Hacker News new | comments | ask | show | jobs | submit login
Wolfram Alpha Personal Analytics for Facebook (stephenwolfram.com)
306 points by ecmendenhall on Aug 30, 2012 | hide | past | web | favorite | 103 comments

> all you have to do is type “facebook report” into the standard Wolfram|Alpha website.

.. and connect with your Facebook account, grant extended permissions, signup for a Wolfram account, go to your mail inbox, validate your Wolfram ID, hit a dead end, sign into Wolfram with your Wolfram account, type "facebook report" in the searchbox again, wait 10 minutes for the page to load and finally.. get to see the report (which is nice by the way).

I also had to switch from Chrome to Firefox to get it to work (in addition to all of the above).

I also had to switch to Firefox (after seeing your post here). I tried with Chrome for hours before seeing this, switching, and it worked right away.

You forgot "reactivate FB account" as step one. Although I'm considering doing that to check this out - it looks very cool.

Yes, and "get an internet connection" as step 0. And "have a working computer" as step -1. etc...

I guess I think it's fair to claim its a simple process by assuming you have a login with Wolfram Alpha and Facebook.

I can't even sign in, it's just a blank dialog box.

Mine was a three click process, I was already signed in to Facebook. Go to app (Facebook), Allow access, Create account. Done!

This is what I thought. Load times were brutal. But I do love Wolfram Alpha.

Heh, my experience exactly.

>.. and connect with your Facebook account, grant extended permissions, signup for a Wolfram account, go to your mail inbox, validate your Wolfram ID, hit a dead end, sign into Wolfram with your Wolfram account, type "facebook report" in the searchbox again, wait 10 minutes for the page to load and finally.. get to see the report (which is nice by the way).

I just had to type “facebook report”, connect with my Facebook account, grant extended permissions, signup for a Wolfram account and then wait for the report to load.

The rest you don't have to do (go to your mail inbox and validate the Wolfram ID). Actually, I haven't YET validated the Wolfram ID I got in the mail. It let's you login and use it without validating.

Do you know that feeling when a project/idea you have been working on a long time gets implemented almost exactly as you imagined it, only by someone else?

Well I do now. I'm not sure if I should be excited, or listen to the sick feeling in my stomach.

Edit: Thanks for the positive support! I'll keep working on the project.

This is just another example of a Wolfram product that is great (almost magical) from an algorithmic/data point of view but misses the mark from a design and product point of view. As with most of his products they are clearly designed by (and for) engineer/scientist types.

This is the first time they've done something with W|A that I could see taking off since it's narcissistic, interesting, and custom for each person who uses it. But, the signup workflow is horrible (it requires a magic incantation of "facebook report"), it's a non-actionable, giant information dump that gives me no reason to come back and nothing to do with it. The blog post linked in this is basically a giant manual. I couldn't even read the whole thing. If your product needs a manual, you are doing it wrong. (I didn't even get this far, I am going on screenshots since the sign up didn't even work.)

Mine their product for ideas, and then design one that doesn't suck and that people will rave about. They've done a lot of the heavy lifting of figuring out all the ways to slice the data, now pick out the best that people will care about and aren't meaningless nerd-trivia. You can execute a million times better than this. They are shackled by their thinking both from the fact they are scientists and engineers and they are using this as a way to funnel people into Wolfram|Alpha, which most normal people have no real use for.(They seem to think we live in a world where an average person, when discussing geopolitics over coffee, gets into an argument over the ratio of GDP between Chile and Ecuador, and needs to know STAT, and pulls up W|A for the answer. Except for those whose lives are like The Big Bang Theory, its a small market.)

Build a standalone product that's well designed, curates the data to the most important parts (and doesn't call it "data"), gives them a reason to keep coming back, and is viral and easy to share and sign up for and you will be light years ahead of this thing. Oh, and if it really has something special that people want, even just one thing, guess what, you can probably charge for it too.

As for Wolfram, if he could manage to hire some creative designers and product people, and could cede his ego to their ideas a bit, it's hard to understate how much insane shit they could be building over there.

Edit: For example, it's truly amazing to me that I can still get to a page like this: http://www.wolframalpha.com/input/?i=akljdfalksjdf -- since there is a certain domain W|A can compute that is much, much smaller than the space of all queries, the user should not be able to press enter until they have a valid query. Ugh.

This is hilarious and spot-on at once. After a few minutes of salivating over the data, I came to one thought: Whom amongst the normals I know would give the slightest fuck about this product? Plugged that query into W|A. Answer: Null set.

That was a really great post (upvoted too, but I wanted to say it).

Just curious, why'd you say to not call it "data"?

Because it sounds too cold/impersonal or something? One thing I could also see such a product work well for is awareness, so that people realize not just how much private data they actively put into FB (people are slowly realizing this), but additionally how much, much more information can be extracted that was not intentionally added, but inferred from it with clever algorithms (which is kind of like a side-channel attack, I always thought).

Yeah the "data" thing is just a rule of thumb, users aren't usually interested in data in and of itself, but how they can use it.

Creating a tool that is an expose' of how you are having your privacy violated by Facebook is a noble idea but you run the risk of creating a site that is creepy and discomforting. You run the risk, basically, of people shooting the messenger. So, tread carefully :) For example, if you're going to go that route, make sure people can only see their own results, not other peoples' (and, most importantly, they realize this is the case), and that you provide overt and helpful information on how to lock down their privacy. (Nobody likes to feel helpless, especially immediately after realizing they've been exposed.)

To me a much more attractive proposition is to figure out what parts of the W|A project (or things you've done) are non-obvious but valuable applications of non-trivial data analysis of the Facebook graph that many people would enjoy. Building an experience like this would avoid the creepy "I'm just telling it like it is" factor as well as open up the possibility to sharing each others' results which takes the experience to another level.

Yup, it sucks. That happened to me with Light Table, though I had only been developing very rough prototypes rather than any kind of polished demo.

On the positive side, there is a sense in which it is still overall a good thing. You might have had your thunder stolen, but at least more people are now looking up at the sky, and you've had the opportunity to think deeply about where the winds are blowing.

Thanks, I definitely feel that way too, plus they totally missed some things here which are pretty important to me like privacy, more friend analysis, openness as a platform and predictive models.

I definitely wouldn't stop on a Light Table like thing if you have made much progress. While I am very excited and optimistic about Light Table, there are a lot of choices to make. Especially as they are somehow going to try to monetize it, I think there is definitely room for competitors.

Don't give up. There is a ton of room in the IDE space, and Light Table isn't going to get it all right.

If you create a tool that saves devs 1 hour a week, you're saving a 10 man shop 25k/year.

Why not work on Light Table? http://kodowa.com/jobs

Write an Eclipse or Vim plugin instead, and have a greater impact.

I'd like to, but I suspect this kind of thing: http://imgur.com/ZIVd4 is best done in an HTML-based IDE like Light Table where you can take advantage of SVG.

This happened to me when Yobongo was announced.

I had been working on basically the exact same concept for several months. Mine was named GeoHello. I was building mine as a web app, meant to be used by mobile devices. I felt that a native app would be better, but I didn't have any iOS experience at that point in time, so I went with web. I had the demo finished and even began trying to get people to try it out. In general, everybody thought it was neat. Then, a few days later, word of Yobongo started to show up on TechCrunch as a project that was being started by a team with somebody that was well known in the Valley. They were also building it as a native iOS app.

I was pretty depressed for a little while, but it actually played to my advantage because I got to see how Yobongo tackled many of the chicken and egg problems that a real time location based app faces. There were a few other apps that appeared shortly after Yobongo that were doing the same thing. I'm pretty sure that none of them are around anymore because that's a pretty difficult problem to overcome.

Yeah, I bet that's a difficult problem. It would be pretty cool to see how WA implemented this, unfortunately they don't say. I did get a good perspective of how useful this kinda of service is and how much people are interested in it, whereas before I doubted it would be useful beyond my own curiosity.

Disrupt it :) Now you know what features they've got (and more importantly, what features they don't). I personally don't feel 100% comfortable linking with facebook - so a little pre-scan pep talk would be great. Plenty of other metrics to check out, too. Don't give up.

I agree. As exciting as this looks, I really don't want to give away every permission to WA. Who knowns what there going to do with it. I would like this kind of analysis to be local, thanks for the support!

Totally agree with this. Competition is good - plus you have the advantage of seeing what they worked on. Remember, first mover advantage might be good in some cases - but in a lot of cases the only thing Pioneers get is arrows in the back.

Had this happen with my Master's thesis. Not fun. The positive side is, it was a good idea (at least based on the attention it's getting). It gave me something to compare to as well.

Yeah, it's good to know that you are at least keeping up with the exciting ideas. Better than being completely surprised by it. It's also nice to get some perspective on how other people think about the topic.

Better than creating something that nobody wants or cares about certainly :)

Also, don't let this stop you. If you really feel like giving up though, consider open sourcing it, I suspect some people (myself included) are interested in this, but hesitant to give our info away to some untrusted 3rd party.

I'm not giving up. There are a few things they totally missed here like the fact I have to give away all my data and that it's not platform based (I can't build on it). I would like to be able to ad-hoc query and visualize my social data on my own terms locally. Not much written yet, but I will probably be open source.

Best of luck :)

Yeah - also take some comfort in the fact that though this sounds amazing (and probably is), it's clearly illustrating the fact that it's easier to have an idea, get it kind of working, and make a big announcement about it than it is to oversee a successful launch.

I don't know what's causing the failures (whether it's a problem at wolfram's end or facebook) but I've seen multiple errors every time I've tried this and judging from the comments here in hn, many others are, too. Maybe you'll be first to announce a reliably working implementation?

Sorry, but don't despair! Facebook did the same to a project/idea I started because they didn't. I thought about giving up, but they didn't implement it exactly as my site. I should be glad they made it because it was something I wanted and they could make it faster. If Wolfram Alpha didn't implement what I wanted with this, I hope yours does.

Thanks! There are definitely some features missing here that I would love to see. I'll keep working on it!

I think I would be on the side of excited.

It would be nice to make a batrillion dollars I suppose, but I already have a day job and a growing list of books that I haven't read, and as much as I like coding in the case of my main idea I'm kicking around it's a means to an end, and I would be a happy chap if that end already existed.

I am totally excited (it's a mixed feeling), but this doesn't exactly fulfil my end goal for the project: let me fiddle with my own facebook data.

It's pretty good inspiration though.

Totally :-) You get to see what works and what doesn't

I know how you feel. We have something like this in an upcoming product we're launching.

But that's fine. It's okay for people to compete with you. However if you want to chat with someone else working on similar stuff pre-launch, always love to talk to people working on stuff we're working on - tends to make us all better.

Yeah, even though they are your "competition" they have to most in common with you in terms of interests.

...also people often complement by going after slightly different things which creates an ecosystem all benefit from. Ours is a platform play with this as a simple use case to onboard people, so this isn't the core focus. We'll probably still do it, as implementation is partly done already, but it's good to see others doing similar.

Happened to me when Apple launched the iPhone, which I had designed for my pre-GCSE project in 1996.

Then again when for my actual GCSE project, I designed a electric car in a Lotus Elise inspired body.

That's life though, suffice to say I'm buying a tesla roadster when they're next available and I have the cash.

Think of this way. Some one build the prototype for you. Now go ask everyone showing this and find out if they will pay for this?

The answers will give you an idea how useful this is and whether to pursue it or not.

I'm sure there are ways you could improve upon this.

I'm sure. I would like it to be more of an open platform for Facebook data, and focus more on friends. Plus it should be local instead of trusting other with all your data.

Here's the direct link to try it out: http://www.wolframalpha.com/input/?i=facebook+report

As someone with a long history of incomplete self-tracking projects, this kind of automated collection and analysis is great. (If only I could get the rest of my data in the same place!) What I'd really like to see is a tool like IFTTT for self-trackers.

I've always thought there was a niche market for a complete, secure, self-hosted self-tracking solution, from keystrokes through to email, file system usage, and HTTP traffic.

If anyone knows of anything along these lines, I'd be interested to hear about it, but I suspect it doesn't exist. Probably the difficulty of doing cool things with the resulting data is one of the main reasons such a product doesn't exist.

Can you name some self-trackers that you use?

Sure. I use CRON-O-Meter (http://cronometer.com/) to track my diet and weight, RunKeeper (http://runkeeper.com/home) and Fitocracy (https://www.fitocracy.com/) to track exercise, Moodscope (http://www.moodscope.com/login) to track my mood, and YourFlowingData (http://your.flowingdata.com/) to track anything else I want to measure. My favorite self-tracking service was an MIT Media Lab project called Mycrocosm (http://mycro.media.mit.edu/), but it's pretty much dead. (I wrote a Python script to extract my old data. I'll clean it up and publish it).

Some of these services are completely siloed, some of them export data as .csv or XML, and some of them actively interact with one another (e.g. Fitocracy imports Runkeeper data, CRONometer can connect to a Withings scale). Each tool works well in its niche, but there's no easy way (yet!) to get all the data in one machine-readable place.

What do you use to track your trackers?

Looks really cool, but it keeps timing out on me. Been longing for this kind of information, so here's hoping it will work soon!

This is the kind of thing they need to get into. Data that lots of people care about, data for the masses, rather than obscure details on bolt sizes or ancient currencies. "Compute things people want".

I could only imagine what they could do with celebrity gossip. Or product comparisons.

I am genuinely surprised we have not seen any "big data" articles about celebrity gossip rags. That has to be a market you could data-mine the crap out of for fun and profit.

I can imagine split testing Tom Cruise vs George Clooney to see who generates more revenue.

You could call it quantitative gossip.

I'm pretty sure media companies do that already. They just won't publish it because of competition and PR reasons.

Media companies are actually doing this already, for example to recommend the fitting celebrity for an advertisement ;-)

Just take a look at http://www.celebrityperformance.com/

I would be surprised if FB does NOT have an analytics portal similar to this it allows NSA types to access.

Why would they spend time on an interface for the government?

Not necessarily built in-house. Or, charged for at sufficiently high rates, or used as an internal research tool ("how are people using our product?").

Also, Palantir.

Yeah - I always forget about Palantir!

I'm imagining they must be like the Spanish inquisition of tech companies.

Nice viral way to data mine personal data here.

That should be Facebook's tagline.

That's against FB developer terms of servoce, so would never happen.

Do you think FB would care about that if it were the other way around?

Ah, where did I see the word 'knowledge domain' before? Here: http://www.nytimes.com/2012/08/23/opinion/the-national-secur...

I can't access it now, alpha is very being slow.

The concept reminds me a bit of this chrome extension: https://chrome.google.com/webstore/detail/nangghhladpnhlllol...

I expected this to be very slow. Access to Facebook's API is slow. My Facebook app that analyzes mutual likes is very simple and has just one stat, but it could take a second to query data for every 10 friends. The more friends you have the slower it is. You can't use just one query for all your friends either. You have split queries and wait for each before querying for the next batch.

I made something related to that extension, but instead of counting the number of hits, it keeps track of how much time you spend on different websites: http://webtimerapp.com

Like the author I'm also not really active on Facebook, but I found the clustering of friends particularly insightful.

I can see this extending to Twitter, Linked in and so on, combining everything into a dynamic scorecard. This is what Klout should have been.

(Personal blog post showing clustering: http://lancewiggs.com/2012/08/31/mapping-your-social-network...)

The clustering is so good it's almost scarily good. I wish there was an option for creating friends list automatically from it.

Like Wolfram, I've been doing some personal analytics for a while, but only with email (http://sluggish.dyndns.org/wiki/Emailgraph). Potentially it would be possible to build a Friendica addon which does similar things to the Facebook Report.

That's pretty cool -- I always suspected people out there had done this kind of thing. Nice work!

P.S. You should try out Mathematica sometime, it is a better fit for doing cool things with rich data than C or even Python (disclosure: I work at Wolfram).

Just... wow.

I love what Wolfram Alpha is doing with data-based search results. So innovative and a natural search space that Google is only dipping their toes into.

Google is doing all of this, but not publicly. Through Google accounts, Google+ and even just search cookies, Google has most of the same or similar information, but their goal isn't to provide it to you in a data analysis, but to use it for ad targeting.

It seems concerning that my personal information can be so easily summed up and displayed in easily digestible format. I'm not sure if you can view other people's Wolfram Alpha Facebook summaries, or what data they could be collecting about me... However, here is one study that demonstrates that your sexual orientation can be determined just by analyzing your Facebook friends.


Wow, the coolest part (in my opinion) is the Friend Network section at the bottom. It accurately mapped and clustered every section of my life, including my first high school, the high school I moved to after that, my college, my family, various work groups and social groups, etc. Very interesting to see visually.

All collected in neat little circles.

Maybe Google+ was onto something after all...

This is interestingly non-viral. Although it's one of the coolest links I've seen in a while, it takes much longer to "consume" this, and sharing doesn't have until after consumption. Interesting to note how much less frequently this was shared than some much interesting "popular links".

Yeah, stuff with a scientific bend doesn't get shared, although it has impact. It creates its own niche, with considerably less noise.

Sorry to broaden the scope, but this made me think...

In stock markets, you have the concept of "noise traders", less well-informed traders. They sort of create their own dynamics, sometimes even generating good return on investments, where better informed traders stay out of the market.

Social dynamics... not sure if it's possible to derive some new fundamental laws, because human behavior is too fluid. Maybe in the end (after years of "big data" number crunching) all we get is some empirical observations and weak correlations.

I really liked the Facebook report from Wolfram Alpha. Logging in was pretty easy for me because I already have a Wolfram log-in. I started the process just before a family meeting and walk with my wife, and less than an hour after I started, I see on screen a DETAILED analysis of my heavy use of Facebook, with a lot of information I've been looking for--for example who is the person who comments the most on my wall, or which post that I ever posted has had the most comments. (Hmm, the link referred to in that post belongs HN if it hasn't been posted before.)

On the whole, it is more user-friendly than Facebook itself for telling me about my activity and connections on Facebook, so I'm glad I signed up for the Facebook report on Wolfram Alpha.

Very soon, Facebook needs to get to a point where when you log in it would ask you "what is your target mood: happy, stimulated to do more work, disturbed, relax,happy" and would tailor what you see and do based on past analytics.

This just made personal search a very interesting space.

Put aside for a moment the fact that you're giving your personal data to (yet another) third party. Imagine you tie all your social online stuff to a service that's good at aggregating/displaying the data from each one. Now I could have a 'dashboard' of my online life as well as being able to query it (e.g when/how did I last interact with Alice or Bob?).

I don't know how good Alpha actually is but if I take the visualisations on faith, then I'm interested to know where they're headed. If I were a startup in the personal data/aggregation space, I'd be paying very close attention.

It's hard to ignore that I've just granted someone most of the permissions available to my facebook, but I totally agree that if I could get an aggregation of all my social data in one place and pipe it into some visualizations to create a dashboard it would be pretty formidable.

I just wish I could have that without having to sacrifice my privacy. Why not have a standalone app like the days of old?

> "... if I could get an aggregation of all my social data in one place and pipe it into some visualizations to create a dashboard it would be pretty formidable."

You mean something like this? (or at least enabled by it). http://perscon.net/overview/dataware.html

It's one of the research topics I'm involved with. Any feedback/opinions appreciated (I can pass it on to the folks actually working on it).

This is what I always wanted out of FB.

I look forward to trying this, Wolfram's site is a little slow right now though.

Not bad, the friend network clustering was pretty good (clustered using mutual friends). You can clearly see clusters of people from each place I've worked, the neighborhood I live in, my school friends etc. each in almost their own cluster (and when they aren't neatly organized, there's a very good reason why, like people who moved between jobs with me...aggregating both clusters together).

Strangely, it gets my place of residence wrong.

I really want to use this service and check out the interesting data it generates. But I don't really want to give away all my data to WolframAlpha. It says 'Your information is only stored for one hour, so each time you return, we'll run fresh analytics on your Facebook data.' but I am not really sure. It's encouraging me to come back and let it continuously mine it? I am conflicted.

It pretty clearly says they are deleting your data after one hour.

I am not sure how many have seen http://inmaps.linkedinlabs.com by Linkedin Labs. The graph is much more neater, zoomable, and shows quirky insights about your network.

I have a small bug with it: I set my city in my Facebook profile and Wolfram Alpha says I'm in a city with the same name but in France. But the map shows the right city.

Not only is the analytical part super neat, but this is an absolutely brilliant way to get people to find out about Wolfram Alpha and create an account.

Times out on me! :(

I'm excited to try it, but I'm a bit disappointed they're having such performance issues right now, regardless of the traffic. I would have expected a better launch from Wolfram Alpha.

same here :(

It timed out on me too, then I just hit the "Submit-button" again and all the data appeared

Have to love those people with ages of 99 years.

(At least they deal with the "hide my birth year" people correctly instead of saying they are 1-year-old.)

This was the first time I have ever seen a loading bar go down both percentage and visually. Why on earth would they do that?

The screenshots look great but I can't access the feature at all, the page just times out.

Wow. That's impressive. Big step forward in social analytics.

related: anyone know any good platforms for collecting personal data so you can make cool graphs like this? i use google spreadsheet right now, but im not a fan.

Have you tried Wolfram|Alpha Pro data input features themselves for the analysis?

Funnily enough, many of the visualizations you see in the Facebook scanner have their origin in Wolfram|Alpha Pro scanners (especially the network analysis and visualization stuff), although they've been heavily adapted and tweaked.

getting this { "result" : "failure", "action" : "", "url" : "" }

Applications are open for YC Summer 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact