One interesting thing about quantum computing is that because it is a very new field, a large amount of the research in the field is on the arXiv (interestingly the worst users have historically been computer scientists.) Back in 2006 whenever I would sit around BSing about the arXiv with other quantum computing people, the idea of improvements that would bring the arXiv more up to date would come up. After hearing repeatedly about such ideas, in January 2007, I got fed up of hearing about these ideas and so I sat down and wrote scirate.com, a Digg-like front end for the arXiv. Okay well mostly I did it to learn PHP and Python. Oh, and because coding is fun and I can actually succeed at it as opposed to opened ended research which if hard. Also I did it because I hated spending time filtering through the arXiv each day and wanted to use the power of group knowledge to help save me time. I figure if I add up the time Scirate has saved me versus the time spent reading it I’m pretty close to having gained time. What you didn’t know the point of this blog is to slow down all you competing quantum researchers and thus effectively increase my own effective research speed? 🙂
After some initial development, however, I mostly stopped working on Scirate. Why? Well first of all because I didn’t think I’d succeeded in a very elegant way. Second there was never much traction: there is a group of quantum computing theorists who use scirate fairly often, but outside of that it is not widely used (though there are around a thousand users signed up.) Probably this is also because the development of scirate was essentially closed, consisting of me, hacking away in his spare time. Third, well this thing called a “real job” called (but I keep getting this “hold” music, heh.) I am, however, very proud that until last week, I basically haven’t had to touch the website in any way (last week my host moved Scirates server and didn’t copy over my crontab jobs, thus there is a day missing where I didn’t catch this) besides fixing a few double votes (that occur via a mechanism I’ve never been able to track down.)
So now the question is: what should I do with Scirate?
Some things I’ve been thinking about.
- One problem with Scirate is it’s closed nature. Thus it seems that it would be useful to open up an API for Scirate, allowing for its integrated functionality in other Science 2.0 websites. Indeed I’ve been thinking a bit about a very general framework for the type of functionality Scirate provides, but haven’t mapped the idea out fully.
- I’d like to learn more about Google App engine. Seems like what I do next would be a good opportunity to achieve this.
- One thing that was clearly missing was the ability to use Scirate for some sort of social networking. I’m a bit of a skeptic of “scientific social networking” sites, simply because I don’t see how scientists are all that special in their needs for social networking. Or to say it another way I don’t quite see how a more general social networking tool can’t be “extended” to be useful for scientists, but also be very useful across a wide swath of society. This would imply that I should investigate integration into other social networking sites. But does anyone really want Scirate on Facebook? (Farmville proves to me, however, that I have no idea what people want with Facebook.) And something like LinkedIn doesn’t seem to me to be as widely used as a social networking site (it’s more of a contacts / job site) nor does it allow for extend-able apps as far as I know. Actually this makes me realize that there is a huge hole in the professional social networking genre, though I’m sure that there are people out there attacking this problem. Anyone have any leads?
- There are rumors that the arXiv will soon be accessible in “the cloud.” What sorts of functionality would this allow that it currently missing?
Anyway it seems that I’m due to be working on something new…and yes I know I need to update my iPhone apps as well 🙂
Trust system. Interesting idea.
One thing I think I want to at least bounce around is how to build the guts of scirate without building scirate. That is, are there other places where this basic functionality would be useful.
Perhaps an idea in terms of social features would be an ability to “trust” other users to various degrees, so that when a given user looks at scirate (when logged in), they would get rankings weighted by how much they trusted the opinions particular users Sciting papers. This would incentivise users to log in, but would also mean that users interested in a particular topic would find the interesting papers in that general area floating to the top (since presumably they would be “trusting” users with similar interests).
Sorry, I know this doesn’t help with the whole issue of opening it up, but I thought I would raise it anyway.
Well, it seems quite closely related to the stack-exchange model, though you deal with chronological ordering of material too.
I would have thought one way such a system could be very interesting is as a control for RSS feeds, allowing users to vote up or down different possible sources of content (or indeed different specific pieces), so that if you subscribe, you get just the best picks. An example would be if you were collecting all quantum blog posts, and you wanted away to filter out my crap from the gems you and Scott sometimes post, without having to know in advance that I tend not to post particularly interesting stuff.
Thanks Matt for the suggestions! Lots for me to think about.
I definitely think that scirate needs to integrate with the mainstream social networks in order to succeed, because that is where the people are. A few ideas:
1. Implement facebook connect, OpenID and Twitter login so that anyone with a facebook, google, or twitter account already has a scirate account.
2. Implement sharing buttons for Twitter, Facebook, Buzz, etc. on scirate that automatically do link shortening so that it is easier to share a preprint from scirate than from the arXiv itself.
3. Import mentions of preprints from the social networks to get better metrics, e.g. you could display number of tweets about a preprint as well as number of scites. You could also get metrics from specialist services like Mendeley and CiteULike.
4. Create an “app” for every possible platform, e.g. Facebook, iphone, android, etc. so it’s possible to do sciting from anywhere. Possibly also browser extensions that let you do sciting directly from the arxiv. Even a “scite this” bookmarklet would be useful.
5. Open source the scirate code. I am guessing that you are not planning to make a huge profit from the service and several of us with PHP/Python skills might be able to help out (BTW, I prefer Mercurial).
6. Oh yeah, I almost forgot, Scirate needs its equivalent of Diggnation. Some sort of regular discussion podcast about the latest preprints. I actually think this would be more useful than scirate itself, as we currently have exactly zero physics podcasts at the technical level aimed at researchers, but hopefully it would also drive traffic to the site.
One thing you definitely DON’T want to do is build your own social network into scirate. People don’t want to have to rebuild their social graph on every site they join. Instead, you can use the APIs to read in social graphs from the mainstream sites. Building an open API for scirate is also a very good idea. It will allow people to build 3rd party apps and integration with other services, which will reduce your workload.
The other thing I would suggest is to look very closely at what changes happen in Digg v4 and copy the good ones. They obviously have to make some major changes to deal with the fact that most people are now doing most of their sharing on sites like facebook and Twitter and Kevin Rose probably has smarter ideas about how to do this than me.
I really like Matt Leifer’s comments, and don’t think I can really improve on them. That said, I do want to point out that like Orkut, MySpace and others, LinkedIn is a part of Google’s OpenSocial initiative, and can (in theory) support any OpenSocial-complaint application or widget. I’m not entirely convinced that OS is the way to go, but it is one way.
Oh, and one comment on open APIs… one of the big advantages to Facebook’s recent Open Graph Initiative (debates about the word “open” aside) is that it builds on the large body of work done on Resource Description Framework (RDF). Similarly, something like arXiv filtering and annotation screams out for exposing as RDF. Just a thought.
First of all, thanks for bringing out Scirate. It is fantastically useful.
I like some of Matt’s suggestions, particularly a bookmarklet and maybe an easier url to guess for when you find a paper somewhere else and want to scite it (scirate.org/abs/xxxx.xxxx). In terms of integrating with other sites, don’t just throw buttons on but think about it and do it right. Also:
1. Open-source it. This is the most effective way of leveraging your own work on Scirate. Keeping a website evolving takes a lot of work, and it is too much for one person with other commitments. A less obvious advantage of opening up the development, though, is that the right partners could help enormously at marketing the site to their communities and broadening the site’s usefulness (just like you originally got quant-ph people into it, a personal connection is important).
2. Introduce tagging. Tagging is a lower pressure way of getting people to interact with the site than comments. It can also be used to create statistics, tag clouds, help organize and search, etc. Tags seem to be very important for a lot of web 2.0 sites. Tagging is also a lightweight way of allowing organization. Perhaps you should be able to search your own tags or those of everybody.
3. Introduce curated lists. A collection of authorities should be able to make selections of papers within a certain area (for example, “computational complexity”). This would facilitate arXiv overlays, and would help draw a larger community of people into Scirate. It could be implemented using special tags, or in something akin to Flickr groups.
4. Market the site more. You can do some simple things without spending money. For example, you or a friend can put together a simple flier that individuals can hang on their doors, departments can post on their walls, and maybe a friendly conference organizer might be willing to advertise. This might be easier if the site becomes a community project instead of a personal project. Creating a Scirate badge might encourage people to link to Scirate, which helps marketing through Google pagerank.
5. Clean up the site. I don’t know exactly what to do, but for example “Title:” and “Authors:” aren’t necessary (and the pluralization is often wrong). There has to be a cleaner presentation of the individual papers than the current shrunken text method. Many of the interface elements look a bit sloppy. It is perfectly fine, but it doesn’t give the best first impression.
There are certainly other things. I would concentrate on ideas that are most likely to increase the userbase.
It is good enough for me the way it is. I use it almost everyday [almost never use arxiv.org as the starting place]. For me its main function is to filter…and to leave vicious comments [well once] and read other vicious comments. So out of selfishness, I’d say please leave it the way it is…ok maybe change the horrible color scheme.
I think the prime example [has no nontrivial components] of a successful science social networking site is mathoverflow. I think it is successful for a variety of reasons:
1. mathematicians are more tolerant of/prone to showing off. 2. it is built upon the idea of questions and conversations about questions [very socratic and pedagogic]. 3. mathematicians are far less reliant on research grants than physicists are, so the spirit of collaboration and enlightenment is healthier as opposed to the secret competitions in theoretical physics. 4. The idea of trust and credit is very nicely implemented. 5. again for selfish reasons, they helped me out once when I really needed it.
mathoverflow and stackoverflow are completely different websites though and I am not convinced that they are really needed [despite being successful]. I am not so much for integration with facebook, etc either. is it for me or is it for others? I could always go to the website directly if I feel like it [or the arxiv or gscholar]. The others might see that I have an app called scirate and they might get excited and get it but if they are not physicists what can they do with it? So maybe it’s best to just promote it to the physics community the old fashioned way…
Thanks by the way!
I’ve just discovered scirate.com, and I think I will use it on a regular basis.
As non-user until yesterday, I can already answer to some of the questions John asks just above:
* In the Recs section, it took me some time to find the way the “weights” are computed. The algorithme should be explicited somewhere.
* There is no obvious place to report a bug. And having a first name containing non-ASCII characters, I hit one : Several first-names containing non-ASCII characters are truncated at the first one here http://scirate.com/who.php?id=0710.2597 . Curiously, it’s not the case here http://scirate.com/who.php?id=0801.0979 , with the same authors. While this bug is not critical for the first names, it changes two last names in http://scirate.com/who.php?id=0712.3823 .
Make it easier to stay on Scirate. Author links should search on Scirate. (The author page can later be expanded to include more information.)
Make it easier to use Scirate occasionally. Right now it is difficult unless you check every day or every other day.
Introduce “informal comments” which are also mirrored on Twitter (on @scirate and/or the user’s account), and thus limited to 140 characters (minus space for paper and user IDs, although the Twitter API will soon allow tweet baggage). Part of the idea is to encourage comments. The space limitation should help sidestep the usual academic hesitancy to comment, by enforcing informality.
Give authors more of an ownership stake. For example, link authors to their scirate IDs. Possibly allow authors some sort of special area on the paper webpage, in which they can post an informal introduction and link to talks or other supplementary material. Show registered authors more statistics (since the arXiv doesn’t). Allow authors (or anybody) to submit their papers to the curated lists, which are maintained by moderators.
Allow users to subscribe to multiple arxiv classes on the same homepage. Once crossover users start to scite more, mainstream users will be attracted in.
Keep the interface simple. Simplify it. What do the calendar and arrows at the upper right do? Why are these icons when every other navigational item is text?
Ask non-users what they find to be missing, and how the site can help. Follow a new user through the site for the first time. What is confusing? What can be streamlined?
I like Scirate but I don’t use it because I prefer the ArXiv website feeling and can’t be bothered to visit two ArXivs on a daily basis.
Ideally, the functionality you offer with Scirate (comments, voting) should be built into the ArXiv proper.
The second best thing would be to make Scirate look like an exact clone of the ArXiv (Scirates current look-and-feel make my eyes bleed) with voting buttons, the commenting function, and future options added as unobtrusively as possible.
As to what other people suggested: Tagging, curated lists, a place for authors to provide comments/additional material for their papers (this is what acawiki – another attempt at science 2.0 is supposed to do) and better integration in existing networks seem like good ideas.