We’re sorry to announce that tiinker will be shutting down in the next few days.

It’s been great fun and we’ve learnt a heap, but it seems that there just isn’t space for a personalising news aggregator to grow large. We still think personalisation is the future, but the eventually successful implementation will probably be at the news sources, the online faces of the newspapers and media companies, or possibly a large content aggregator. The power of the news brands is strong, with good reason. Future combinations of trusted editorial with intelligent personalisation will provide a very compelling source of information, but for now the traditional media seem reluctant to pursue this approach.

Thanks to all our users for joining us thus far.

The tiinker team.

We’re back online. Flexiscale eventually recovered our server, but we took the opportunity to move tiinker to Slicehost, something we were mere days away from doing anyway. We’ve recently re-implemented tiinker’s data storage with an object database, rather than the traditional relational database. The move was a bit of a rush so we haven’t finished performance tuning yet, but already tiinker is more responsive.

We’ve learnt a lesson with tiinker’s various hosting infrastructures. When moving to Flexiscale we thought the ability to switch instances on and off on demand (and not pay for switched-off instances) would be really useful, but we far overestimated how much we’d be doing that. tiinker experiences a fairly stable load so there’s no need for quick scaling with demand. In particular, there’s no call for scaling back down. We now prepay slicehost monthly, but similar hardware is about half the cost, more than offsetting any amount we could have saved by switching off Flexiscale instances in quiet periods.

Our apologies that tiinker has recently been unavailable. We’re having trouble with our hosting provider and a timezone difference plus the weekend is making resolution a slow process.

We’ll update here when tiinker is back online.

We pushed out another update today in response to some of the feedback we’ve been receiving — specifically to give you more feedback. Next to each story you see on tiinker you’ll now find a little orange bar indicating how interesting we think the story is to you; a little insight into what tiinker is thinking. You can can use this to guide your ratings and help tiinker learn.

You’ll probably notice that you don’t see all the top-rated articles first; a few more lowly rated stories will be scattered around. We do this on purpose to provide some variety between the various topics that have different levels of interest. If we didn’t, you might get all the stories from one topic area in a huge monotonous bunch, followed by a bunch from a second, and so on. We found it to be much more engaging to have stories from a range of (personalised) topics on each page, as discussed earlier.

We also made a great improvement to the look and feel of stories sent by email, and you can now choose to have stories emailed to you daily, weekly, or not at all.

Digg recently announced changes to their algorithm for promoting stories to the front page, citing a need for diversity in content. I used to visit Digg frequently (back when someone else was paying my salary) but over time I found that the type of stories I’d find there narrowed to just a few popular topics, many of which didn’t interest me. Digg has built up an amazing community around their popular-election model of content and has truly changed the world of online news, but any socially-recommended method of content selection finds itself on the horns of a dilemma: you can improve the service to your core community or you can diversify and appeal to more people, but it’s very difficult to do both.

If you’re a regular Digger who’s interests closely match those of the core community members then Digg is just great; an endless stream of stories, images, opinion and occasionally even news. But if you don’t match that profile or have significantly different interests then Digg is far less relevant. Digg’s growth speaks of a very large and valuable community, but growth seems to have peaked around July last year. Perhaps they reached saturation.

They have a problem, though, as evidenced by the backlash of some core Diggers (discussion on VentureBeat and TechCrunch). The algorithmic changes reduce the influence of core contributors, who used to get a hidden bonus to their vote. The changes seem to move the story selection closer to a fair vote and the upset Diggers feel (rightly) that the collective control they had over the community is under threat. This illuminates the problem brilliantly: giving highly contributing members more influence serves that community well, taking advantage of good track records to bump good content to the front page sooner. But this continues to narrow the range content available, and hence broadness of appeal, as more people who like those stories are motivated to contribute and hence gain influence. The changes to diversify content are an attempt to appeal to a wider audience, but they do not serve this core group well. Diversification weakens the community that Digg has built up, but without it appeal is narrow.

There are alternative models of content selection of course, and tiinker captures one of them: treating each user as an individual and learning what they like. Our story selection algorithms pays little attention to how popular an article is but pay close attention to what it’s about and how you have rated previous similar stories. There are middle grounds too, such as that adopted by Reddit, which compares you to other Reddit users and chooses stories based on shared interest. But they too have some narrow popular topics, and the fact that a user’s votes affect content delivered to others opens up the system to gaming.

We think personalisation will be the next stage in our evolving online media, and it seems plenty of big Internet properties think similarly. It’s going to be a lot of fun finding out.

Conoscete già Tiinker? E’ un aggregatore di news che raccoglie le preferenze alla maniera di Digg, analizza i tuoi voti e dopo un po’ impara ad offrirti solo le notizie che ti interessano. Carino.

Well I’ll start of by saying Happy late new years! Been busy since new year working through some new years resolution lists ~_^
Well what kept me busy? Work obviously. Been trying to clear up loads of junk on my PC, backing up files etc. Also had to revise for an examination on the last week, and doing more work on my projects. Examination was okay. There was fair amount to write about, which usually lead to getting cramps on my hand.
No updates posted on these as of yet. All I can say that AGT project is way ahead of it’s schedule and my level 3 AI project is pretty much in sync, although appearing to have some bugs and issues which should be managable. As for Air Roar, I received a nice comment from the guy behind Pyrogine2D and he is eager for me to use their API for Air Roar, although I’m already in the process of putting together my work for XNA, I will eventually try an implementation of this game through Pyrogine API sometime in the near future, but highest priorities to my Uni projects.

Up on my list of todos is to begin putting up my old work on my blog. List includes:

• RPG Game Engine made in OpenGL.
• Some old maps I created for CS:S and Half-Life single-player level.
• Very old but cool console text-based adventure game.
• A number of 3D rendering works.
• Video files.
• A number of C++/C# tools I’ve made since.
• Oh and Rm2k games, which aren’t finished but have nice fully functioning custom battle systems.

Well next section of this post is usually stuff I found on the net.

See these images below:

All done in less than 10 mins, no joke intended.
You can make hi-res wallpapers like these here: http://www.wallpapermaker.net
Free to use wallpaper maker tool. Have fun! ^_^

————————–

Streaming and listening to full albums from bands is never easier if you actually check up on http://www.deezer.com/

For the past week I’ve recently been listening to whole band albums through this site. Nice quality songs. You don’t really need to register for the site to start listening to music. Just add the artist/album in the search file and start listening. Signing up is only for creating your own playlist and sharing your playlist to other members. I haven’t registered therefor I am unsure on whether your playlist can be imbedded.

Another I always recommend is Pandora. I’m sure many of you have heard of this awesome site but for those that haven’t, simply enter in some favourite bands/music and with Pandora it will play songs that sound appropriate to the band/song you’ve entered. Pretty neat and interesting use in querying songs of similar sounds.

——————–

Tiinker

Usually early in the morning I spend a couple of hours reading news, blogs etc and it can sometimes eat more than just a couple of hours. Eventually you’ve spent 3+ hours without even having breakfast(LOL) and you eventually figured out that you’ve wasted most of your time reading inappropriate news. This happens regulary.
Then eventually I read up on http://tiinker.com/ and decided to give this new tool a try.
What it does is simply gives you a number of articles appropriate in different news/blog categories such as science, sports, technology etc and you give thumbs up or thumbs down for each short brief article summary. Depending on these ratings it will eventually spew out articles that interest you most. From here on you can continue fine-tuning your articles.
I’ve seen some improvements as I’ve been using it for the past couple of days. Your feed the articles in RSS and put onto your Google home page, which is very handy for me.

——————–

Thank You Notes

If you ever felt the need to thank someone for something and you can’t clearly put right a good “thankyou” note, you can check http://www.thank-you-note-samples.com. It has samples of thank you notes. I dont know when these would come in handy, but its something I thought was pretty interesting.

Ron

Se via piace Digg, può essere interessante perdere qualche minuto su Tiinker.

I love seeing more and more copycat “intelligent” personalized news sites. The good news is that means that there are funders out there who still know in their gut that there’s money to be made on innovation in the news business. They just need the one idea that will stick. And go pop.

Meantime, more than a six months ago, Mike Arrington wrote about a site called Thoof. Back then, I was also writing and thinking about Streamy and FeedEachOther and other unmemorable twists on feed readers and personalized news sites. No matter their differences, they all seem the same. I just came across yet another—Tiinker—and I just can’t bear it any more.

In his write-up of Thoof, Arrington frames the debate as taking place between two competing positions. He believes that “the masses want popular news,” while the Thoof’s CEO believes that “the masses want tailored news.”

I think they’re both wrong and come at the issue the wrong way.

People want their news based on others’ interests—specialized news from friends (those who have similar interests) and widely popular news from the masses (everyone else). And they want their news based on their own interests, even if their friends don’t share those interests.

Now suppose there’s a continuum of users—from RegularJoe on one end to PowerUser on the other.

RegularJoe wants his news from other people. Although he has relatively few “friends” online, and is thinly connected to the ones he has, he wants them to put in most of the effort to help him get specialized news. (He likes read the “Most Emailed” news articles but doesn’t email them, or he likes visiting Digg but doesn’t log in and vote.) RegularJoe is mostly interested in widely popular news.

PowerUser is different and wants his news mostly based on his own interests. But it would be a mistake to think that he pursues his interests alone (no man is an island, says Donne). He has relatively many friends and enjoys pushing and pulling mutually interesting news to and from them. Of course, PowerUser also has news interests that his friends don’t share or don’t share as strongly, and so he pursue his news independently from his friends as well. Because he enjoys consuming a lot of information, moreover, PowerUser is also interested in widely popular news (he wants to keep his finger to the pulse).

These purely black-box algorithmic personalized news sites don’t really fit either guy.

RegularJoe: They’re too hardcore for RegularJoe. He doesn’t want his own news because his interests just aren’t sufficiently deeply cultivated. RegularJoe isn’t motivated enough to build up a profile by clicking “thumbs up” all the time (as tiinker would have him). When he is motivated enough, he isn’t sufficiently consistent over time for these fancy algorithms to get him what he wants before he strays back to cnn.com because it’s easier to let someone else decide (a person-editor, in this case).

PowerUser: They’re too secret for PowerUser. He wants to put in more effort cultivating his interests and doesn’t want to trust an (anti-social) algorithm from some start-up that might disappear tomorrow. PowerUser also wants to get specialized news from niche groups of friends. For him, the fact that friends X, Y, and Z read some blog post makes it inherently more interesting because they can have a conversation about it (broadly speaking). The personalized news sites just aren’t sufficiently social for the PowerUser who wants to interact with friends around the news.

This isn’t meant to be a slam-dunk argument. I’m not sure about what happens with the group of users who are in the hypothetical middle of the continuum. Maybe there’s some number of users (1) who care enough about the news to have non-trivial interests that don’t shift or fade over time but (2) who also don’t care very much for a transparent or social experience of the news. Ultimately, however, I really doubt that this group of users is big enough to support this kind of personalized news site.

Oleg and I have just released tiinker. It’s now available to anyone, no beta invitations required. In fact, there’s no need to create an account just to try it out.

We got a few good reviews during the beta period, such as these from PCMag’s AppScout, BetaFlow, The Other Review,  bloggers somewhatfrank, rockusnarus, and this hilariously flattering question from Matt Hayden.

Anyway, check out tiinker. I think personalisation is going to change how we consume media. We’re only at the very start now.

One of my fellow Sydney start-ups has just announced that their site tiinker has left beta and gone public.

We are proud to announce that, after slightly more than a year’s hard work, tiinker has graduated from beta testing and been released to the public. No more beta invitations required. It still works just like before — you browse recent news and tiinker will find interesting stories for your based on your feedback.

You can read more on their blog.

Congratulations guys!

We made a tough decision over the past couple of days: we removed a feature from tiinker, discarding weeks of effort and perhaps making tiinker a mite less useful for some users.

It’s for the better, though. We removed the subscribed sources feature, also known as the feed reader. Up until now, tiinker was a hybrid of two applications: a personalised, learning news aggregator and a feed reader. It wasn’t the best feed reader in the world (although we think it is the world’s best personalised news), and it was a feature that only a small fraction of our regular beta users had ever used. It also led to some confusion over what tiinker was trying to achieve, and was causing a couple of technical headaches.

It probably shouldn’t have been there to start with. We initially thought a feed reader would be a great power-user tool, making tiinker better for technically sophisticated users. But our main audience aren’t power users; tiinker is personalised news for everyone, and we’re proud that you don’t have to know a thing about RSS (a standard for publishing streams of content) to use it. Sophisticated users instead use tiinker for their initial news hit, getting lots of interesting stories quickly, but then visit an existing feed reader (like Google’s or Bloglines) for watching particular sources.

So the feed reader is gone, although it will be back, much improved, sometime in the future. This change allows us to focus on the essence of tiinker — the news personalisation that is its core value — in preparation for public release. There’s a lesson in here too, one that’s been espoused many times before: do the simplest, smallest thing that will work first, and add features if required later.

An interesting issue that’s comes up in designing a system that learns which articles a user likes, and then presents those articles, is how to order them. For example, we could order the articles purely by how much we think you will like each article. But this means that you would see a large chunk of your most favourite topic area in a contiguous block, followed by a chunk of your next most favourite, and so on. It’s quite boring for me to get three pages of stories all about, say, motorsport. A selection of stories on different (but interesting) topics is far more engaging. We try to balance the stories presented to you as much as possible while still presenting high quality articles, which poses somewhat of a trade-off. The algorithms by which tiinker orders articles that are presented to you have been by far the most re-evaluated and re-written parts of our code.

Lately we have been experimenting with various ways of doing this, so you may notice a wider selection of articles coming up than you’re used to. We welcome comments regarding this, and any other aspect of our system. We’re constantly trying to find a balance between presenting a wide variety of articles and providing high quality articles and, of course, learning what balance suits you.

I quit my job at the end of last year and have been devoting all my energy to a business/product I started with a partner. We’ve built an intelligent, personalised news aggregator, tiinker. tiinker pulls articles from news sites and blogs from the world over as they are published and then learns what stories each individual user is interested in. It acts both as a means to discover new, interesting articles and also as a filter so you don’t have to waste time reading and ignoring the majority of what’s online.

We’re now weeks away from releasing tiinker from it’s private beta stage (where we limit how many people can use it) and opening it up to the world. You can be one of the cool cats who get’s in before the rush though. Check it out now — we’d love to hear your feedback. Just stick your email address in the “invite me” field and we’ll enable an account for you in short order.

There’s also a blog where you can keep up to date with our progress: the tiinker blog.

Flexiscale suffered a power spike yesterday, and tiinker dropped off the Internet. We’re back up now, after one and a half days or so of downtime. There is nothing more frustrating that having a problem such as this and not being able to do a thing about it, relying on the staff at our service provider on the other side of the world to fix things as quickly as possible.

Anyway, apologies for any inconvenience caused by the downtime. We hope that all of our (rapidly growing!) set of users will continue to enjoy tiinker. We have some sympathy for the very nice people over at XCalibre: Flexiscale is a new service, like tiinker, and we knew it was a risk. No doubt they’ll learn from this and have a more reliable service going forward.

P.S. Our march towards a full release is proceeding well and we cannot wait to share tiinker with the rest of the world. We’re excited!

We have finally begun promoting tiinker to the “general public”, although we are still keeping it as a private beta for a short while so that we have some control over the number of users (to forestall any problems that may be caused by the user base growing unexpectedly fast).

In addition to the basic challenge of actually getting the word out, we also need to make sure that we get the right word out about what tiinker actually does. tiinker is not intended as primarily an RSS reader (although it does that just fine). It is an intelligent, personalised news aggregator. We are aiming tiinker at people who don’t necessarily know what an RSS feed is; users who just want to read news and explore beyond their local paper, but who do not want to spend time filtering out stories which do not interest them.

This is a bit of a catch for us. Many potential early adopters, who are already advanced enough to use feed readers and the like, may initially be unimpressed by tiinker as a pure feed reader. Subscribing to feeds is not core functionality for many of our users; we haven’t focused on the reader, in favour of optimising our site for non power-users. To keep it simple, the reader lacks some of the bells and whistles attached to dedicated feed readers like Google’s. It’s simple, functional, but not a bounty of features.

An additional challenge (which we knew was coming) will be keeping new users interested long enough for them to appreciate the advantages of personalised news. At the beginning, when we knew most of our users personally, we could count on them visiting the site a couple of times, giving the underlying AI time to adapt to their interests. However, we can’t count on that for most Internet users. People are far less likely to come back a second time if the personalisation does not adapt quickly enough, so we’re working now on getting that initial “wow” factor up as high as we can.

These challenges are not new in the land of web startups. Getting users and keeping them isn’t an original problem, but one we must now face .

We presented tiinker to the Sydney Data Miners Meetup group last week, discussing some of the expected and unexpected challenges we have encountered while building it. It generated quite a bit of discussion, particularly around the future of online news and personalisation, and how the technology underlying tiinker could be used in other ways. Quite a few good ideas were thrown up, some of which we’ll be trying out just as soon as we get the chance (thanks data miners!).

Our goal all along has been to create something useful. That means we actually need a finished product — the sooner the better — even if it’s not perfect. At almost every stage in the development of tiinker we have thought of new, better ways to do things. Most of these we’ve had to put aside to be investigated later, a wrenching task for our research-loving brains. After the public release, however, we’ll get back to the whiteboard and the content processing and personalisation technology that drives tiinker will continue to improve.

We’re transitioning tiinker.com over to Flexiscale this afternoon (sorry about the brief downtime). We’ve been running a clone of the site there for a few days and everything looks good. It’s much faster, as predicted, and will give us a lot more room to grow easily.

With this new hosting sorted out, we’re now staring down the barrel of a public release. There are just a few more things we’d like to fix up, along with a refresh of our default source index, before tiinker is ready for the world.

We’ve been experiencing a little grief with the dedicated servers hosting tiinker, so for this and other reasons we’re going to shortly try out Flexiscale, a relatively new pay-as-you-go virtual hosting provider. This means that instead of running on a small number of fairly crappy computers dedicated to us, tiinker will be spread across a number of more powerful computers, but share those computers with other customers. Flexiscale have a utility model of computing that allows us to turn virtual machines on and off at will, paying only for what we use (rather like Amazon’s EC2).

We think this will suit us quite well though. tiinker does a lot of work in the background regardless of how many people are accessing the web site, but Flexiscale should be able to move things around on the fly so that we have CPU available to do it without impacting their other customers. We’ll also be running on much faster disks, albeit over a network, so database access (a bit of a bottleneck right now) should improve.

It’s not all roses of course. Flexiscale don’t yet have a Debian image going, so we’ve been re-packaging all our applications and supporting libraries for CentOS. This has been more of a struggle than anticipated but it’s also helped us generalize our build system. All in all we think it’s a good move, and we’re pleased to be one of Flexiscale’s early customers.

Before tiinker goes public, we’re giving our fantastic beta users the chance to invite their friends to check it out early. Along with a few graphical changes we sent live a couple of days ago (oooh, icons!) you might now notice a big “invite a friend to tiinker” button over on the right hand side of the front page. Clicking that will let you send an invitation code to a friend and they’ll be able to start using tiinker right away. The more the merrier!

You can now also tell tiinker to look out for news from your home region plus any other countries you specify, tailoring your news stream to regions of interest. Other updates over the past few weeks include RSS feeds of your news, clearer marking of rated stories, reorganisation of the front page’s “feature” area and new icons.

It’s quite annoying sometimes when you’re sitting at your computer and you have a really cool idea for a new feature, or a way of implementing an existing feature which is more efficient, more accurate, more bad-ass. Then you snap out of fantasy-land and realise that there are more important things at the moment than writing new features. I have spent the last few days looking for a mysterious memory leak, only to find a mystery crash in the process. It’s frustrating to be looking for a specific error, only to find others; its like you’ve just done negative work, at least it feels that way. Oh well — fixing them feels good.

In other news, I have to say, I’m quite happy with how our beta is going; it seems now like a few of the features which were a bit iffy earlier on in the piece are finally coming together. A lot of our users seem to like tiinker and genuinely find it useful. Hopefully we can fix all of our bugs and implement a few remaining features soon, and release tiinker to the world as soon as possible.

What a busy time! We’ve been enabling a heap of accounts on tiinker recently, and things seem to be working well. We’ve also rolled out a few updates improving the usability and look-and-feel of the site (with another coming soon).

The list of things we still want to do before unleashing tiinker on the public is steadily shrinking. We’ll be looking to expand the user-base a little more rapidly too soon, to make sure things scale ok (current beta users stay tuned!).