Explorsquare
We arrived in Palm Springs really late last night and I woke up too early (I always have trouble adjusting to new time zones). So I got up and went out looking for a good espresso. I'm not a fan of Starbucks coffee and since I only have one cup of coffee a day, I try to make it a really good one.
So I pulled out my phone and launched Foursquare and selected the Explore tab and typed in espresso. It looks sort of like this (I pulled this image off the internet):
I was directed to a place called Just Java where I was able to get a very nice macchiato.
I don't use the explore tab when I'm in NYC because I know where I want to go. But whenever I find myself in a new place looking for something in particular, I've been using the explore tab in Foursquare. And it works pretty well.
And as I checkin more, my friends checkin more, and the places I like to frequent get more checkins, the explore tab will get better and better. This is the power of social metadata at work.
Comments (Archived):
This is where a big chunk of the FourSquare monetization, I believe, will happen.
I was visiting Paris last fall and wished that this was around then. It would have been great to know what around me was highly frequented by Parisians using foursquare to help with finding non tourist cafe’s, restaurants, shops, etc. The explore tab is definitely a great new path for 4sq to follow and one that opens up many ancillary forms of monetization.
Great validation for Groupon’s forthcoming “I’m hungry/bored” app…
There is a ton to get rediscovered in social metadata. I was discussing this with a friend last tonight about how it is very powerful especially between children. (Granted, he was on the opposing side of data usage, law student clinic thing,…) In the end, we came to no agreement over the limits we should use this data. Unlike my friend, I am for more usage, not less…When does social metadata usage become exploitative rather than useful?
Funny, I use Foursquare, Yelp, and a couple of other apps like that when I travel. However, I never use any of them when I’m around areas I know (including to do checkins).
I actually find the explore stuff is *most interesting* in neighborhoods you already know. e.g. I’ve been using it in the East Village to find *new* brunch places and *new* sushi spots… basically as a tool to assist me in breaking out of my regular routines.Very interesting to hear a few comments here from people who don’t think about using it in neighborhoods they feel they already know, but I think this is where the product really shines.-d
If I lived in the East Village instead of a Boston bedroom community there might actually be places to explore that I hadn’t already been to 87 times 😉
this is true. the big cities still need discovering – most others dont.
@dens in the AVC comments!!!!
as he should be if his company is being disqused
totally agree
True. But. It’s the optical rendering I find awkward/dated – even with something as relatively elegant as Disqus…
Shame Disqus doesn’t have an Explore feature 😉
absolutely Carl. Although a comment stream could be perceived as outof context there’s also a serendipitous element to them
i used to think fb would not be a threat here, but i now find goog and fb to be too in this space. in my social graph i see the following breakdown:1. nerds (like me and beyond); tested foursquare, no usage2. trendy marketers (i.e. tech as fashion, but not tech as nerds): foursquare users3. casual internet people, have a smartphone: facebook4. no smartphone: nothingfoursquare is still playing the application layer trajectory….such a risky bet at this stage in the game IMHO. incumbents are firmly entrenched with formidable barriers to entry.there is also the issue that foursquare is simply too hard to use, in my opinion. the value of integrated systems in a mobile environment is immense.not hatin’ though. i like foursquare. they obviously have a lot going for them and many opportunities.
Curious which parts are too hard to use? Particular apps or features?
The whole thing….install, sign up, learn the app….mobile is anenvironment that needs to be far more instant and simple than the immobileweb (laptop, tablet, desktop). Google maps, gmail, gcal, goog places arepractcially effortless. Youtube too. I don’t think mobile apps can competewith native integration and expect that to be a major ongoing issue.
what is the point. Really – what is the point – dont tell me i get a badge.a tiny bit of discovery – ok – but the back has been largely broken on discovery – at least for me – in my town i am not out “discovering”.Unless they pivot directly in to a commerce channel – this is all becomes another layer of semi latent noise…
Mark – I would like to know as well. “Beyond just what is the point?”, how do you monetize it?
Firstly – Get OFF the notion that your app and its features lead to a better way to push to people. Build for pull. Let someone gesture an intent (not a check in – thats not an intent – thats a latent not active commercial gesture) and then envelop them in a world of beautifully created features that offer them or allow them to communicate with what is it they are looking for. That is the future. Its not – oh you are here – let me go and see who is close that can annoy you with stuff that likely has no context at all. Fail.Second – “discovery” is an abstract for commercial behaviour. It should not take 4 steps in a flow to make a purchasing decision. Fail. Part of discovery – is just that – discovery – trying a place and not liking it, the randomness, adds to the human element. Discovery with apps today – tries to compute for an act of randomness that frankly adds to the human experience.FS has clearly done something right there is no doubt and huge kudo’s for that. I dont use it – its too difficult as kid said. But while they are delighting people with beautiful design – build a business around it as well that is frictionless and authentic -Please dot take me down a path to push – that would be so disappointing. (asuming i can figure out how it adds value in my life now!)
i tend to agree, though i think there is a big oppty for foursquare to move into commerce…..but that will be a competitive space, and i don’t think they are positioned to make that transition. but with a large engaged userbase i feel they are still a formidable threat.
the problem is kid – you get people huddled around one use of your service – it is extremely difficult to get the to use it for another – its not in the company DNA.4SQ is known for checking in. how does it excite users about another use?
I get the “hard to use” thing… signup has been wonky, checkin can be slow.Our internal Q1 rally point was “make it feel like every checkin counts”, wich I think we nailed w/ Explore and the new game mechanics (points, leaderboard)Our Q2 rally point is “make the whole thing easier / lighter / faster”. Check out the new signup flow we just launched on the web (!!). A similar streamlined approach coming to mobile and there’s 3-4 prototypes of *super fast* checkin flows making the rounds in the office.#staytuned
i believe that this market will prove once again that the company thatfocuses on a category as a business instead of a feature will prevail
I agree though we likely have different definitions of category and feature.To me all apps are features and categories are determined by communityvalues/niche focus. Ie yammer has the same features as twitter, but in adifferent category. I agree that category focus, which using the definitionI espouse is the same as niche focus, prevails.
LOCATION AS APP FAIL. WHEN LOCATION PART OF PHONE, THAT PHONE WIN.
damn grimlock i think i agree with you but you need to speak up i can barely hear you….oh i forgot i got you downcased! damn! #OHSNAP!lol yes i totally agree with what you are saying…..location app needs to be a part of the phone. not only will it work a million times better that way, but i think there will be big money there, and the phone OS will thus have an extra incentive to build that into the phone in some way, and perhaps block competing apps.
WHAT SNAP? YOU BREAK BONES AGAIN? CALL 911, KID FALL DOWN, HIM NO CAN GET UP.AFTER BONES ALL BETTER, LOOK UP IOS5 RUMORS. MAYBE THEM LOCK OUT FOURSQUARE ETC., MAYBE NOT. EITHER WAY, IT MAKE BIG WAVE IF HAPPEN.
I hope 4SQ save all the explore search queries made and hook them to a user and location.Once they start tracking & publicising completed cycles- will be sick.’Explore’ search query ===> Recommendation ===> Visit/Purchase at recommended locationHardcore data mining and acutely targeted ads – FTWGoogle for the real world – innit.
Absolutely.
Ads – why ads? why do you think people want to be advertised to FFS.you create this cool contextual flow in your comment and at the end you say that the nirvana is that i get “better ads”gotta think better than that.
Touche.I didn’t say people *want* to be advertised to.I just see that as the end game to the explore feature. Some kind of merchant auction-style marketplace.Where do you think this goes?
I’ve been saying this a while – and i have put my money where my mouth is so to speak – we are building for gestures of intent and pull based commerce.I am doing this because in my world of one – thats what i want – my wife uses and how i see buying stuff in the future. I tell the world of what i am looking for – that world is local and realtime and responds with an incentive to tip my choice – the back draft of that world becomes social with a tight concentric circle of close friends who see my purchase and follow me in to a deal.When i travel – i borrow a friends profile and use his pull world to get my stuff.I absolutely dont want to be s zombie with a police light on my head drawing other “28 days” type marketing characters towards me with irrelevant messages.I literally see that world of marketing as just like the movie 28 days.
Wait, do you mean “28 Days Later?” because “28 Days” is a movie about an alcoholic Sandra Bullock going to rehab. And if you really are talking about the Sandra Bullock movie, I just don’t get you at all.
what do you think?
I want advertisements that are vetted recommendations from trustworthy sources, precisely when I need them and at no other time. But ads really only get you so far.What I’m curious about is the not so distant future when customers can conjure products merely by thinking about them. Why not have 3D printers or rapid aggregate software programs just waiting for the moment when individual need meets a willingness to pay. On demand products with a highly tuned range of designs specialized to my tastes. Too expensive? Maybe today it is 🙂
feedback loops are powerful
Rendering the iterative process as being fun is key.
Foursquare nailed it with “Explore.”There are still a few things I want to see for more robust recs, but the human element is the most important to me….
I notice that many places that are well liked on yelp end up converging towards boring, bland and average, which makes sense when you are looking at averages.Seemingly, what this space needs is more personalized recommendations a la netflix or lastfm. Interestingly, 4sq has the data necessary to make a similar system work: frequency of visits, which was crucial for lastfm (frequency of plays). It will be interesting to see how these types of services evolve.
foursquare is last.fm for places
A most excellent comment.So, how do we provide a *personalized* recommendation based on aggregate feedback?I don’t know the answer, but whoever does will soon be the new Larry.
For your last two paragraphs, now, now.Larry who?How to relate the personal and the aggregate? Well, von Neumann has a nice proof of the Radon-Nikodym theorem, and one result is, with very meager assumptions,E[Y] = E[E[Y|X]]So, X is for a person, and Y is the data. So, if for each person X you get data Y and then average over X, then you get the average of Y.So, for each person X, E[Y|X} is the personalized data for X and, then, for everyone E[Y] is the aggregate average.But I can absolutely, positively guarantee you that, your excitement aside, and whatever value there may be there, there is not a single VC in the country willing to pay any attention at all to any such things. Period.Is this stuff important? Well, it is in the collection of ‘ideas’, and as we have heard so often ideas are “worthless”!Is this all? Actually, no! Much, much more is true and important for ‘manipulating’ data to get results, e.g., in ‘personalization’, such as you suggest would be valuable. But, again, all such things are just ‘ideas’ all of which are “easy” and “plentiful”. Instead, if you want something in ‘personalization’, just f’get about ‘ideas’ and just “execute”. Here’s your keyboard: F’get about ideas and just start typiing. Execute!!!For von Neumann, he was just a researcher who had a lot of ideas, clearly all worthless! Apparently one of his ideas was for ‘stored program digital computers’ — clearly worthless. Another of his ideas was for the geometry of a small H-bomb.
Larry Page.Ideas by themselves are worthless.Ideas + ability to execute are valuable.If you can translate E[Y] = E[E[Y|X]] into traction VCs will listen.
Okay, Larry Page. I believe your estimate is low.For how to connect things likeE[Y] = E[E[Y|X]]with ‘traction’, users, ad revenue, ad targeting, earnings, etc., yes, there are ways to know. I have about 40 pages of such math, all nicely typed in with D. Knuth’s TeX so that the math looks good.The math is full of theorems and proofs and convinced me.For ‘traction’, yes, I’ve concluded the same. Or to a VC, some math ‘secret sauce’ and some good ComScore numbers are worth maybe as much as the good ComScore numbers alone.Actually I can believe that the VCs would be much happier just to hear that there was just some ‘software’ without any mention of math: What they want to pay attention to is just the traction; they are used to ‘software’ but afraid of and not used to ‘math’. Then why would I want such a person on my Board?The situation was similar in ‘The Maltese Falcon’ (1941) where, as athttp://webcache.googleuserc…the Humphrey Bogart character Samuel Spade told the Mary Astor character Brigid O’Shaughnessy:”We didn’t believe your story, Mrs. O’Shaughnessy, we believed your two hundred dollars.”The VCs won’t believe anything about powerful ‘secret sauce’ and will believe ‘traction’.Or they believe in Markov processes: The secret sauce and the future of the business are conditionally independent given the current traction. Or, if X is the past, Y is the future, and Z is the current traction, then they believe thatE{XY|Z] = E[X|Z] E[Y|Z]The expression E[Y|X] is ‘conditional expectation’ and is the grown up version of conditional probability.It’s a bit amazing: There must be some function f so thatE[Y|X] = f(X)Then this f minimizesE[(Y – f(X))^2]so that f(X) is, from X, the best nonlinear least squares approximation of Y. Nice. Take all the computer science ‘machine learning’ you want, and for the best nonlinear approximation the secret that sits in the middle and knows is justE[Y|X] = f(X)Did I mention that for the future of computing, math is more important than computer science?Where to use such a thing? Suppose for some user and some ad, event A is that the user clicks on the ad. Suppose we have data X. Then we want to use X to estimate the probability of A, that is, P(A). So, our least squares estimate is P(A|X) = f(X). Our estimate is ‘good’ in that it is ‘unbiased’:E[P(A|X)] = P(A)just what we want. That is, on average we get exactly the right answer. Next, from the least squares, our estimate is ‘minimum variance’. So, it’s about the best estimate we could hope for.The X doesn’t have to be just some one number but can be a lot of data, all the data we have. Indeed, X can be the ‘sigma-algebra’ generated by such data. So, this means of making an estimate is quite general. Thank you A. Kolmogorov.When doing ad targeting, better estimates show up in the bank. That we have unbiased estimates means that when we pair off users and ads and give each user the ads best on our estimate, then we do have (in the context) the greatest ad revenue. The minimum variance helps make our estimates more accurate with also helps revenue.Much more is true.Yes, my work does make progress in ‘personalization’ and associated ‘ad targeting’.The personalization software is ready for production.The software for today is to find why one of my ASP.NET Web pages is communicating with my server with GET instead of POST although I clearly said POST in the statement FORM. Routine stuff.
Ideas alone are not worth millions of dollars. VCs by definition are looking to invest millions of dollars, which means your company valuation must already be in the millions of dollars. Of course they do not want to talk to a guy with some equations. You’re talking to the wrong kind of investor.Build working prototype that proves that you’re onto something, and demo it to some *angel investors*.If you’re right and you can outperform the best machine learning folks (for example, by winning a personalized recommendation contest similar to the Netflix or GitHub contests of the past), you won’t have a hard time raising seed money.Then make a product that people want, get some users, make it better so that they tell all their friends, and *then* hit up the VCs for some rocket fuel.A few slogans for you to consider:”Smart and gets stuff done.””Show, don’t tell.””Genius is one percent inspiration, ninety-nine percent perspiration.”
To be still more clear, you asked if I could connectE[Y] = E[E[Y|X]]and such things to ‘personalization’. Yes. I’ve got the math done and have the corresponding software ready for production.You made the point:”Ideas + ability to execute are valuable.”If you can translate E[Y] = E[E[Y|X]] into traction VCs will listen.”So, can I “translate” the math into “traction”? Apparently I can. All the code is ready for production except a few more Web pages, and am typing those now.You believe that doing well on personalization is important; so do I; and I believe that many millions of Internet users will also. Of course, you and I might be wrong, but it does appear that we have some relatively solid ground to stand on.I did get delayed by some absurd ‘system management and administration’ problems, general and not specific to my work. As my company grows, I will ‘outsource’ that work, e.g., get the best support available from Microsoft, Cisco, etc. and for any issues pick up a phone and have one of their best people walk me through what to do.For the Web page problem I was working on today, just solved that: In the HTML code, there was a missing double quote mark in a STYLE specification!That page is likely the most complicated at my site: It’s mostly ‘dynamic’. So, the usual HTML ‘controls’ of text box, table, etc. are calculated in code just before the page is ‘rendered’ and sent to the user. The way this works is, in ASP.NET, there are .NET software (object-oriented software) classes for each of these controls. So, there are classes for tables, table rows, and table cells. Since a table row is essentially a list of cells, an instance of the row class has a collection of cells, and, thus, uses a .NET collection class.Progress: I’ve got the code writing the page, and now I just saw how to use the classes, especially the collection classes, to extract from text boxes in tables the data entered by a user. Should get that code working tomorrow! It’s routine, but ASP.NET can be fun.What do I have to do to get the “traction”? Finish the code and get some more initial ‘base’ data and load it. Go live. Get publicity and then users, ads, and ad revenue.Grow the number of users and do some special ad targeting.Grow the business.Doing those, about as fast as I can type. So far the difficulties are on the order of a missing double quote mark.Again, at this point the work is easier and cheaper than starting a pizza carryout restaurant.Still, the key, the crucial core, is the math, just the math. The rest is routine. The difference between me and a million hackers who know more about ASP.NET than I do is the math, especially my original math. A million hackers can do most of the computing (I will have some advantages in some advanced aspects of high-end scaling) better than I can; a million business people can do the COO work better than I can.Apparently typically to VCs ‘traction’ is at least 100,000 unique users a month.But my Web pages are simple, say, 1 Mb per page to upload. Then for $55 a month I can get 15 Mbps of upload bandwidth. So, with the 15 Mbps half full, that’s 7.5 pages a second or7.5 * 3600 * 24 * 30 = 19,440,000Web pages a month. There should be 100,000 ‘uniques’ in there somewhere. I’ve implemented some nice logging software with some total sweetheart output, super easy to ‘parse’ and analyze, and that logging output should be able to estimate the number of ‘uniques’.That computing should be from one or a few servers by my left knee in my living room.One server? HP has offered me a loan of a 32 core server.Windows Server and SQL Server? The Microsoft BizSpark program will let me have the needed copies for free for a while.If I put one ad per page and get a click through rate good enough for $2 per 1000 pages sent, then that would be2 * 7.5 * 3600 * 24 * 30 / 1000 = 38,880dollars a month in revenue. At that point I will begin to wonder why I want to take equity funding, convert from an S-corporation to a C-corporation, and report to a Board that doesn’t understand my work, my approach to business, or my business and believes that the math and ‘ideas’ more generally are “worthless”.The $38,880 a month would be much more than I need to get some more servers, get a 100 square foot cage at a high end colocation site a few miles away, and grow to, say, 75 pages a second at which point I wouldn’t walk across the street for a Series A term sheet.Net, my interests are in business, the money making kind, but my view is that far and away the most valuable thing in my business, now and over the horizon, is the math, just the math. Given the math, the rest is routine. Without the math, your prediction of “another Larry” is just a very wild shot in the dark.Making money with math? That’s precisely, exactly, and unambiguously what I had in mind when I went to grad school and selected my areas of study. I’ve never wanted status, prestige, praise, approval, fame, or tenure. Instead my goal has long been a yacht, etc.Again, what the VCs think of the ‘worth’ of ‘ideas’, math, etc. is for my project rapidly becoming moot.To me, the claim that ideas are worthless is outrageous and absurd, so I’ve argued the point here if only as a ‘public service’.But, and with some surprise to me, it’s clear enough that I won’t change any minds in the business community. As I’ve long since concluded, so be it.That so many others believe that ideas and math are “worthless” is one of the best aspects of my business opportunity.
We need to be more clear:You pointed to the financial value of better work in ‘personalization’, e.g., another Larry Page. Call it “another Google”. We’re both interested in how to do personalization.But, as you pointed out and as I long since concluded, at least for a Series A VCs are interested only in ‘traction’, plus maybe market size and team. As I’ve long thought and sometimes written, this situation is so uniform that it seems to have a single cause in common; so my guess has been that these criteria are wanted by the LPs.But just from traction, market size, and team, there is next to nothing to estimate the financial potential you pointed to. That is, the VCs are not interested in the details of a better way to do personalization except via the quite indirect, and inaccurate, means of traction, team, and market size.So, we are saying that for “another Google”, the VCs flatly refuse to consider information crucial for that goal.Yup, sounds like the LPs.But, again, for me, the situation is curious but moot.
Quite a challenge indeed, David – even with people close to me that I have parallels with in terms of interests, business, culture, etc, have very much different expectations to me re: food, bar ambience, art, service standards, what is ‘cool’, etc, etc.Their ‘Good’ can be my ‘Bad’ – and vice versa!Life’s Rich Tapestry…Or, “Nowt as queer as folk” as we say around here! 😉
That’s a great feature, but I would like to see them take it one step further to make it even more relevant.I’d like to be able to slice these Explore recommendations by a) my friends (or see them), b) by coffee afficionados.
On your first question, you can do that already: within the explore tab there’s a little slider control that lets you limit its scope to only your own checkins, your friends’ checkins, or everyone’s — that’s what made it a useful feature for me.And I agree with the second point. I’d love to be able to “tag” people within Foursquare, so that I can filter based on people whose opinions I trust on coffee/food/bars/what-have-you.If there were an established taxonomy for this sort of tag, you could even have interesting inheritance effects: I trust Mike White’s opinion on where to get coffee, so I tag him “coffee,” and he tags the people he trusts on that topic the same way.With that in place, Foursquare could assume a transitive relationship for the tags and let my “coffee” results include places frequented by people that Mike White has tagged as “coffee.” I wouldn’t even have to know who those people are, I could just depend on the chain of trust for a particular domain.It’d be a nice way for me to be able to benefit from my Foursquare friends’ friends and their experience, while allowing me to keep my actual “friends” list down to the people I actually know.
I didn’t know about the slider. Thanks. Maybe it will do what I need.The road to lots of friends’ metadata means my friend networks need to get broader and noisier and I need more filters. Seems counterproductive and messy.What I want to see is the Interest Graph sit on top of the Friend Graph. Both are social but my friends aren’t necessarily the best choice of coffee, for example.I want interest-based interlaced with friend-based. Expert informed where I can then both share with friends and make friends with experts of like ilk.This may be a bit whack but social noise is beating me down today,
“social noise is beating me down today”yet another reason you should be using http://knowabout.it ;-)…sorry shameless plug…I know, I know…but I can’t pass up a lob like that now can I!?! 😉
OK…nudge accepted.
Arnaldo, check out William’s answer below:When I’m looking for the best coffee in neighborhood X, I really don’t care about what my friends say, but want to know the experts opinions. Same for wine recommendations or food stuff.Indeed.
Sorry Kevin, I hit the wrong reply button 😉
Arnold- You hit the nail on the head, and that could lead to the holy grail in social graphonomics, if it ever gets handled eventually.When I’m looking for the best coffee in neighborhood X, I really don’t care about what my friends say, but want to know the experts opinions. Same for wine recommendations or food stuff.Maybe the Chowhounds, Snooth et. al should be selling/passing “expert graphs” that Foursquare can consume. So, I would see a 3rd tab that says “Experts recommendations”. For me, that’s worth more than any other recommendation. So, we’re combining social+local+expert data, and that’s powerful.
That’s a very interesting comment, William, thank you.I’m apparently part of a very small minority who think the whole “what do my friends like” metaphor is bollocks.Your comment goes a small way to making me feel a little less odd.
you and me both
I find it most annoying that I can reply to myself, but not to Richard below – Disqus, come on!To Richard:Attaboy And then there were two.It’s how the best revolutions start…
Thanks. Social graph segmentation and discrimination is perhaps somethingyou develop over time.When I was under 25, I used to think that all my (real) friends were ahomogeneous bunch. Think the Facebook crowd where 40% are under that agebracket and probably account for 80% of the activity there.But, as one grows older, or becomes more experienced, demanding orsophisticated in certain things, you start to want to discern segmentationof “friendships” based on tastes, personal hobbies, professional interests,etc… And therefore I’m jaded like you that the word “friend” is startingto be generalized and means less if you put that into a social settingwithout the right context.
You are right on David.This maps of course to our conversations and it is all starting to congeal and make sense.Really powerful idea as a guideline. Massive win if it can be executed on as a basis for interest based social communities.I do believe that the time is right and that the market is ready for it.
I too will join in the conga line comments to you David, even thought it’s to William below.Yes!Why do software companies keep trying to sell me what my friends are using, how about sell me what I’m looking for, what I need: which is timely, targeted, high quality information.
Really well said William!The distinction between the Interest graph and the Social graph seems artificial. The distinction between the Interest and the Friend graph is where the power lies. Both are social.Facebook is a good example as they will need to surface the Interest graph from its social graph to monetize the platform…which they are doing somewhat already.For marketers and business owners and even for curators, we can use the Interest graph as a design element for connections and the friend graph as a the fuel for viral spread beneath it.This is powerful stuff.Thanks again for articulating this so crisply.
YOU NOT WANT TO KNOW WHAT EXPERTS THINK.IT LIKE AMAZON. YOU LIKE COFFEE PLACE X. YOU WANT TO KNOW PEOPLE THAT LIKE COFFEE PLACE X ALSO LIKE COFFEE PLACE Y.CHALLENGE IS FIND PEOPLE THAT HAVE SAME TASTES AS YOU. MAYBE THOSE PEOPLE NOT EXPERTS.
Yes and No.If I go to a coffee place, it’s for the coffee, not for the people there.Same for food.Challenge is to find experts that can be your friends too. I agree. It takesa lot of work, but experts usually recognize each other and are typicallyopen to one another.
STUPID @DISQUS REPLY FUNCTION. HOW ME HAVE IMPORTANT DISCUSSION ABOUT COFFEE?ME, GRIMLOCK, SAY THERE NO SUCH THING AS EXPERT WHEN COME TO PREFERENCES.THERE ONLY OTHER PEOPLE WITH SAME PREFERENCES.
But depending on whether these people are experts or not, their preferences will be different.I don’t mind disagreeing with other wine experts about both of our preferences, but I will certainly disagree with the non-experts.
bingo grimlock.trying to compute for human semantics associated with friendship affinities is not realistic.you like this coffee – so do i – so what?
WILLIAM NOT UNDERSTAND WHAT GRIMLOCK SAY.MATCH BY PREFERENCE. NOT EXPERT. THEN PREFERENCES MATCH. OPINION OF EXPERT MEAN NOTHING, ONLY MATCH MATTER.UNLESS MAYBE YOU SAYING IF SOMEONE LIKE SAME THING YOU LIKE, THEN YOU CHANGE MIND AND NOT LIKE IT IF THEM NOT EXPERT?THAT SEEM RUDE TO GRIMLOCK.
I didn’t mean the later GRIMLOCK.I like you, but if you don’t know anything about subject X, I will likely discount your opinion.What I’m saying is it would be nice to have another filter which reveals the user’s level of expertise or authority. I think StackExchange has that type of metric built into how they surface content based on a pure meritocracy and feedback response quality.
GRIMLOCK PACK LOTS OF THINK IN SMALL WORDS.TAKE LOTS OF THINK FOR UNPACK WHEN READ.
I couldn’t agree with this sentiment more. I only ask my friends for recommendations because I know they’re more expert on something than I am. If they aren’t, I don’t care what their favorites are. With food as an example, I don’t have a single inner circle friend who could recommend me food that I’d love…I eat differently than they do (I like to think better). In fact, in most cases I’m the ‘recommender’, again, based on my reputation.Like I said on my own blog, reputation is the piece of identity that’s underdeveloped at this point. I want expert/reliable recommendations, and until the reputation element is fine-tuned it seems most attempts rely on ‘celebrity’, which I think misses the mark.
Derek, +1 for agreeing with me. +2 for being in Vancouver (my home town).Indeed, what’s missing from the social graph/interest clutter is the reputation variable. We’re still at the “low-res” stage in social graphonomics. Perhaps the reputation parameter might help to kick it up a notch, and I’m not talking about the Klout type of scoring which is a measure of the effort you put into social or your celebrity status mostly.
Yeah, Klout never smelled right to me, and I spend a lot of time thinking about social capital, credibility, and reputation on a personal level and as it relates to our community of food geeks @foodtree.Reputation is hard, and complicated, but for that reason also crucial to value creation for tomorrow’s technologies. The true challenge I see is taking reputation outside of niche environments and recognizing them on a wider scale. I want to know that someone who’s established a deep history of quality insights and experience over at Chowhound can take that credibility and improve the Exploresquare experience for me. In the same vein I’ve taken vocal issue with integrations that rob communities of this potential or transport it into the closed walls of another community who won’t share its value (Facebook comments on Techcrunch, for instance).Cheers from Vancouver! Beautiful day up here:)
“I want interest-based intersected with friend-based. Sort of expert based where I can then both share with friends and make friends with experts of like ilk.”To some extent I think the classification + transitive trust model I’m thinking about would do that (within the limits of Foursquare, anyway): what I’m talking about isn’t exactly (or maybe isn’t just) a filter.By allowing the transitive property into it (I’ve said Alice is someone I trust about coffee, Alice trusts Bob about coffee, therefore I can see coffee places that Bob frequents when I’m searching for my next fix), I can take advantage of info from people interested in coffee without actually being “friends” with them. I’m adding *new* information sources by exploiting my current coffee-based relationships, in addition to filtering out non-coffee sources.I’m actually a lot more enthusiastic about that than the possibility of accessing the larger set of information from self-identified “coffee people.” I don’t know those people nor have any way to gauge their opinions and tastes, so it’s hard to work with that data set.If none of your existing friends like coffee, then yeah — my model doesn’t do much for you, but even one like-minded friend could connect you to a lot of others, and you’ve got that initial trusted relationship as a quality filter.Man…after writing this I’m really wanting a good cup of coffee…
Thanks for the pointer — it’s a fascinating topic, I’ll get over and readit!
My post that came out of this discussion.Thought you might be interested post this: “Choosing context over friendship to filter the social web” @ http://bt.io/GwKV
Doesn’t the Explore tab already do that? Explore recommendations show who among your friends have frequented a venue and, if you follow tastemakers, they’ll show up there as well.
What is tastemakers?(I did see the explore by friends which I didn’t know about- thanks)
WilliamYou know this site:http://www.foodspotting.com… Pl W, New York, NY 10282, USA/I’m curious how you would think of this in relation to your thought processes around the intersection of interest and friends.There is something here.
The explore tab is the most useful 4sq feature for me. I think helping people find something to do (pre-checkin) is much more valuable than sharing your location or getting a tip (post checkin). Using social meta data to make this feature better is icing on the cake!
Right on Fred. And the “Tips” feature helps you explore further once you’ve chosen a venue.
But it could have been lousy.
well that hasn’t been the case yet with the places foursquare explore hassent me. that’s the power of social metadata in making recommendations
If it’s ‘people who go to independent coffee shops in NYC that you check in @ go to Just Java’, then its better than random (i.e., an Amazon recommends for local) and easier to scale than Yelp and it might have some legs – is that what you are saying? I buy that, although I don’t know that Just Java would ;[)Unfortunately (fortunately?) my palette is not accurate enough to tell the difference between Sbux and other coffee ( and there is no shortage of Green & White in the Valley). Maybe Explorsquare might be limited to what the Brand Mgmt types call ‘overspenders in the category’? Still a good segment.Enjoy the festival.
All I have to say is… http://images1.memegenerato…
Love the “Explore” feature and wrote about it here: http://bit.ly/gJnQ4m (Explore vs Google Places). I find using it in NYC helpful because the location density allows for relevant check-in info about my friends just about everywhere. It is handy for deciding between a few places and also random discovery.They really did a great job with this feature – it make the whole service more useful.
Nice post, liked this line especially:”Not only do users not care about the average, they expect (even hope) that they will deviate from it and that their experiences will be unique.”
Explorsquare really came through for me recently in an excellent use case:Had the gfriend fly out and meet me in San Francisco, where I proposed to her.*Then we went to Sonoma Valley. Used 4[] to suggest only the wineries that had specials or something unique noted about them. Worked fantastic, waaay better than the envelope of paper coupons our BnB host had handed us.*(the existence of that last paragraph implies she said yes– phew!)
wow. congrats!!!!!
Congrats indeed!Makes my 2003 experience seem archaic, now – an era long before this stuff, lol – I proposed to my (now) wife in NYC – at Langan’s brasserie/bar – had earlier in week bumped into a lovely jazz singer at another jazz bar and she said she was playing at Langan’s later in week – so, I had a private word with her and asked if she could sing a song for us just before I popped the question!Wishing you both a long and happy life!
FourSquare should add more stuff like this. Discovery of information based on what your friends are doing, for tourism and other things. Where I’ve Been was a nice app on facebook, and I think it could have been a lot more!
Discovery is a big space that has been pretty much neglected in terms of elegant solutions. I think a lot of what Foursquare, Facebook, and Twitter will do in the future revolves around relative discovery. That’s what we’re focusing on at explorer.io, but limited to the outdoors.
As someone that just moved to NYC, this feature has been *really* useful.
I find it hard to respect a man who restricts himself to one coffee per day – get thee a life Frederico.Wrt to 4sq – you raised an interesting point.At the moment it’s all about deals, but in the long term I suspect locations will become web pages – and curation will become key.
i quite coffee for a decadethen i went to italyand i couldn’t resistit makes me edgy and wiredbut i love the tasteso one a day is all i allow myself
Almost a haiku – a few more cups and you wudda nailed it.
I see a typo in there
Wait till seo folks get ahold of location based recommendations….seems like a recipe for disaster.
I was just talking about this the other day with someone who was saying checkins were dead. The checkin mechanism isn’t the point, this is what many of us have always wanted: to be somewhere new and get insider tips from people they know/trust. I’m seeding data for everyone’s next trip to San Juan Island right now! 🙂
Thank for for those bread crumbs. I would follow your trail most anywheredaryn
Absolutely, Daryn!To say check-ins are dead in this context is to say altruism is dead.Boy, that would be a sad world.I also love to spread the word of a service/bar/restaurant/gallery/whatever that I feel does a great job and deserves recommendations and positive publicity.
Do they have all of the places pre populated?If not, they should think of combining the two …
What’s your favorite coffee place in the Village? Mine is Joe.
It is so cool that FourSquare recommends. I was wondering what are best sites that recommend on the intersection of both the social graph and the implicit and explicit interest graph.Is foursquare building the implicit interest graph; ie if I check in to yoga studies, then I am likely a vegan. Because most people who check into yoga also check into vegan restaurants.Amazon does a great job with its recommendation engine, but I would love to find other examples of the best recommendation engines.
thx to your post i’ve finally realised what foursquare is useful for…finding good coffee.
hahaha. lol.
VALLEY NERDS WORKING ON STANDARD FOR GEOLOCATION DATA.WHAT US REALLY NEED IS STANDARD FOR LIKES.THEN US HAVE ONE PLACE US CAN STORE THINGS US LIKE FROM FOURSQUARE, AMAZON, LAST.FM, GOOGLE SEARCH, EVERYTHING.WHEN LIKE DATA FOR PERSON SHARED ACROSS ALL SERVICES, INTEREST GRAPH BECOME REALLY USEFUL.
love.
Right on the money.Being from the future must rock.
🙂
I have to start using Foursquare a bit more often. I don’t check in at a lot of places. It kind of stopped, since I have to grab my phone and manually press “check in” the moment it becomes automatic with a certain degree of accuracy I think it will take off even more.But I should start using it a lot of more*grabs HTC Desire Android and drags Foursquare to main menu* 🙂
Is it possible for tourists to search for a good coffee in NYC in another language?When the info is in your own language, is easier to feel like a “local”.Maybe my site needs a FS space in Portuguese for tourists in NYC.
Fred, have you been to La Colombe on Church st in Tribeca? It’s awesome.
We’d like to know if you went back to Just Java this morning, or found another place via Explore?
Jamie Flournoy,Since your response to my post is at the maximum depth of the tree permitted in this thread, I’ll respond at the top of the tree.Most of what you note and advise I’ve already discussed at length in my several posts in this thread.You didn’t post anything as bad as ideas are easy, plentiful, and worthless, but in some places you get a little off track.You wrote:”Ideas alone are not worth millions of dollars. … You’re talking to the wrong kind of investor.”No, you are misinterpreting the nature of my contacts with investors: My contacts have usually closely followed the sparse and pragmatic ‘foil deck’ outlined by G. Kawasaki where there is just one foil on ‘secret sauce’. The decks have presented a business proposal and have never suggested that the investors should fund just my math.But, yes, the secret sauce is the crucial core that permits delivering much better results than others. I’m doing the rest, too, but so far it’s been just routine, and my view is that the rest will remain routine. The US is awash in ability to do routine work well. Yes, not every startup team does.Still some ideas alone or nearly so can be quite valuable: E.g., apparently Myhrvold funds high end cooking experiments and, now, a trip to El Bulli in Spain with essentially just ‘ideas’. It appears that for at least some years the lab at Yorktown Heights got enough revenue just from licensing its patent portfolio to fund the lab. There are ‘intellectual property’ (IP) lawyers who can explain some of the ‘value’ of ‘ideas’, and even a venture funded company with crucial, core ‘intellectual property’ will try to protect it. If the company is sold, then certainly the buyer will want the IP, and the seller will want to claim it is valuable. If a US company with US DoD contracts gets some work from another country and leaks out US DoD ‘ideas’, then the US DoD can get ‘unhappy’; it’s not nice to make the US DoD unhappy.”Build working prototype that proves that you’re onto something, and demo it to some *angel investors*.”Here there is a serious flaw in your thinking: Quite generally, even if my company is as successful as I hope, for nearly any investor at all skeptical and serious about ‘due diligence’, there is no very good way now or at any time in future years for them to conclude that I am or have been “onto something”. I know I’m “onto something” because of the theorems. The investors cannot even direct a competent evaluation of the theorems and, no matter what level of financial success I achieve, will never assign any of that success to theorems.Besides, the core software has been running and ready for production for some months. I’ve shown some venture partners some of the output, but that output has been of no interest. That is, the core work is way past a prototype, is running, and output is available but of no interest. The investors don’t want to look at the core software; they just want to look at the UI. But I did the hard work first, left the UI for last, and am doing that now.In VC, it’s an axiom (their story and they’re sticking with it): “Ideas are easy, plentiful, and worthless. Execution is difficult and everything.” So, note: According to this story, at a table for an ‘early stage’ round, the entrepreneur has yet to do much in “execution” so has ‘nothing’! You did notice that, right?”If you’re right and you can outperform the best machine learning folks (for example, by winning a personalized recommendation contest similar to the Netflix or GitHub contests of the past), you won’t have a hard time raising seed money.”For “machine learning”, f’get about that! That work is computer science people who didn’t get very far in pure and applied math and mathematical statistics throwing stuff against the wall to see if any of it appears to stick. On a really good day, some of it will appear to work well for a while (indeed, that’s long been the main criterion of research in ‘artificial intelligence’), but, by analogy, the work is snake oil instead biochemistry. The good people in mathematical statistics laugh at the work in ‘machine learning’.Example? Once I was in an artificial intelligence (AI) project, got sick of that nonsense, took our main problem, borrowed measure preserving transformations from ergodic theory, got a finite group of them, summed over the group, proved some theorems, and got something that knocked the socks off our AI work. Yes, the paper was published; and, yes, now it could be the basis for a good business, say, something more specialized but better thanhttp://loggr.comWhat is it? The total cat’s meow for nearly real-time, ‘zero-day’, multi-dimensional, distribution-free, anomaly detection for more important server farms and networks. False alarm rate is known exactly in advance and adjustable over a wide range. Asymptotically for large data, there are some good things to say about detection rate comparable with the classic, best possible Neyman-Pearson result but with less than half as much data. I may deploy the work in the server farm for my present project. Then I might spin the detection work off as a side business and sell it. I would have pursued that work as a business before except the infrastructure and selling cycle were too grim; both are much easier now.Again, for the future of computing, applied math (sometimes original and sometimes with advanced prerequisites from pure math) is more important than computer science. Computer science profs need to return to school, get a good Master’s in pure math and a Ph.D. in applied math. But they won’t do that.My guess is that the Netflix contest got a lot of flavors of old ‘collaborative filtering’. My math is quite different: The prerequisites are much more advanced, and some of the crucial work is original.How’d I do that? I never even heard of ‘collaborative filtering’ until I had my core work done. For my work, I looked at Last.FM, thought “This sucks”, put my feet up, reviewed some of my best grad school material (some advanced pure math, some of it rarely taught), had some ideas, wrote them out, typed them into D. Knuth’s TeX, decided to make a business out of it, got the core software ready for production, etc.The Netflix goal was not ‘well posed’ and was quite narrow. My ‘business model’ has a goal that is better ‘posed’ and brings in some new data. My work and the Netflix system are not directly comparable, but generally I should do better for the users.The main bottleneck in my project is my working my way for the first time through some thousands of Web pages of documentation at Microsoft for system administration and management and .NET, ASP.NET, and ADO.NET. I’ve done a lot of programming, but this is my first .NET project. Still, the work is only routine and not really difficult. When things work, it’s a lot of fun. When get through the documentation, .NET is a lot of fun.When I started, a good round of equity funding could have let this project move more quickly, but that time has passed. I’ve accepted that I will have to do with my own fingers essentially all the work particular to my project, and there are some major advantages in this approach.Just now, I’m not seeking equity funding. Instead, my posts on this thread are just a ‘public service’ to push back against the claims of “worthless”.Besides, my lawyer has advised me on some tax issues about the time from formation of a C-corporation to getting a term sheet. So, I’m some months away from being ready for a term sheet.A little more generally, as I’ve outlined on this tread, the equity funding issue is getting to be: By the time the investors see what they want to write a check, I may be quite close to no longer being willing to accept a check. Such an example is Plenty of Fish, and it will be curious to see how many more examples there will be.So, there is a short time window when both sides will shake hands. That window is likely shorter than the usual elapsed time from first contact of a venture firm and a check. If my business is anything like as successful as in my planning, then that time window is so short that the probability of a check is small.Besides, as I’ve outlined on AVC, during that time window I will be busy with my growing business and reluctant to agree that as CEO “my main job is just fund raising”.Indeed, in case I do need a check, and to avoid some such ‘crunch time’, I’d like a list of VCs I know are interested in my project and that I’d like on my Board. So occasionally now I send a succinct foil deck to VCs and ask if they are “interested”. Overwhelmingly they are not. It looks like we’re not in related businesses. So be it.In more detail, as I go live, I’d need the check for, say, a server and software from Microsoft? But HP will loan me a 32 core server, and the Microsoft BizSpark program will let me have the software for free for a while.So far I don’t need a check for ‘Raman noodles’.For another analogy, by now my project should take less in time, money, and effort to start than a pizza carryout restaurant. Actually, I look at this whole project, as a business, as not much different than such a restaurant except for Moore’s law, the Internet, and my math.Someone might guess: “But, if you are successful, then you will need a big data center, maybe thousands of people. etc.?”Maybe not: The new FaceBook data center athttp://opencompute.org/serv…is amazing, but from my software timings, server farm architecture, and planning arithmetic, four standard racks ‘on-line’ should be enough for $4.5 billion a year in revenue. With more racks for reliability, backup, staging, development, and test, still we’re talking only a few thousand square feet of commercial space somewhere. I don’t see a need to take over a bend in the Columbia River powered by Bonneville Dam.Besides, Microsoft is able to run a huge data center with automation of nearly everything and only a few people. That’s one reason I’ve stayed with Microsoft instead of some flavor of Unix.I’m not sure that my ‘org chart’ need be larger than some Main Street guy running a few pizza shops. So, mostly the Ben Horowitz notes to CEOs on how to run a big software organization need not apply very well to me. How to manage five pizza shops? Maybe!For the COO work, I’ll hire someone; the US is awash in people who can do such work very well for organizations up to some thousands of people.”Then make a product that people want, get some users, make it better so that they tell all their friends, and *then* hit up the VCs for some rocket fuel.”You seem to be assuming that I’ve been trying to get funded just on the basis of some math. This is part of a general assumption that someone who can do something advanced and specialized can’t also tie shoes and boil water. It’s a bad assumption.My project, proposals, and foil decks have always been for a business and not a ‘research project, math project, or invest in some equations’.For your slogan about showing instead of telling, that’s what investors want. But, it’s not good to do first and plan later. Instead, we should plan first and do later. So, the planning, i.e., telling, should come first.It’s common to denigrate planning because it is not yet doing so can be described as just empty talk, but such denigration is a way to try to insult people and is dumb. There’s another slogan, “Meausre twice. Saw once.” Or “Don’t start vast projects with half-vast planning.”.Besides, as the US DoD has shown very well for 70 years now, it’s quite possible to plan just on paper, to evaluate those plans quite accurately, and, given a good evaluation, to move to good implementation with quite high reliability. My technical background is partly from such DoD work, and I have been surprised that information technology venture capital cannot evaluate projects on paper. Also, at this point, I don’t want my project to have a Board of such people. If they can’t ‘get it’ from paper, then the most I would want from them is their check, and I hope not to need that. As my plane takes off, I don’t want someone in the jump seat who is thinking only with the seat of his pants.For your other slogans, I doubt that they would ‘cut any ice’ with investors. Instead, from all I can tell, the investors are mostly working from some rules on one side of a 3 x 5″ card:General rules: Have a ‘good’ team going for a big market. ‘Stage’ rules: For a seed round, look a the UI/UX and evaluate how well millions of users will like it. For a Series A, look at ComScore numbers. For a Series B, look at audited financials.”Math”? The word may never be mentioned on the Web site of even a single venture firm with the possible exception of the bio of one venture partner who spent some time in a Ph.D. math program.Here I’ve pushed back against “worthless” about as hard as I can and care to and done about my best as a public service.On changing any minds in the investment community, there’s no hope: The successful investors have no reason to change. The rest will soon be “pursuing other interests”.
Guess I’m late to the ball.Reading (quickly), let me put it simple. Everyone is chasing their tail trying to prop up claims a little (read into that what you want).Go out and as the average joe if they would like their own assistant, suggesting where to go per data (expert, age group, gender…) as directed and get you there (directions).Now the phone that does that will be the phone the average joe wants…and nice thing, there are a lot of average joe’s out there.To get the data everyone wants in the end… yes we have to get past FB dictated likes and so on. The way to do this is with the consumer and providing the consumer with something that is flexible to fit their wishes and easy to use. Simple.
Might be a little late here, but if you’re in the mood for Bob Hope’s breakfast of choice/a great early bird dinner, check out Sherman’s (401 E Tahquitz Canyon Way, Palm Springs). Old family favorite.