back to drinking excessive coffee (a new habit) to pump through everything to reach that tipping point; I shall have to start a donate-to-my-coffee fund soon.

    9. Esteban Contreras

      The question is who will become an “expert” and who will be “the masses.” Curation might be “overhyped” in some circles, but the need for it is anything but fluffy. Curation, when done well, is a huge differentiator and it provides valuable new meaning/insight/expertise/pov.

  2. Neil Braithwaite

    I believe curated pages help create two essential elements of a web site:1. A more engaged user2. A more sticky web site.Crowd source sites that don’t have curated pages/aspects will not last.

  3. Bryan Everly

    You might want to check out my little side project I’m working on then.http://statuscurator.comThe idea is to pull all of the sources of information you frequent (facebook, linkedin, twitter, google reader and instapaper for now) into one place and then allow you to share the things that interest you back out to the various networks you belong to.It’s a little rough right now but anyone who wants to can sign up using their Twitter credendials in my public beta and provide me feedback.

    1. fredwilson

      i’m taking a look now

  4. andyswan

    Does anyone have examples of excellent curation beyond what Fred has in this post?Is Stack Overflow a good example of dual curation? Where they are categorizing the content niche via URL/brand (stackoverflow, onstartups, etc) but then within that the users are curating using tags, etc? And quora is more of pure community curation?William Mougayar I’d love to hear your thoughts on this and pls post link to your blog once you get it out.

    1. William Mougayar

      Thanks Andy. Some thoughts already here from last month.”Curation is a Means to an End, Not the Objective”…New post coming up in a few minutes.

    2. William Mougayar

      Andy, here it is “It’s Official: Curation is Overhyped. 4 Reasons Why.”

      1. awaldstein

        Link is not working William.

        1. William Mougayar

          fixed it. sorry. pls try again?

          1. awaldstein

            will do…retweet as tweet link was broken as well I believe.

      2. Dave W Baldwin

        Well written William. You just have to be patient with all the ‘Me Too’ folks that are vanilla re creativity.

  5. Emil

    Great post. I believe in open curation with no experts. Over time that will bring out the wisdom of crowds which is the ultimate expert. Free market of ideas.

  6. kirklove

    Art stored randomly in a room is a warehouse. That same art curated and distilled becomes a gallery.That’s the power of curation. It make things more digestible. You spend your time appreciating the art instead of combing through it. You may miss a few pieces as a result, but the overall experience is much less time consuming and far more enjoyable.

    1. fredwilson

      that’s a great way to put it kirk

    2. Mark Essel

      The same could be said for technology and design.Technology tucked away in inaccessible undocumented libraries is warehoused potential. Once tech is designed and documented it can flourish into an impactful society shaping force.

      1. awaldstein

        Hmm…wouldn’t you say that it’s still curation and that tech is a tool and design is the organizational principal?The three don’t seem to be equal in my mind with tech and design just serving curation.

        1. Mark Essel

          I was speaking to the relationship between curation and data, being comparable to design and technology. Just a passing thought. Terms with many meanings lead to confusion, what is useful is to apply to context to solidify the concepts and terminology.

          1. awaldstein

            Interesting thought…as this whole discussion.

          2. Mark Essel

            That it is :D, now back to taxes and code.

        2. FAKE GRIMLOCK


    3. Sam Birmingham

      Great analogy, Kirk!This whole curation debate picks up on the discussion in Fred’s “Defence Of Note Taking” piece last week.As I explained at the time (…, what we’re looking for here is a solution to the Context, Audience and Information Overload problems which arose as the consumers of media also became its producers (hat tip: Clay Shirky).Curation is one such solution, albeit that it doesn’t necessarily scale that well if we are relying on individuals to curate our feeds for us.After giving this more thought, I am convinced that we need to look further “upstream” for the solution (……To develop my own analogy (albeit not as elegant as Kirk’s):Twitter is one big lake. We’ve got thousands of species of fish swimming around in there. Sure, if I want to catch a salmon I can wet my line and eventually I will get lucky.Perhaps someone could even filter the salmon for me and I could fish in a pond downstream…But wouldn’t it be great if we could go further upstream, to where all the salmon go to spawn? That’s where I want to fish…

  7. Simon Olson

    My favorite example of social curation to date is Quora. I think that they have really cracked the code on how to implement a curation model that scales.

  8. Rick Bullotta

    Curation of content is as old as the printing press, perhaps older! The owners/editors of traditional print media and their staff have always had a curation process, and with it, an implicit curation bias. This is not inherently a bad thing, but a reality. It is reasonable to expect that the same behaviors will continue with online media, which we of course have seen with HuffPo, Tech Crunch, etc…The hypocrisy comes when some service claims to be “unbiased”. The moment that the human element is introduced into the process, biases find their way in.

    1. fredwilson

      my partner Brad told me a story maybe five years ago. his nephew (just out of college at the time) told him he preferred reading sports blogs to sports pages in the newspapers. brad asked why. his nephew said “everyone has a bias, at least with blogs, you know what it is right upfront”

      1. Rick Bullotta

        Ha. Classic. I think you’re on the mark here, though. Reducing the signal-to-noise ratio in a time of exploding “content” and opinion is critical to the perceived value of a service. Witness the gyrations Google has had to go through in its attempt to maintain the value of its core search service vis-a-vis filtering through content farms. I think you may have found a real business model for an innovative company who can build a general purpose framework + process + algos for content curation.

  9. Derick Rhodes

    Curation is a huge topic for us with the list-oriented social network we’re building, Listgeeks ( At the moment, while it’s still early, it’s important to allow the users to seek out the people they want to follow on their own, but it’s obvious – as with Twitter and Kickstarter – once the noise level gets too intense, you’ve got to find ways to organized for appeal. Great post!

  10. chrisdorr

    Great piece and great comments. Steve Rosenbaum, who has posted several comments today has just written a very good book on the topic. called Curation Nation. Steve is too modest a guy to mention his own work but I am happy to recommend. Here is the link to Amazon to anyone who is interested. Curation Nation: How to Win in a World Where Consumers are Creators (9780071760393): Steven Rosenbaum: Books

    1. fredwilson

      Nice callout on Steve’s book

  11. paramendra

    Tumblr also needs lists.

  12. im2b_dl

    My intuition tells me Color is kind of taking this to a different localized level. My guess is they are going to curate based on location…but curate in a structure that will (my guess) eventually brings in the dimensions of existence (video music, dialogue, sound, etc) to kind of ….recreate the event/location. something tells me “RE-augmented reality” through curation.



      1. im2b_dl

        Agree. & I also think Mozilla is setting up that this will best be done in a browser / cloud structure than in an app. (There work recently on CSS3 and HTML(5) is laying the groundwork). I think that is where some curation investment is going to miss as we head into modules / hubs of identity intersections.

        1. FAKE GRIMLOCK


      2. Dave W Baldwin

        Yes, you are right re Disquis…Disquis no have tech that lead to potential. Found the experts (I am best in Math vs. Machines are stupid) humorous.There are many who will find reality around the corner shocking.Re Color, most do not see what the bigger equation is. They are intimidating the competition (like Micro, Intel and SRI did with VA’s) and in the end, $41mm?…there are enough smartphones out there to get that money back and more, just a matter of marketing fad.

        1. FAKE GRIMLOCK


          1. Dave W Baldwin

            Plans amount to talk and talk is cheap. As said, they do not have the tech…finding out no one else does which places me in the position I thought 😉

          2. FAKE GRIMLOCK


          3. Dave W Baldwin

            Reply to comment re Disquis not hiring Grim… Sorry my friend, you not know the fellow from Cape. The ultimate end run that would have Disquis become its own highway would be done by my team… but what is nice is a highway is a highway, no matter the name.

      3. ShanaC

        Grimlock, apparently color hs been very successful in documenting what has been going on in Japan….

        1. FAKE GRIMLOCK


  13. Simon Cast

    We add

  14. falicon

    Now if I only knew someone who was taking this general approach to the information overload problem… ;-)Note: this was supposed to be filed under ricks point…all I can say is that I am thrilled success is not entirely dependent on intelligence…some day I’ll learn to use my iPad with disqus

    1. sigmaalgebra

      “Now if I only knew someone who was taking this general approach to the information overload problem… ;-)”See:

    2. Mark Essel

      Smarts is for profs.You’ve got super powers when it comes to rapid implementation and design. Uber brains are envious of those skills.

  15. Steven Rosenbaum

    Here’s the thing I’m sure of. The volume of ‘stuff’ (data, content, social media, call it what you will) is on a curve to continue to grow exponentially for the foreseeable future.Eric Schmidt’s quote that “from the beginning of time till 2003 we created 5 exabytes of data – how we create that amount every 5 days” makes it painfully clear.So – we’re not going to go back to the days were information is ‘manageable’ by any measure we’ve previously understood.What that means is that we’ll have to embrace a new world where, what matters most finds you, and you simply won’t be able to ‘keep up’ with everything in your neighborhood, workplace, field of expertise, or hobbies.In corporate America the the endless CC’s of email will force companies to end the practice or grind to a halt. How many of us already get more email that we can read in a day. Fred? *grin*What’s exciting about Curation – and why I stand by embryonic – is that we haven’t even come close to solving the content question. And even as we poke around at it -the number of devices that tweet, blog, post, check in, instagram, and video are just going to explode.The iPad hasn’t had it’s first birthday yet. And iPad2 is a content creation tool.

    1. sigmaalgebra

      “Here’s the thing I’m sure of. The volume of ‘stuff’ (data, content, social media, call it what you will) is on a curve to continue to grow exponentially for the foreseeable future.”Yup. So, to ‘curate’ all that stuff, have to ‘automate’ ‘curation’. Now, how to do that? If very many people knew, then it would have been done by now. So, not very many people know. For a start, see:…”What that means is that we’ll have to embrace a new world where, what matters most finds you, and you simply won’t be able to ‘keep up’ with everything in your neighborhood, workplace, field of expertise, or hobbies.”Or, for each of your interests, have ‘automatic curation’ software be able to work effectively with the ‘meaning’ of both your interest and all the content on the Internet and, then, direct focused content to each of your focused, personal interests. In this way, make the “message” conform again to the “medium”, this time the medium of the Internet, and, thus, take a big bite out of the “message” of old ‘one size fits millions’ of the “medium” of old media.The challenge is how to do that. But I can 100% absolutely assure you that no one anywhere in US venture capital, north, south, east, or west, gives even as much as a weak little hollow hoot about either doing it or how to do it. Instead, they first want to play with the UI/UX and second look at ComScore numbers. The ‘how to do that’ is ignored. This fact, however, is close to inevitable: If many VCs could evaluate the “how to do that”, then too many entrepreneurs would already know “how to do that” and the opportunity would be gone.Another, and a related, issue is, each VC spends a huge fraction of their time looking at average ideas and starts to see the world in what they see in the average while, as we know, both solutions to difficult problems and corresponding successful businesses are quite exceptional. So, the exceptional gets ignored until it’s wildly successful, and then the public image is from ‘Time’ or some such nonsense.There are segments of our society good at attacking difficult problems, getting powerful solutions, evaluating those solutions, and moving to implementation with good reliability, but those segments are a bit far from VC funded Internet entrepreneurship.At least on this thread we are seeing some of the importance and challenges of ‘curation’. If USV would fund a project with some powerful ideas on automating curation, ignoring UI/UX and ComScore, then it would be about the only venture firm in the country to do so.So, except possibly for USV, for business, our discussion of curation here is intellectual self-abuse that nothing will come from, in nine months or otherwise.So, yes, the flip side of this sad situation is a good opportunity: Get a good UI/UX and ComScore numbers, keep burn rate low, and grow organically. Generally, given good ComScore numbers and a low burn rate, between late enough to receive equity funding and too late to need it can be shorter than the time from initial contact to a check in essentially all of the VC funding processes.

    2. fredwilson

      I’m with you on this one steve

    3. Anshuman Bapna

      Completely agree. My take is that because of the content explosion, in at least some corners of the web we’ll see a resurgence of the “middle-men” who’ll do this curation – travel agents for trip planning, headhunters for finding jobs, etc – in a new avatar.This time these middle-men will be truly powered by the web.They’ll have tools to smartly aggregate content / deals from around the web but provide the incredibly important layer of human judgment on top to personalize this for your specific need.We do something similar at – a human guide does all the online research for your travel questions. The interesting thing is that in the majority of cases, someone who’s never stepped outside their own country is planning a backpacker’s trip to, say, Costa Rica. This is possible because what’s needed is the technical skill of finding and making sense of enormous mounds of online data (aided by our tools) plus empathy for how a backpacker travels. These are not skills that come naturally to travel agents of yore. As a result, none of our hundreds of guides is a travel agent in the traditional sense, but simply travel enthusiasts who live & breathe online.My sense is we’re entering a new age of the professional curator on the web.

    4. JBrookeAker

      I rather think of the iPad as a content consumption tool and the apps on it the curation tool that helps me manage my info overload. Use that analogy and you arrive at why Facebook growth is exceeding Google … FB is “somewhat curated” where Google is generation behind.

  16. mcenedella

    An interesting aspect of curation is the difference between product models that enable curation from the wisdom of crowds (“soft-wired”) and those in which the curation is accomplished *by* the product on behalf of the crowds (“hard-wired”).“Soft-wired” sites such as Twitter, Etsy, and Facebook, enable the curation of a thousand clicks. Those things which are liked, selected, or listed by users become the curation output. The sites themselves are not determining what to curate, are silent as to what is good or bad or worthy.For Patch, TheLadders, or TechCrunch the curation will be done unto ye. It is hard-wired into the product experience itself, with the sense of these products being that the curation should be done on behalf of the users by those with specialized knowledge, insight or experience.Then there are the go-between curators such as HuffPo, Amazon, or TripAdvisor where the curation is partially accomplished by what is published, sold or presented by the site, and partially by what is favored by the users. (Blekko may be considered similar, although the steps are reversed – what is favored by the user is part of the input that then produces an output.)Indeed’s JobRoll (a widget for providing continuously updating results against a canned search) is an example of this middle way. The developer tool is soft-wired – you can “roll your own” search against Indeed’s larger database; but this JobRoll then becomes a hard-wired experience for the users of the site on which the JobRoll appears.What will be the characteristics of spaces where these different models along the spectrum will succeed? Which businesses require which level and style of curation?Interesting questions to consider and yet another example of why Internet work is only getting more compelling with each passing year.

    1. fredwilson

      Great comment. Kickstarter is a curated market already. But they still havea need for more curation

  17. Mesut Çelik

    We are implementing somehow biased-curation process for unofficially verified twitter accounts…We implemented the idea first for politicians ( )… very early stageWhat we do is :1- listen tweets from Politicians2- check retweets + replies from their followers to calculate a rating (might be added lots of arguments in future)3- define a top list with the rating…4- distribute curated content to related party fb,twitter accounts etc.What we saw so far is that the idea works really well for some countries and some content (tweets with links) but for some it is really mess.. The main reason of the mess is the volume of the reaction to tweets mostly… If volume is too low, then you cant easily distinguish valuable content. Culture is another big affect…For instance, Some Politicians in Turkey have so many followers who react to all actions taken by politician… However, even if Politician tweets meaningless content or content which has some meaning attached to other tweets are retweeted alot and cause problems in our curation process….I have so many such experience so you can ask me directly @mesutcelik

  18. sigmaalgebra

    To improve the signal to noise ratio, use the law of large numbers!What is that? Suppose n is a positive integer and, for i = 1, 2, …, n, X(i) is a random variable taking on real values. Suppose the X(i) all have the same distribution, that distribution has finite variance and expectation 0. Then(1/n)(X(1) + X(2) + … + X(n))converges (in every sense, probability, L^2, almost surely) to 0 and, thus, ‘goes away’.Regard the X(i) as the ‘noise’; now the law of large numbers has made it go away. What is left is the signal.Doing ‘curation’ of content means getting at and then appraising the ‘meaning’ of the content.Doing this work with humans can be “noisy”, subject to ‘bias’, and possibly expensive.So, it would be good for software to find a way to get at and appraise the ‘meaning’ of Internet content, text, sound, images, video, etc. Yes, one might guess that having software do well getting at ‘meaning’ would be a ‘grand’ problem.Some efforts are based on crude, gross ‘popularity’ (e.g., diggs, likes, links). That approach might be good for the old ‘one size fits millions’ “message” of old media from the old “medium” (McLuhan) but conflicts strongly with goals of ‘personalization’ and ‘focused interests’ of a person’s ‘interest graph’ (not much like graph theory!).Other approaches involve singular value decomposition, speech recognition, struggling with ‘data sparsity’, heuristics, hacking code until get tired or the frig is out of sugared, caffeinated, CO2 water, etc.Something new is needed!How to do that? (1) Get the room quiet. (2) Get a cold can of diet soda with caffeine. (3) Lean back and put feet up. (4) Pop open the cold can and glub, glub. (5) Review what data have or might be able to get. (6) Review what know about means of manipulating data. (7) Have some ideas. (8) Check and confirm the quality of the ideas. Easy!But, to continue, (9) don’t bother telling anyone because nearly no one would believe it or have any means to evaluate any of it anyway! Or for getting computers to do better working with ‘meaning’ as humans understand it, what fraction of the population would be able to do this work or even evaluate the work once it is done? The flip side of this situation is much of the opportunity! More generally, for a large fraction of good things, necessarily not many people know how to do them or evaluate them. (10) Write code, bring up a corresponding Web site, get publicity, users, good ComScore numbers, ads, ad revenue, and take the money to the bank. When the late stage VCs come, be sure never to return any of their contacts! Or, just quote to them, “Dance ’round and ’round and suppose while the secret sits in the middle, and knows”!Uh, the “idea” (that is, the thing that’s “easy”): Have computers do well automating ‘curation’ of all content, text or not, on the Internet. Then, for Act II, use that curation to direct focused content to focused, personal interests and, thus, have a new “message” for the new “medium” and take a big bite out of old media.If can do so well with ‘interests’ (the ‘interest graph’), then might tweak a little and do well with ad targeting? It would seem so.Ah, back to writing some more ASP.NET code! Actually, it’s fun! E.g., just found again the ASP.NETHttpRequest.UserHostAddressI want to write to my Web site’s log file (the one I created, not the one IIS writes!)!The ‘meaning’ software? That’s long since done! It was much easier than making sense out of the documentation for SQL Server administration!Ah, Snow’s ‘Two Cultures’: ‘Writers’, trained in ‘belle lettre’, don’t know ‘hacking’, and hackers don’t know ‘belle lettre’ writing, and neither really knows technical writing. Ah, usually progress encounters some twists in the road!

    1. Dave W Baldwin

      You are getting closer than you think, though you make too complicated. Real meal deal is having the vehicles in place that enable what you see in HD. Good job.

      1. sigmaalgebra

        Thanks.I omitted the (usually implicit) part that the convergence to 0 is as the n –> infinity. Of course it’s the n –> infinity that are the ‘large numbers’. Of course with more assumptions, can say a lot about how fast the convergence is to 0. In practice, it tends to be darned fast, maybe where n a few dozen is ‘large’ enough.For the noise to go away via the law of large numbers, want the ‘signal’, say, Y and the ‘noise’, say, X to be added as in X + Y. So, we observe X + Y. All we want is Y.Can we make signal and noise additive in ‘curation’? Actually, yes. One way is via just the ‘average’ we take in various cases of ‘popularity voting’, but other cases are possible and in my view more important.By “Real meal deal is having the vehicles in place that enable what you see in HD.” I assume you mean that, beyond just trying to do ‘noise cancellation’, it is also important to work to get an especially good, i.e., “HD”, ‘signal’. Yes.One old issue was to take person A, say that they are a lot like person B, observe that person B likes X, Y, and Z, then ‘recommend’ X, Y, and Z to person A. This is not ‘seeing in HD’:First, the ‘matching’ of A to B will be awash in errors. Second, saying that B likes X, Y, and Z is even poorly defined as being relevant to what A will like. Why? A is expected to like X for what? For what purpose or interest of A? For ad targeting, that A might like X for some one of their interests might be enough but not for ‘recommendation’, ‘curation’, etc. So, A may want something for their daughter, mother, sister, for her and husband jointly, etc. Indeed, maybe B actually didn’t like X at all but got it for their father. So, due to such problems, not ‘seeing in HD’.Broadly, using a ‘social graph’ as base data for curation, recommendation, discovery, and search for Internet content (instead of just people) is quite ‘indirect’ and, thus, awash in ‘noise’. So, for curation, …, really want to get at a person’s ‘interest graph’ instead of their ‘social graph’.So, get at their ‘interest graph’. Now we are getting closer to “HD”. Next, focus on a particular ‘interest’. Now we are getting closer. Next, …. Next, …. Now we can focus in “HD” ‘up close and personal’.

        1. Dave W Baldwin

          Ha…actually the HD was a compliment meaning you see a clear picture of some random setting in the middle part of this decade.My piece of advice is to remember the human brain and all habits like emotion, attention span and so forth.With the speed factor delivered, it is a matter of flexibility and doing what the individual user wishes at any given moment.Since we are passing to the other side of the curve where big media tries to make news via sensation and their ratings drop, it is a matter of the public sharing with each other. The vehicles are in place to allow this since there will be those that are curious to opine and/or share about any given person, place, topic and so on.

  19. reece

    i love curation.Twitter is a fire-hose, but when you curate lists within it, it’s a powerful way to filter the (real-time) web.

  20. LIAD

    We’re going through this now with Shoply.We’re beginning to amass a volume of sellers and products where things can become unruly, and to deliver on our promise of ‘efficient discovery’ we need to start putting in place various curation options. They will take the form of a light-touch mixture of a implicit and explicit social curation and algorithms (based on your demo/geo etc).Another aspect of curation which we’re considering is enforcing quality controls on products being submitted (i.e. length of product description,image quality etc) – this is comparable in the art gallery analogy above to deciding what you allow to enter in the first place rather than just re-arranging stuff once it’s there. This is another tool in the marketplace arsenal. – But not one we’ve decided to utilise just yet as it contravenes our value of an open marketplace and creating a ‘shopping democracy’- interesting times.

  21. Guest

    On Navegas you will have curated playlists. (with the direct pendant of artists claimed = offical once). Imagine Lady Gaga content curated from all over the web (youtube, soundcloud, twitter, blogs, interviews, lolcats) organized be the real fans, upvoted by those who enjoy it and directly playable in your new music player. Navegas, = do stuff cool.

  22. Simon Cast

    We added curated groups about 2 weeks ago to PeerIndex and it has been fascinating to watch our users create all sorts of groups. What we are seeing is very much like Kirk’s point about Art being stored in a warehouse versus art being curated into a gallery.In our case people a curating a community. Or more accurately their perception of the community. Some examples of groups being curated are: * Su Butcher’s group on Architects… * Lucy P. Marcus’s group on Corporate Governance professionals…Other groups are simply location based, for example: * Martin Reid’s group on Hampshire, UK…To me one of the most powerful aspect of curation is unlocks knowledge and understanding that is contained in people’s heads which would otherwise be hard to discover or determine algorithmically. Most importantly curation unlocks human judgement. Case in point is the group Technology Elite curated by Jack Schofield…This is Jack’s judgement of who he feels is part of the technology elite. Different people will have different judgements to who is in the community which is perfectly valid. Whether you consider a group or list valid will be partly reflected by your judgement of the person doing curating. I don’t see the expert vs hobbyist as much of an issue as long as people can make judgements about the person doing the curation.I’m really excited to see how our users evolve the use of groups and how we can fold the human judgement into PeerIndex to improve the service.

  23. Bala

    Definition: CuratorCurator (from Latin cura, care), means manager, overseer.Traditionally, a curator or keeper of a cultural heritage institution (e.g., gallery, museum, library or archive) is a content specialist responsible for an institution’s collections. The object of a traditional curator’s concern necessarily involves tangible objects of some sort, whether it be inter alia artwork, collectibles, historic items or scientific collections. More recently, new kinds of curators are emerging: curators of digital data objects, and biocurators.Definition: Expert— n1. a person who has extensive skill or knowledge in a particular field— adj2. skilful or knowledgeable3. of, involving, or done by an expert: an expert jobLets not confuse the words… do I really care about something if I “Like” it? not sure… you need to have skill to be good at something not everyone can be good at everything.

  24. William Mougayar

    My post. It’s Official: Curation is Overhyped. Here’s why.

  25. Dan Epstein

    Great idea for Kickstarter. Curation should help funders and projects find each other. And opening it up to individuals will take advantage of social networks (everyone becomes a taste maker).Perhaps too clever by half, but the step after that seems to me to be helping individuals find curators they’ll like. For example, I’ve started reading several investors because they were mentioned on this blog (msuster, cdixon, and bsabet, to name a few). But I can’t remember how I first got here to AVC. Through .vc I discovered aweissman, who has great taste in music. But who knows how l would have discovered him if I hadn’t started reading AVC, which is an interesting way to find a favorite music blog.

  26. Steven Rosenbaum

    Not angry a bit. Delighted to explore, but let’s just agree to disagree here.My position is that curation is a human endeavor. It’s art, not science.

    1. PeterisP

      My point is that “art is a human endeavour” was true earlier, is only 95%-99% true today, and will become obviously false within my lifetime. There are no practical reasons why machines can’t do art. There are no practical reasons why machines can’t do it better than humans – we just aren’t there. Yet.

    2. sigmaalgebra

      “My position is that curation is a human endeavor. It’s art, not science.”Well, it long has been!”But curation is creative. It is taking diverse, often incongruous things and creating a unique, new creation out of sources and voices.”With that definition, ‘curation’ of ‘art’ is essentially also ‘art’, and for software alone to do such things is for all or nearly all of art not promising now.E.g., I bought a CD of violin music intended to be a good ‘curation’ of some of the best of violin music — best on each of composers, performers, and performances, and ‘curated’ to give some broad coverage, at least from Tartini to Massenet. So, there’s the astounding Grumiaux performance of Tartini’s ‘Devil’s Trill’, the ‘seamless’, polished Kennedy performance of Massenet’s ‘Meditation’, an elegant Oistrakh performance of one of the Beethoven ‘Romances’, and a nicely balanced and mature Szeryng performance of the Bach ‘Chaconne’. It’s a nice CD.’Recommending’ or ‘curating’ mostly via just software just the performances would not be too difficult (indeed, I’d heard the Grumiaux, Kennedy, and, Szeryng performances likely before the CD was issued), but selecting for that CD would be: The person who selected the tracks for that CD knew a lot about violin music.But in the examples Fred gave at the start of this thread, the ‘curation’ was simpler, more of just collecting some content good for a purpose.All such things said, there is a ‘middle ground’: (1) Even if we accept that curation needs humans, it’s not so clear just how much is needed from the humans. In particular can do a lot where what get from humans is much ‘weaker’ than we would expect from a ‘curator’. (2) Even though software by itself cannot do all of good curation, with some seemingly minor input from humans, in a broad range of cases software can do a lot of quite good curating.E.g., for a first cut, suppose we want a ‘curation’ of Heifetz playing Mozart (yes, he did that). So, at YouTUBE just give search:Mozart Heifetzand get 695 hits. As one being a Heifetz fan for a long time, fan enough that when my dorm was next to the music school at Indiana University, I checked out a violin and got started (too late, too little talent, too little practice, but a LOT of fun), just the first page of YouTUBE results actually is a good ‘curation’ as a first cut.Right, it’s just a first cut; we know that.So what was the ‘human’ role? Sure, one way and another people had some Heifetz recordings they liked for whatever reasons and uploaded them to YouTUBE. Usually for an upload, they entered some keywords, e.g., ‘Heifetz’ and ‘Mozart’. I doubt that music critic G. B. Shaw, who made at least one famous remark about Heifetz, was one of the ones doing an upload; that is, the uploaders were mostly not experts; likely only a small fraction of the uploaders even knew what notes the open violin strings play (the first G below middle C and, then, rising in ‘perfect fifths’, D, A, and E).Note: The YouTUBE software didn’t pay attention to the recordings and, in particular, didn’t try to analyze the waveforms. Instead, the YouTUBE software paid attention to data from people who paid attention to the recordings. So, there was data, crucial data, from people. It doesn’t take much of a music fan to recognize Mozart or much of a violin fan to recognize Heifetz from just from a few seconds, but computers will have a tough time recognizing violin at all. The people uploading were not really ‘curating’ and were not experts. Note this point.Is such YouTUBE ‘curation’ all that’s possible for ‘curation’ via ‘software’ with some data from humans? No, not nearly. It’s possible to do much more and much better.How, by humans listening and organizing a taxonomy, popularity voting, Diggs, page ranking based on links? Maybe, but that’s not nearly all that’s possible.What else? Well, some of it’s not yet on the shelves of the libraries.When my software goes live, maybe you will like its abilities to do ‘curation’ along with ‘discovery, recommendation, and search’ for Internet content, both with text and without.

  27. maverickny

    One of the challenges of web 2.0 curation is that data can disappear into the ether faster than you can gather it. For example, we’re only in March and already there have been 3 key cancer meetings that scientists and researchers live tweeted. I planned to gather the data from the aggregated hashtags of all 3 events into a database, parse the data by cancer type and curate the analysed and relevant data before sharing it with the community.The big problem? Twitter only saves data from 8 days ago, arrrgh! The result? A huge amount of valuable information is lost.It’s not lists many of us need in Twitter, but a way to extract information and discussions around hashtags or events. Real time data is less meaningful to scientists like me.

    1. fredwilson

      Twitter has all the data. They just don’t make it available via search. Yet

      1. maverickny

        Ah I see. It used to be available via Summize, but once bought by Twitter the feature oddly disappeared. Can’t wait for the data to be made available again.

        1. FAKE GRIMLOCK


    2. Dave W Baldwin

      You will enjoy the real solution involving delimna. There has been a lot of argument over human vs. machine which really moves attention from true answer.If you could train ten people to sift thru the data of the 3 key cancer meetings and match key points with linked data showing relevence, for example, your life would be easier (more productive).There is nothing to be worried over training ten ‘artificials’ to do that for you.For then, if the ten people mentioned above each trained ten ‘artificials’, you’d be surprised how soon cancer would be a one day outpatient routine 😉

  28. Gytis Labasauskas

    Etsy should allow to add the whole search results list to your favorites

  29. Steven Rosenbaum

    PeterisP”Pandora’s chief curator has not seen my tastes or created anything for me – the machines are curating my playlist for this evening. And they are succeeding. Better than many humans could.”My point was that music, which lends it self to structured data – Pandora has created a taxonomy – a music Genome by organizing over 800,000 pieces of music based on 400 musicological attributes, requires human input. And, with no comment on your musical tastes or mine, the odds are if you like Elvis Costello you probably like Marshall Crenshaw (same period, rockabilly is a cousin of new wave, etc). Which is to say, Human’s at Pandora have created a framework, and then you either like or dislike. But much like movie recommendations – the ‘noise’ isn’t that hard to filter out – Pandora and Netflix both start with ‘pro’ content. But, the web is an unqualified mix of random posts, links, tweets, and ‘pro’, and that’s the power of it’s diversity.It isn’t going to be solved by a computer alone. Even a super smart one. Humans are a required ingredient.

  30. Youssef Rahoui

    Interesting.Curation: a new word for something not that new. It seems to me that there are two types of curation:- People powered: favorites, lists, likes…- Algorithmic: Pandora or… the most famous one, Google!One can imagine that they could be blended (Facebook’s Edge rank). Anyway, if a startup is to provide a curation service, it should IMO, for scalability reasons, favour the algorithmic type or make it user generated.

  31. ShanaC

    Every time I hear duration as a term I think of “7 Days in the Art World.” And I don’t necessarily think of it in a positive manner.One of the basic ways duration works is that therein an in crowd and an out crowd. More entrants into curator land means in some ways a narrowing of creativity as more people have to agree on what is cool. In order to be seen, you will have to spend more time pandering to many curator, none of them powerful enough to take a risk with taste, instead of making art. (It seems to me that is why I often feel there is less great art to be seen in contemporary art land even though there are more talents out there.) You end up not seeing a lot of the really cool stuff.I fear what happens sometimes if UGC goes down that route. Or if code does..

    1. Dave W Baldwin

      Good point/concern. What society has on our side is the rise of the mobile platform. Yes, there will always be the ‘fad’ driven items. It makes you cool that you like whichever thing.At the same time, the circles of those that push the edge, love the fringe, combine these elements to compromise for commercial sake will grow also. To have creative people no longer seperated by geography will truly help.Eventually, the impact will be felt.

      1. ShanaC

        🙂 One of the reasons I chose that example is that the art in question has always been involved with a Jet set crowd, since, well, since the beginning of art.I’m just not totally sure that curation in a multi-polar world means quality, even though it does mean expansiveness of choice….even though you will get more people who like fringe stuff……..Then again, you’re talking to a person who really wants to see what kind of image a robot/computer would take after feeding it raw images of “the best photos of the 20th century)….

        1. Dave W Baldwin

          I understand. Art is just one example of where we change with the ‘in the moment’ sharing which will take us out of our current karaoke music scene (whatever instrument machine and a poem) over to the more fleshy… the same goes for those that try to prove they’re smarter because they know of some app.As we move thru middle decade, the transfer of thought will be so natural we will be able to do much more.Unfortunately, what I’ve seen via comments here, their robot/computer would only show what their programmer wishes, hence the curation problem… on the other hand, the machines over in my camp would simply make a best guess that means something at the moment, changing depending on the mood 😉

  32. Gregg Freishtat

    Fred – Great thread and right up my alley…Interesting the “business of curation” is absent in the discussion. While curating “links” has been the norm for those who care about staying within the law of “fair use”, new models are emerging where sending consumers to numerous curated links/sites just doesn’t cut it from the consumer experience and/or business model for content creators (of any sort). Just as Google unwound the portals and now Facebook seems to be unwinding a bit of Google’s “search as the center of the Internet”, now new and compelling models and consumer experiences are being created which will literally unwind how we think about content discovery and “web sites”.In this new paradigm, content will move to consumers rather than consumers having to move to the various sites and places that the content they desire happens to reside. This means that consumers will be able to explore portions of the net through the lens of brands, publishers, or anyone who makes money by connecting with and serving an audience online. Many new partnerships will be required to broker this content mobility and content will “route” to where it is desired and can be best be monetized. The era of “web site” is coming to an end as the walls of the site become more and more porous allowing content in from other sources and content to go out to other audiences. Curation is the catalyst for some really big changes afoot in the economics of online content.We call this “Content Logistics”. To make it work both Digital Curation and Partner Management are required. Full disclosure — I am working on this full time at Vertical Acuity.

    1. Dan Palevski

      Hey Gregg – I liked your comment and thoroughly enjoyed the video on your web-site. My only additional comment would be that mass-media, which we seem to be focusing on in this conversation, may be best ‘crowd-sourced’ as I believe Steven Rosenbaum is advocating.I think the real ‘value’ for expertly curated content is in the depths of highly specialized information. Getting help in finding the relevant information on moving your equipment manufacturing operations to China is far more valuable than knowing what the most insightful update on the Japanese nuclear situation is today.The second is probably far easier to find today, than the first, given all of the tools already available. The first example, however, would seem to always require specialized skills and knowledge that could be charged for at a premium.

  33. David Haber

    What are your thoughts on curation from the get-go? i.e. Kickstarter has specific regulations as to who can be accepted into the community marketplace.I feel limiting the aesthetic / supply within the context of a marketplace is extremely important for creating the right customer experience. Kickstarter has successfully elevated the entry-threshold so that the vast majority of their projects are of superior quality to their competitors… this brings an aspirational quality to their service.I wonder if this were true of sites like Etsy, Svpply, etc?

    1. fredwilson

      Its a great question. I’ve not been a big fan of Kickstarter’s curatedapproach to building the market. But like david karp’s conviction thattumblr didn’t need native comments, the Kickstarter team has proven mewrong. I’m wrong a lot and the best entrepreneurs trust their instincts morethan mine. And I trust them

  34. Donna Brewington White

    One of the things I love about AVC is how you can read a blog post knowing very little about a topic and by the time you’ve gotten through the comments (and links), you can feel like an expert. I just earned the “curation certification” at AVC U.Great thread!

    1. fredwilson

      Hacking education one post/thread at a time

  35. Steve Poland

    Fred, I think most USV companies have launched before you invest, but if they haven’t, do you spend time with them talking launch strategy? Just reading all the Color PR stuff:

    1. fredwilson

      Very few launch after our investment. I would not recommend color’sstrategy. Too high risk for my taste

  36. whitneymcn

    I’m late to the party on this post, but it’s a topic close to my heart so I’ve got to stop by and think out loud.”Curation” is, in my view, critical to a lot of online services, but I think that we often hold a view of what that word means that’s too limited.Back when Kevin and I created Conversationlists, we took some flak because “Twitter lists should be curated, not automatically generated;” my view of the issue was (and is) that Conversationlists are curated by your actions on Twitter — you don’t know that you’re “curating” when you @reply, but that’s exactly what you’re doing.Taking a look at two examples of Kickstarter curated pages that are close to my heart (pages by the City of Portland, Oregon and by the Pratt Institute), there are a total of fifteen projects selected…and of those only one is still active: the other fourteen have already been funded successfully and closed out.Now I love the projects selected for these curated pages, but the pages don’t really help me find more interesting projects on Kickstarter. They don’t help me become a more active user of the service.But those pages could be a starting point rather than a dead end: are the people who created those pages backing projects that don’t appear in the curated list? Have the creators of the listed projects themselves backed projects? And the people who backed projects on the curated list: are there other projects that they’ve tended to cluster around?Asking people to curate is worthwhile, but accept that most people will have a limited amount of time available to do so. Recognize that the actions people are *already* taking are meaningful to them, and so should be meaningful to you as well.Maybe you can’t always make recommendations, but you can expose connections as a way to supplement and enhance the manual curation. Make use of the data exhaust.The thing that frustrates me a little when looking for projects on Kickstarter is that I never feel like I’m diving down into the rabbit hole: I can browse, but I don’t yet feel like I can discover.

    1. whitneymcn

      Replying to myself because I realized I didn’t address one significant factor: Kickstarter does have to be careful when building to support this kind of discovery: you don’t want creators to feel like the Kickstarter platform is distracting people from their project by pointing out the other pretty, shiny things *right over there,* just a click away.That said, I think Kickstarter can retain its focus on individual projects while adding a little more of the Wikipedia-esque “wow, I don’t remember how I got here, but I’m glad I did” feeling to browsing.

    2. fredwilson

      Damn this is fucking great comment. I agree 1000%

    3. falicon

      Of course I agree a lot with this train of thought…and I have carried this core concept from conversationlist over to the stuff (it’s one of the primary thesis actually – you’ve already do enough ‘work’ within your social networks…let’s get you some more value out of it all)

  37. Victor Cho

    Awesome thread. My friend Charlie Crystle shared the link. I currently run what used to be for Kodak (now We have 12 petabytes of highly personalized images in cloud and the problem of curation is one I have thought about at length as it relates to the overwhelming volume of photos being taken (by you and others). I have some some random contributions to this thinking:How I think about curation (some definitions which have some important nuance):* You have a really big set of thing (Y). * You have a person (X). * The subset of Y that delivers optimal happiness/joy/etc. for X (call is C) is the curated set.What that simple frame, here are some adds:* Finding C small sets is easy. Finding C against volume is the challenge.* C must be indidivdalized to the X. * C changes for X based on time, intent, location, etc. It’s not absolute. * The distributions of C’s tends to be very flat for large sets. So you get a lot of room to deliver something that will be perceived to be ‘wow’, even if it’s not the true global maxima.* What X thinks is the optimal C might not actually deliver the optimal happiness. Algorithms that can figure this out feel like ‘magic’.Victor Cho

    1. kidmercury

      really good comment. thanks for taking the time to articulate.

    2. Dave W Baldwin

      Very wise. It comes down to X calling up what they want at the moment (real time curation) and those things in either C’s or Y (following habits) being delivered.Sure, you can try to do the big engine matched against the X’s or C/Y’s, meaning you force X or the C/Y’s to play by strict rules…….or you develop the appropriate engine that adjusts in between. Which brings us to my area.You seem to get it Victor!

    3. rich caccappolo

      yes, great to see you comment here, Victor – yes, time, intent, location, – I think of them as “state” or “mood”

    4. fredwilson

      You might be able to find a startup in those learnings /insights

  38. Catalin Stefanovici

    Instead of using features in each service to curate, you can now centralize curation of content from different is a service for curation. Curation of any type of objects/things you can thing of. Users define the objects they are curating.

  39. Justin Kuepper

    I think we need to democratize curation more before it is truly effective – that is, curation has to be effectively combined with scoring/ranking/trust algorithms and tailored for the individual consumer.For example, Twitter should ideally weed out spam within the curated lists and give preference to higher-quality users, while also giving more preference to content that it thinks I would like based on my past (ie. users I follow or read).Curation is good now, but not great.

  40. Brian Hayashi @connectme

    The real power of curation is addressability.You don’t want to have to “think” about what to get, these startups have the opportunity to have their name become a kind of short-hand for a specific type of experience.IMO, curation is really a fancy way to say “landgrab”. Like the movie Highlander, there can really be only one authoritative expert in any given field. Once an industry figures out how to use data exhaust to predict consumer behavior, the playing field automatically shifts to favor the participant that has the most comprehensive dataset, corpus, whatever you want to call it.Once upon a time someone invented the term “celestial jukebox” as a way to talk about a hypothetical experience where all of the world’s music would be available on demand. With Napster that hypothetical became reality. Similarly, curation is about making a specific vertical addressable. Since it is impractical to start out of the gate with 100% of everything in the vertical indexed and ready to go, curation is about securing the rights to the most illustrative/in-demand uses of that vertical, and then filling in the gaps over time.As a wise guy at Intel once said, build it, fortify it, then you can leverage it.

  41. falicon

    This topic is def. Inline with the stuff you are working on…what I like about your approach is that you are thinking of curation as a starting point (most are thinking of it as an ending point)…I think you are going to push it to a really interesting place very soon.Oh and thanks for the nod as well :-)P.s. iPad auto correct kills me…

  42. sigmaalgebra

    Okay.I’m anonymous on this blog, but if Fred would like to connect us, fine with me. Yup, Disqus could use a ‘connection’ feature!

  43. Mark Essel

    I see some API tagging swag potential between your apps. There’s certainly orthogonal boosts from finding the best stuff and social search trail blazing.

  44. ShanaC

    I’m not getting interesting links, i only get empty emails 🙁

  45. whitneymcn

    I’m a little reluctant to call it “passive,” exactly, but it’s a fair point:one is taking an action by the user as a curatorial signal, even when they might not (initially) see it that way. Maybe it’s “signalled” curation?Implicit curation, maybe?In the case of Kickstarter, if the goal is to present collections ofprojects that capture or explore particular themes (projects based inPortland, OR, non-traditional approaches to photography, projects exploring communication, or what-have-you) then I absolutely agree that explicitcuration will be far more effective.On the other hand, if the goal is to give users ways to explore projectsthat isn’t based entirely on geography or general category, then implicit curation may be just as effective as explicit. The weird, personal, orquirky connections between projects can expose connections that don’t fit into a more formal strategy.

  46. falicon

    Doh! Not good…I will look into this right away and get the problem fixed for you. Sorry about that!

  47. falicon

    OK – your account is fixed up now…for some reason it had daily emails turned on but specifically configured for zero results per section (I set them back to the proper defaults but you can actually change them if you like -> ‘settings’, then ‘collection summary’).As a side note, you don’t have very many links passing through your account right now (our average user has between 300 and 500 a day)…so there’s a good chance the service won’t surface a lot of interesting stuff for you from within your collection (because you are probably seeing it all already)…but you *should* still be able to get some value out of the ‘outside your collection’ recommendations…In any case, you are a really interesting use case for us right now…so would *love* to chat more with you about your experience and such…I’ll email you (so we don’t hijack the comment board more than we already have here). Thanks! 😉

  48. fredwilson

    Charlie. Let me know if you want an email intro

  49. fredwilson

    I will do it if you send me an email