phirate

Making a News Bulletin

"@CateOwen l'd love the "not news" whingers to put together a bulletin for us one day. I suspect no-one would watch." - @Hilary_Barry

I thought it was a fascinating question - what would I put on TV for the news slot if I had the opportunity just once?

I think I'd do news, but in reverse. In essence when reporting the news we deliver the exceptions - those things that happened on that day that are so unusual that they stand out. Instead, I would deliver the average day in a way that helped us understand our place in it.

Imagine a visualisation of a home, top-down, in which the people appear as glowing bodies visible through the roof. The day starts with their waking and grabbing a shower etc, and the narrator begins to paint a picture of average family waking (imagine David Attenborough or someone - you'd need a particular kind of narrator I expect), it then zooms out to show the neighbourhood, with all the various people as glows moving around, some have left for work already some are still asleep some are coming back from nightshift etc - all roughly matching the statistical likelihoods for an NZ suburb.

We keep narrating, telling the story about people, all the while expanding out slowly - the neighbourhood becomes a suburb becomes a city becomes a country, eventually we're watching the world rotate below us, billions of tiny glows doing things, different colours sweeping through to highlight various activities, red for war or death, green for humans helping other humans, cars, bikes, farming, factory, office work, drinking coffee, drinking tea, getting wasted on a friday night, dancing, singing, bargaining, praying, running, walking a dog, earning money, spending money, laughing, crying, painting. Every piece either links the viewer more strongly to others ("They're just like me!") or makes them see how different we are on a global scale.

The regular news lacks any real sense of scale - the kid of down the road looks just the same as the UN diplomat, the views of war, disaster or aid are always at the human scale that leaves so much missing. The idea is to put the events of the day in a much truer context, the scale of 7 billion bright, creative minds interacting on a globe of 510 million square kilometers.

We all wonder about this - "Where are they all *going*?" we say as we drive somewhere on a saturday and observe all the other cars around us going…somewhere. We fly into a city and get a glimpse of the true scale of our home town, or more shockingly the scale of a city like London where the buildings stretch on and on to the horizon. And yet the regular news continues to deliver the world in tiny bite sized morsels that serve only to keep us comfortable - no matter how many millions of interesting things happened on earth that day, you will only hear about the tiny number that can be fit into an hour-long broadcast with reporting, and it can become all too easy to believe that this is all that did happen ("Slow news day" - that's not even possible in a real sense anymore).

I don't think this would be the way to report every day, but it would be nice to see it done once, especially by professionals who could take an idea like that and really make it riveting for an hour, a combination of Earth (the BCC series), the TED visualisation talks, Koyaanisqatsi and news reporting that would help expand our perspective a bit (it'd be quite awesome to fit mini-reports into the right places in the scale as well, just to bring the news into the context)

Oh, and me being me it'd be broadcast and webcast at the same time with all kinds of fancy ways of delving into data source and reinterpreting the visuals, but that's probably not strictly necessary.

Fortunately for everyone, nobody will let me do that because it probably seems like a much better idea in my head than it would be in reality :)

What "HashDoS" is / is not

Lots of people running around in a fuss about the whole HashDoS thing. For those not familiar with the issue, let me explain. No, there is too much. Let me sum up:

Imagine you are being given parcels randomly, each of them numbered between 1 and 10,000, and at any time someone might come and ask for a particular parcel. In the first instance you just stack them all on top of each other, but then when someone asks for #888 and it's at the bottom of the pile it takes a long time to get to it to hand it over. So you think for a bit and you split it up into 10 piles. you put anything numbered 1-1000 in the first pile, 1001-2000 in the second etc.

This makes it quicker to get to a given number. Someone asks you for #888 now, you're only looking through an average of 1/10th of the number of items.

In computing terms, the ten boxes and their number guides are called a "hash function" and it forms the basis of many data types. It turns out there's a bit of a headache case though. Imagine that the person giving you the boxes really wants to make your day suck, and imagine that they also know how you've laid out your boxes. All they need to do is keep giving you fake parcels numbered between 1 and 1,000. Those keep going into the same box and it gets harder and harder for you to get anything out of that bin anymore - eventually the pile grows so high that asking for #888 could take you all day.

Getting back on point, the problem this whole "HashDoS" thing talks about is that when you make a web request, you often do the following:

http://foo.bar.com/query?parcel9682=1&parcel1837=2

Those bits at the end, parcel9682=1 and parcel1837=2, are usually stored in the computer memory in a set of boxes using a hash function. The big fuss is that someone has pointed out that if you choose particular values for those keys (imagine "parcel0001", "parcel0002", "parcel0003" etc) the computer has to work very hard to store and retrieve them - for the same reason you would have above.

This isn't a security issue, it's what we call a force-multiplier. On the internet it's a given that any attacking computer can knock any defending computer off the internet, if the attacking computer has more bandwidth (this is a vast oversimplification but you get the picture, the attacker just stuffs the pipe full). However there are a number of force-multipliers that allow an attacking computer with less bandwidth to overload a defending computer with more. The traditional way of doing this is to have the attacker ask the defender to do something very bandwidth-intensive, or very computationally difficult.

Lets imagine for example that you offered a website that lets people on the internet test prime numbers. I could send you a number, and you'd reply telling me whether it was a prime number or not. It turns out that the effort involved in me asking about any given number is enormously less than the effort required by you to determine the answer. This means I can trivially saturate your computer by asking about lots and lots of big numbers - My work is easy, your work is hard.

Anyway, the interesting bit about HashDoS is not that it's a new thing to discover that you can make web servers work hard. Almost any web application has this problem. The interesting bit is how early in the process this happens. Well engineered web applications attempt to prevent attackers from making expensive calls using various practices and defenses, but the HashDoS problem in many frameworks occurs very early in the processing, often before the web application itself gets involved. As a result many applications that might otherwise not be terribly vulnerable to force-multipliers may be vulnerable to denial of service using this technique.

That said, the number of people who care about this is actually going to be very small. In practice the defense for the purpose of the web framework is trivially simple - a HashDoS attack has to deliver a truly ridiculous number of parcels before it makes any obvious difference to the request processing, something that can be easily limited without any impact on the normal operation of the service. More importantly, the vast majority of web applications are already utterly vulnerable to denial-of-service attacks. There is almost no focus on avoiding force-multipliers for any but the most obvious target sites, and the majority of the sites you use on a daily basis offer attackers force-multipliers that are an order-of-magnitude more effective than a HashDoS.

In other words, this is mostly a storm in a teacup. It will result in various web frameworks getting their shit together on this topic, and it may also result in some languages adopting more defensive measures for their hashing functions (perl, for example, turned out to be harder to attack because it semi-randomises its hashing function every time it starts. Historically however, this has only ever granted a platform a temporary reprieve. Nothing short of a truly secure hash function is actually immune to this attack).

New Pledge

@GarethMP wanted to pledge to Kiwis, and I thought that was kinda cool so I came up with a pledge for him and the rest of the MPs to recite in parliament.

As an MP I swear:

To serve all the people of Aotearoa New Zealand as equal individuals, regardless of how much I like or dislike them, how much money they have or do not have, whether they voted for me, live near me, look like me, suffer from a physical or mental illness that dramatically impacts on their lives, or have made and continue to make decisions that are short-sighted.

To not shout, yell, boo, or generally behave like a pre-schooler at any time, most particularly when the school children of our country are visiting parliament to see how their elected representatives and leaders serve their interests.

To retain my faith in the people I represent, ensuring that all information regarding upcoming decisions is available to them so that they can determine for themselves whether their interests are best served by the path I intend to take.

To advocate strongly and effectively for the policy I believe in, and was voted in to achieve - but never to forget that it is the whole of our country and everything in it that is my trust now, not my voters or those policies.

To act with long sight, understanding that short term gain for long term suffering serves only myself and my political interests rather than the people.

To at all times put the maintenance of freedom and democracy as my highest calling, knowing that to fail in this is to act against all of us.

To be courageous when surrounded by fear, cautious in the face of confidence and forgiving when the people are lost to hate.

And finally to remember at all times: each person is a whole person who has lived, loved, felt joy and suffering, and that they deserve my utter respect. Even the media.

Should I fail in this sacred charge, I swear I will wear a neon clown suit for the rest of my days.

Honestly, in writing this although it's pretty much just a long-winded joke I admit I wouldn't mind seeing a citizen-created pledge that MPs would be required to recite along with the regular one, particularly focused on behaviour and relevant high ideals. No doubt it'd turn into a morass of political BS tho.

Hell, for all I know we already have one, but if we do then it's not sinking in.

Make your interview work

Hosting a public interview can be a tricky thing, regardless of whether you're doing it online or face to face.

At buzzumi we've done a few of those now and I figured we'd share some of the stuff we've learned as a result. All of this is aimed at the buzzumi video chat platform but most of it is probably relevant no matter how or where you're doing the interview online.

Preparation

I cannot stress this enough - preparation is absolutely key. This is especially true when there is any kind of technology involved.

When we're looking at running an interview, we make sure we get the whole gig set up well in advance - days if possible. We set up the environment (on buzzumi you can use relevant wallpaper with logos etc), we get the timing all nailed down, we organise a general pattern of promotion and most importantly we get the interviewer and the interviewee into the interview space and we make damn sure everything works.

For any kind of web video/audio broadcast there are a number of subtle things you need to make sure you get right:

1. Ensure you have sufficient reliable bandwidth for the interviewer and interviewee.
2. Ensure you have a good position for you and the camera, so that you aren't back-lit, don't have your laundry in the background etc.
3. Make sure you're somewhere *quiet*. There is no noise cancellation that will deal effectively with a cafe or an open plan office.
4. Ensure your camera and microphone work, that the voice pickup is good etc. If not, go get stuff that works.

When you do this test, it is best to do it at the same time of day, and preferably wearing (roughly) the same clothing as you will for the interview. The position of the sun can dramatically alter lighting conditions, and a bright white shirt has a whole different effect than a dark red one.

And of course on top of this there are all the usual things - make sure you know basically what you're going to be talking about, have backup points in case things are running short, make sure the interviewee has a basic understanding of the topics etc so they can prepare.

In action

With all the preparation, you're now in a good position to make the interview a success, but Murphy is still looking for an excuse to make your life miserable. The most important bit is that the interviewer and interviewee are in the space *early*. Nothing looks less professional than arriving late, and worse if there are any last-second technical problems it's impossible to address them without time.

You also want a private channel for the interviewer, interviewee and technical support. In the event that there are problems, you don't want to be talking about it in the public channel - get it sorted in private.

Lastly, make sure that both the interviewer and interviewee close down everything else on their system prior to the interview (at least everything that makes noise). Preferably do a fresh reboot before start. Constant beeps from various software does nothing to enhance the experience, and some apps may interfere with the camera or audio at the wrong moment.

Ok, all the tech is working, you start your interview, it's going great and then ... someone asks a question in the text chat.

Interaction

Interaction in an online interview, particularly on something like buzzumi that gives participants a text chat space to work with, can be a delicate balance. Much depends on the interviewee and the topic - there is no question that a comment at the wrong time can derail a really interesting answer.

As a general rule of thumb, the interviewee should probably not be responding to participant questions - the interviewer should act as a moderator in a sense, picking up questions from the text chat and posing them to the interviewee at an appropriate moment. Don't get too fixated on this however - some interviewees simply love audience interaction and will be eager to engage directly.

In some cases, the questions can present an interesting opportunity to bring a third participant in for a moment. In one particularly memorable case in a recent interview one of the viewers wanted to ask a question about a newspaper article. Using buzzumi we were able to temporarily enable their camera and let them show the article to everyone. The lack of preparation for a random participant means that it may not work, and the interviewer should keep that in mind and be ready to move on promptly if it fails, but the added dimension it can add to an interview should not be discounted.

Closing up

Once the official interview is over, it is worth considering sticking around for a few minutes to see whether further discussion evolves from the participants. While it's possible to do the old "And now we'll take questions" format, this really isn't necessary in an online, asynchronous environment - people with questions can pose them at any time while others are being discussed.

Once this is over, remember to close up cleanly - close the cameras first and then leave the chat space open for a while to allow any post-event chatter to come to a conclusion (If it doesn't look like it's going to finish you can always leave the chat space open). Once it's done, close the room completely.

Hopefully, at this point you get to feel that lovely glow when something done on-the-fly goes right. Remember to thank your interviewee!

When Skype is just Too Hard

Sometimes we just have to talk face to face. When there's too much emotion involved, or it's basically a chat where the words matter less than your ability to decide whether someone is trustworthy/friendly, the presence of audio and video in realtime can make all the difference.

Skype and other real-time chat solutions went a long way to making this a reality over the internet. The video and audio quality is sufficient to substitute (mostly) for a physical meeting. Unfortunately, sometimes it's just a pain.

If you've used Skype for this kind of problem you're familiar with the Skype Dance. This is the one where you suggest meeting on Skype, the other person asks what your skype ID is, you tell them, they find you and request approval, you authorize them, they call you and finally you get to talk.

That's what happens on a good day anyway. On a bad day, you or they have several skype accounts and are logged into the wrong one, or got the skype ID wrong, or one of you doesn't even have skype installed and the delays just stack up.

The project I'm working on at the moment solves this (and several other) problems, reducing the time-to-face dramatically. What follows is a simple tutorial to get you signed up with Buzzumi and into a One-to-one chat.

The key to the buzzumi way of doing things is that one person - you - is the Host. Only the host needs an account with buzzumi. Guests don't need an account so there's no need for them to sign up or even to have visited the site before.

First lets look at getting you a Host account. This will only take a few moments. First jump to buzzumi.com and click the (nice yellow friendly) signup button.

From here, signup should be straightforward. The first field is your username - pick this with care, because it will be your "profile page" as well - ie buzzumi.com/username - and shows up in your chat URLs as well. Your email address is just for the welcome email and password recovery.

Once you're signed up you're taken straight to your new profile page. You can customise things here - add details about yourself, your websites, blogs, twitter etc. For the moment we'll skip that and go straight to the fun bit - creating your chat.

Lets do this real quick - click "launch chat" on your profile, then select one-to-one for chat type, and lastly hit "create" (You could do lots more using the rest of the tabs but..later). That's it, you're done and now you're in your new chat space. We just have one more task - invite your friend! there are two simple ways to do this.

If you're chatting with your friend via another medium - say twitter or facebook - you can simply copy & paste the link from your URL bar to them. When they follow it, buzzumi will ask for their name and that's it - no signup or anything - and then they'll be straight in.

If your friend is on email we have a simple shortcut - just hit settings and put their email address in the invite box and click invite and buzzumi will send them a friendly message straight away with a link.

Once you're both in the chat space you can chat using text, or click enable camera for the full video/audio experience. Chat away and when you're done just close the window - no fuss, nothing to install!

And in the future, you can chat with anybody just by sending them a link. No Skype dance for you!

I won a Kindle from @tokbox for this idea

OpenTok, a service that allows you to embed realtime video chat into your web app, recently decided to run a competition for great application ideas based on their platform. The prize was an Amazon Kindle.

Back in 2005, a short blog article entitled "Ideas are just a multiplier of execution" was posted by Derek Sivers, and it has recently done the rounds again thanks to a repost on Daring Fireball. I agree with the statement, but I think it's worth saying more.

Ideas and *timing* are multipliers of execution. Ideas are easy to come up with. Good ideas are somewhat rarer, and good ideas at the right time are rarer still.

So write the damn things down.

The idea you think of today may not be right for now. But the future is rather unpredictable and honestly it's difficult to say what might have value at a given moment. I keep a big list of ideas for this reason - ideas are cheap, but having the right one at the right time can make a big difference.

Of course, having all these ideas and being able to execute on the ones that are timely would require that I have a lot of time - which I don't. I can only do a few things at a time and I'm a busy guy. When the competition came up I looked at my list, grabbed the best relevant idea I could find and threw it out there.

The concept is simple - turn any old laptop or system with a webcam into an instant, no-download webapp-based security camera. Using the OpenTok features and a bit of ingenuity it should be possible to record, do motion detection, and then add on SMS and audio alerting, logging etc. In a sense an idea I had 4 years ago is now much much more practical - the time has (probably) come.

It turns out the peeps at TokBox really loved the idea and gave me the prize. There were other great ideas submitted, you can see a list by searching the #opentokideas hashtag.

But...I don't have time to do it. I'm busy with Buzzumi and my other work, adding another project just isn't possible right now.

I think the world would be a better place with a simple, no-effort security camera solution like the above. It would be useful for all kinds of scenarios, not just security. I'd really like someone to do it.

I'm probably not the first to think of it, but I'm taking the opportunity to make clear that I want you to take this idea, and make a fortune from it. I don't want a penny, I don't want credit. Regardless of the fact that I have no legal right to it anyway, I want anyone who is thinking "Man, I could do that.." to be absolutely free of any moral qualms. Take the idea and be awesome and I will be happy.

In addition, since it's a service I'd love to see: If you write up a *good* Kickstarter or whatever and send me the link, I will contribute. If you build the service, I will be your first paying customer.

Andrew Sell from Hipcycle did a great interview on Buzzumi today and afterwards I took a look at his Twitter feed. His most recent tweet was the following:

"Far and away the best prize that life offers is the chance to work hard at work worth doing." -Teddy Roosevelt

Among the many projects out there worth doing, I think this is one. Got some time on your hands? why not give it a shot.

We've been alone too long

How to send me design mockups

The state of Javascript

You're laughing at the labour web guy now..