The famous large flat thing slightly more useful than a rock

Here at Konstellation, we just launched the PageGangster for mobile devices including iPad, iPhone, iPod touch and Android. I developed the clientside for this in JavaScript and we launched it with automatic user agent detection serverside. But this had some problems.

I read some info about this and conclude, that some iPad user agent strings include the word “iPhone” and some users of all types of devices fake their user agent strings.

But then I read on the above link, that no matter what, the navigator.platform attribute would always be set to the right type of device. So in the regular web version of the publication viewer we have now (which uses Flash), we have added a JavaScript detecting if the serverside script did not redirect the client properly:

if (navigator && navigator.platform && navigator.platform.match(/^(iPad|iPod|iPhone)$/)) {
	location.replace(location.href+"?force-mobile");
}

It seems simple, but solves the problem of some users seeing the dreaded blue box instead of their publication optimized for their device.

On another node, this experience of developing a web application for the iPad/iPhone was very rewarding and I gained many insights, which might spawn other posts. And if you have any questions about it in any way, feel free to ask here – but please do not submit bug reports about the PageGangster for mobile – do that via info@pagegangster.com instead.

No related posts.

I will now update the previous entries to reflect this change and restructure the current content.

The next Month of Ideas will be either April or May 2010 – and I will start now collecting ideas for this just keeping them hidden until I am confident, that I can complete the task for the given month. But at least I tried and have actually produced a wealth of cool things in my opinion – I hoped you have enjoyed it as well.

No related posts.

For that purpose, Wikipedia categories seem very useful and given the openness of the Wikipedia philosophy, it would actually be quite simple. The site could be as simple as a single input field with input suggestions based on the Wikipedia category system. For example you could input “1950s in film” and maybe even intersect it with another category like “Best Picture Academy Award winners” and the resulting list (including all sub-category entries as well of course) would then be looked up in recent tweets on Twitter and the results would be displayed in different ways – including a list of most popular items, time-based breakdowns, geographic breakdowns, related links, most dominant profiles etc.

The technology behind it is quite simple and Wikipedia category search and even intersection has been seen in many different applications before. But using it for creating lists of items in a certain category to lookup in a different medium all-together is a new in my book.

Oh, and yes, the name used in the headline is quite stupid, agreed, but with Twitter and Wikipedia involved, Twiki-something seems appropriate.

Why?

Trending topics on Twitter are highly popular and the target of many services, lots of discussion and even some succesful spamming attempts. But the general trends are way to general to be interesting for analyzing what’s happening in a certain industry or area of interest. This application/service/mashup is a low-hanging fruit, and might carry some hype once deployed, but might not have a long-standing potential. I do however personally find it very interesting and believe that a lot of “social media experts”, trend researchers and marketeers will as well.

What’s next?

Do with this idea whatever you like – expand, implement, trash or forget. Just remember, that if you use it in anyway make sure to attribute me according to the Creative Commons Attribution 3.0 License, that all these 365 Social Ideas are published under.

No related posts.

Introduction

In a regular (i.e. useful) number system, the digit values are powers of a given number starting with the right-most digit representing a digit value of the given base number to the power of 0, the second digit from the right representing a digit value of the given base number to the power of 1 etc. This has a lot of advantages beyond the topic of this post. Thus, in a base 10 number system, the right-most digit value is 1, the second from the right is 10, then 100 etc. There is a simple rule for how we go from one digit value to the next and for how to find an arbitrary digit value just by knowing the digit position.

But the simple point I’m making here is, that we could invent a number system, where the digits represent arbitrary values – e.g. the first digit from the right represent 1s, the second digit represent 4s, the third represent 178s, the fourth represent 179s etc. We can express any number in this (we could just extend the available numbers with letters or other symbols as we see fit) – in the example given here, the decimal number 2000 would be represented as 11*179 + 0*178 + 7*4 + 3*1 = B073. It is quite a useless set of digit values, and it is a very hard representational number system to do any math in.

For the XKCD book page numbering, Randall has chosen a number system, where the first digit from the right represent 1s, the second digit represent 3s, the third represent 7s, the fourth represent 15s and so on – thus there is actually a rule for finding the digit value for the n‘th digit (from the right) = 2i-1 – compare this to the binary system, in which the digit values can be calculated as 2(i-1).

UPDATE 2009-11-26: As it has been pointed out to me, this number system (which is skewed one off binary) is called skew binary. The only interesting property that I can find anyone reference is the fact, that increment only ever has one carry-over, which makes increment very easy to implement (and I could thus implement it in regexp as given further below). It is however only a “scientific” numbering system, and I can find no reference to any (other) usage in the wild. The article linked before is the first Google hit for “skew binary” and it is an 11 year old slide – which indicates further, that it is a rarely used system with only theoretical purposes.

Conversions

So, how to find the value, that a given XKCD page number represent? Simply multiply each digit by the digit value – if the i‘th digit (where i=1 represent the right-most digit) has the value xi the translation of an n digit XKCD page number can be done via this formula:

General conversion rule from XKCD page numbers to decimal numbers

The reverse translation can’t be done like regular conversion between regular bases bulding the number from the right (using modulus and division), but has to be done from the left. Thus we need to find the maximum i for which 2i-1 is less than the given number – which we do via log2(number+1). Then we go from this maximum i (which denotes the number of digits n in the XKCD page number) to 1 and for each step find the i‘th digit by evenly dividing the target number with 2i-1 and storing the remainder as the new target number – here both implemented in JavaScript:

function dec2xkcd(d) {
  var i = Math.floor(Math.log(d+1)/Math.LN2);
  var r, x = "";
  for (; i; i--) {
     r = Math.pow(2,i)-1;
     x += Math.floor(d/r);
     d = d%r;
  }
  return x;
}
function xkcd2dec(x) {
  var d = 0, l = x.length, i = l;
  while(i--) {
    d += parseInt(x.charAt(i))*(Math.pow(2,l-i)-1);
  }
  return d;
}

We can verify these by running some tests converting back and forth:

function c(d) {
  var x = dec2xkcd(d);
  var d_ = xkcd2dec(x);
  return d+" => "+x+" => "+d_;
}
c(1); // 1 => 1 => 1
c(6); // 6 => 20 => 6
c(15); // 15 => 1000 => 15
c(16); // 16 => 1001 => 16
c(119); // 119 => 111110 => 119
c(120); // 120 => 111111 => 120
c(6542); // 6542 => 110011001010 => 6542
c(5469835); // 5469835 => 1010011011101101001020 => 5469835

Properties

The selection of digit values are not completely random, they do follow some system, but not the systems of regular power-based number systems. So the properties, that regular power-based number systems have are not present in this, but it does have other properties.

First of all, the notion 2^n-1 is known from Mersenne primes. What this tells us however, is nothing useful about the system except that some digit values are prime – like 3, 7, 31 and maybe, maybe not an infinite number of the infinite other digit values.

Another property is, that from any digit value dn, the next digit value can always be calculated as 2dn + 1. This means, that the largest number the can be represented by n digits is a left-most 2 followed by n-1 0s – the next number in sequence is a left-most 1 followed by n 0s. But again, this is quite trivial and not really that useful.

The repunit numbers (numbers consisting solely of 1s) in this representational system represent the numbers 1 (1), 4 (11), 11 (111), 26 (1111), 57 (11111) and this order is actually the second column of the Eulerian numbers (defined by sequence A000295 in OEIS) – but this is still quite useless.

Algebra

So, is this usable as a number system? Can we do math in it? Can we in a simple way multiply numbers? Let’s start by making an algorithm for adding 1 to (incrementing) an arbitrary number. Observe, that we cannot simply safely increment the last digit if it is a zero – because if the number has a 2 in it (which only occurs in at most one position in the number and with no trailing 1′s), increment is done by converting the 2 to a 1 and incrementing the digit to the left of the 2 – examples:

• Increment 2 => 10
• Increment 12 => 20
• Increment 102 => 110
• Increment 120 => 200
• Increment 1012 => 1020
• Increment 2000 => 10000

It is not trivial, but then again, not that hard. For numbers without a 2, simply increment the right-most digit (which has to be a 0 incremented to a 1 or a 1 incremented to a 2).

Then, how do we subtract 1 (decrement)? Here the rule is, if the number has trailing 0s (one or more right-most digits are 0), decrement the right-most non-zero digit and change the left-most of the trailing 0s (the number just to the right of the right-most non-zero) to a 2 and keep the remaining as 0. If the number does not end in a 0, simply decrement the last digit – examples:

• Decrement 2 => 1
• Decrement 10 => 2
• Decrement 12 => 11
• Decrement 112 => 111
• Decrement 1010 => 1002
• Decrement 1020 => 1012
• Decrement 10000 => 2000

Still not trivial, but not impossible. UPDATE 2009-11-26: The fact that you at most have to alter two connected digits when decrementing or incrementing is what makes this number system interesting as this is a lot easier to implement that normal “base-based” number systems, where you worst-case have to alter all digits of the number when in- or decrementing – compare incrementing the decimal number 999999.

Comparison of numbers is done exactly as for numbers in any (regular) base – if one has more digits that the other, the “longest” number is higher. If not, compare digit by digit starting from the left, if a digit is higher in one number, than the same in the other, that number is larger, if equal, go to next digit. If all digits are the same, the numbers are equal.

All right, so far so good. Then we can add and subtract (but as we won’t consider negative numbers, subtraction cannot be negative). We add x to y by incrementing y and decrementing x at the same time until x equals 0 – the value of y is then the result. For subtracting x from y (i.e. y - x), we decrement both numbers until either is zero – the value of y is then the result (thus 1 – 2 is 0, as we don’t do negative number).

The next logical step is to consider multiplication (of positive numbers only of course). We multiply x by y by creating a result variable r and add x to it y times (decrement y each time until 0).

Implementation

Note: this gets very pointless, so feel free to skip to the conclusion – which is that all of this is just silly and pointless.

First, increment and decrement in JavaScript:

function inc(x) {
  if (!/2/.test(x)) {
    return x
      .replace(/1$/, "2")
      .replace(/0$/, "1");
  }
  return x
    .replace(/(0|^)2/, "10")
    .replace(/12/ , "20");
}
function dcr(x) {
  if (!/0$/.test(x)) {
    return x
      .replace(/1$/, "0")
      .replace(/2$/, "1");
  }
  return x
    .replace(/20(0*)$/, "12$1")
    .replace(/10(0*)$/, "02$1")
    .replace(/^0+/, "");
}
function id_test(d) {
  var x = dec2xkcd(d);
  var x2 = inc(x);
  var x0 = dcr(x)
  return x0+" ("+xkcd2dec(x0)+") => "+x+" ("+d+") => "+x2+" ("+xkcd2dec(x2)+")";
}
id_test(1);   // 0 (0) => 1 (1) => 2 (2)
id_test(2);   // 1 (1) => 2 (2) => 10 (3)
id_test(6);   // 12 (5) => 20 (6) => 100 (7)
id_test(7);   // 20 (6) => 100 (7) => 101 (8)
id_test(117); // 111100 (116) => 111101 (117) => 111102 (118)
id_test(123); // 111120 (122) => 111200 (123) => 112000 (124)
id_test(127); // 200000 (126) => 1000000 (127) => 1000001 (128)
id_test(128); // 1000000 (127) => 1000001 (128) => 1000002 (129)
id_test(603); // 100101120 (602) => 100101200 (603) => 100102000 (604)

Then comparison:

function cmp(x1,x2) {
  var l = x1.length, l2 = x2.length;
  if (l != l2) {
    return l > l2 ? 1 : -1;
  }
  var i, c1, c2;
  for (i=0; i<l; i++) {
    c1 = x1.charAt(i);
    c2 = x2.charAt(i);
    if (c1 > c2) {
       return 1;
    }
    if (c2 > c1) {
       return -1;
    } 
  }
  return 0;
}
function cmp_test(d1, d2) {
  var x1 = dec2xkcd(d1);
  var x2 = dec2xkcd(d2);
  var c = cmp(x1, x2);
  if (c > 0) {
    return x1+" ("+d1+") is greater than "+x2+" ("+d2+")";
  }
  if (c < 0) {
    return x2+" ("+d2+") is greater than "+x1+" ("+d1+")";
  }
  return x1+" ("+d1+") equals "+x2+" ("+d2+")";
}
cmp_test(1,1);     // 1 (1) equals 1 (1)
cmp_test(1,2);     // 2 (2) is greater than 1 (1)
cmp_test(2,1);     // 2 (2) is greater than 1 (1)
cmp_test(117,117); // 111101 (117) equals 111101 (117)
cmp_test(117,121); // 111112 (121) is greater than 111101 (117)
cmp_test(127,128); // 1000001 (128) is greater than 1000000 (127)
cmp_test(127,126); // 1000000 (127) is greater than 200000 (126)

Then addition:

function add(x1, x2) {
  while (cmp(x2, "0")) {
    x1 = inc(x1);
    x2 = dcr(x2);
  }
  return x1;
}
function add_test(d1, d2) {
  var x1 = dec2xkcd(d1);
  var x2 = dec2xkcd(d2);
  var xs = add(x1, x2);
  var ds = xkcd2dec(xs);
  return x1+" ("+d1+") + "+x2+" ("+d2+") = "+xs+" ("+ds+")";
}
add_test(1, 1);   // 1 (1) + 1 (1) = 2 (2)
add_test(1, 2);   // 1 (1) + 2 (2) = 10 (3)
add_test(2, 1);   // 2 (2) + 1 (1) = 10 (3)
add_test(7, 7);   // 100 (7) + 100 (7) = 200 (14)
add_test(8, 17);  // 101 (8) + 1002 (17) = 1110 (25)
add_test(53, 17); // 11100 (53) + 1002 (17) = 100100 (70)

Then subtraction:

function sub(x1, x2) {
  while (cmp(x1, "0") && cmp(x2, "0")) {
    x1 = dcr(x1);
    x2 = dcr(x2);
  }
  return x1;
}
function sub_test(d1, d2) {
  var x1 = dec2xkcd(d1);
  var x2 = dec2xkcd(d2);
  var xs = sub(x1, x2);
  var ds = xkcd2dec(xs);
  return x1+" ("+d1+") - "+x2+" ("+d2+") = "+xs+" ("+ds+")";
}
sub_test(1, 1);   // 1 (1) - 1 (1) = 0 (0)
sub_test(1, 2);   // 1 (1) - 2 (2) = 0 (0)
sub_test(2, 1);   // 2 (2) - 1 (1) = 1 (1)
sub_test(7, 7);   // 100 (7) - 100 (7) = 0 (0)
sub_test(17, 8);  // 1002 (17) - 101 (8) = 102 (9)
sub_test(53, 17); // 11100 (53) - 1002 (17) = 10012 (36)

Finally multiplication:

function mul(x1, x2) {
  if (!cmp(x2, "0")) return x2;
  // x2 is not zero, go
  var r = x1;
  while (cmp(x2, "1")) {
    r = add(r, x1);
    x2 = dcr(x2);
  }
  return r;
}
function mul_test(d1, d2) {
  var x1 = dec2xkcd(d1);
  var x2 = dec2xkcd(d2);
  var xs = mul(x1, x2);
  var ds = xkcd2dec(xs);
  return x1+" ("+d1+") * "+x2+" ("+d2+") = "+xs+" ("+ds+")";
}
mul_test(1, 1);   // 1 (1) * 1 (1) = 1 (1)
mul_test(1, 2);   // 1 (1) * 2 (2) = 2 (2)
mul_test(2, 1);   // 2 (2) * 1 (1) = 2 (2)
mul_test(7, 7);   // 100 (7) * 100 (7) = 11010 (49)
mul_test(17, 8);  // 1002 (17) * 101 (8) = 1000102 (136)
mul_test(53, 17); // 11100 (53) * 1002 (17) = 111000101 (901)

Conclusion

As expected, it is just a silly and very useless number system – but fun to play around with though. If you however find other interesting properties of this representational system, please feel free to comment.

However, it should be clear to anyone, that this has nothing to do with a ternary number system just because the only digits are 0, 1 and 2 – it is closer related to the binary number system.

UPDATE 2009-11-26: Albeit still useless, the number system actually has the name “skew binary” as mentioned in the introduction – and thus has been the target of some scientific research previously.

No related posts.

It is a “simple” case of “click-jacking” and the site tricks you to click a Facebook share button, but disguises this as some other button. Please read on for full description.

UPDATE 2009-12-2: “Press the button or dog dies”/”Push the button or this dog dies” (located at pressthebuttonordogdies.com, but don’t go there) is a new such site. The target website is “thisblogrules.com” and the measures used are a little different but all in all the same anyway.

Furthermore, I have used bit.ly for tracking how much these links have been used so far on Facebook – it is pretty inflicting: The “hot” girl has been shared almost 59,000 times and the poor dog has been shared 5,309 times as of this writing. You can see the direct stats from the Facebook link.getStats API here: Somthing Hot and Or Dog Dies

The original post looked like this (taken from my own feed):

Don't go to this page if you see this post

The page it linked to was http://3dvv.com/somethinghot/ (don’t go there, nothing to see) and looks like:

The page tricking you to share the link

When you press the “button” you actually press this (opacity on iframe set to 1 instead of 0):

The real button that you actually press kind of looks like a Facebook button, right?

The trick here is, that they have used the layout of this page about sharing the link (safe to go to, just don’t actually press the share button): http://www.facebook.com/share.php?u=http://3dvv.com/somethinghot/. Then, they included this page inside an iframe that was offset the right amount negatively to the left and to the top, to leave just the share button visible in the corner here (no need to visit): http://3dvv.com/somethinghot/ngr.php, which looks simply like:

How to only show the share button by iframing the share.php page

If we on this page remove the constraints, we can see the whole bottom corner of the page like this:

The rest of the share cutout with the button in the corner

Well, this is not all. In order not to have the page simply redirect to what facebook normally would after having shared a link, this page with only the share button visible is wrapped in another iframe, that steals the redirect Facebook creates and make their own redirect to the actual target (porn) site.

All in all, it is a very clever albeit strictly illegal way of using the Facebook sharing functionality for viral spamming purposes.

What can Facebook do about it? One thing would be to not allow share.php to be embedded in iframes via a simple javascript, but that might break desired functionality in other places. Otherwise, they could shift the placement of the sharebutton slightly of different page loads, but that could probably be circumvented by detecting the position of the button via script and isn’t a viable solution. Or they could require that the user had to type some captcha before being able to post links to new sites not posted before, but that creates a lot of other problems (and a good scammer could simply require the user to answer the captcha as well as click the button – users would do that). I’m sure Facebook can think of something, though.

What can you as a user do about it? Nothing really. Only click on “links” on foreign pages that you know for sure are legitimate and that you trust that your friends really did post on their own. But it is pretty hard to protect against. It really is Facebook’s job to make sure this doesn’t happen.

Please share this with everyone!

Update 2009-11-24: Several sites report this as an example of CSRF/XSRF (Cross-Site Request Forgery) and most use this AVG blog post as source. Not that it really matters, but it is not XSRF – it is clickjacking. There is a slight difference, the main difference being, that the countermeasures are very different.

No related posts.

But that craving for getting hands on building something for the iPhone is gone – completely. I’m sorry Apple, you might not miss me, but as things stand right now, I won’t miss you either. And I am certainly not alone – there is a large front in the developer community building up resentment against the policy of the App Store review process. A veritable boycott is brewing.

First of all here are some horror stories about how incredibly tiresome, ridiculous and illogical the review process is:

• The iPhone Development Story by Mike Ash – about the entire app review process and how time-consuming it is.
• Airfoil Speakers Touch 1.0.1 Finally Ships from Rogue Amoeba – about how a simple bug fix ended up making the developer choose between a less functional or a less user friendly version of his app.
• Apple’s FCC Response Infuriates Google Voice App Developer from MacLife.com – about how Apple randomly picks out already approved applications for removal, if they see fit (and of course give no real reason, only air in response).
• How Apple’s App Store review process hurt Occipital’s RedLaser SDK and my very own ‘avit iPhone app by Aral Balkan – about how Apple’s policies are hurting everyone and not making any business sense whatsoever…

Second, how developers are fleeing the platform:

• On Middle Men by Joe Hewitt – about how his experiences with Apple and the App Store during his development of the Facebook App for iPhone has made him quit developing for iPhone.
• The iPhone Development Story: One Year Later by Mike Ash – about how Mike’s experience and Apple’s lack of response one year after his initial bad experience has caused him to skip the platform.
• Back To The Mac from CarpeAqua.com – about how the overall experience has caused yet another value-adding developer to only develop for Mac, not iPhone.

Third, how commentators are warning Apple about the no-go of trying to be a “software publisher” as that is bound to go bad:

• The War For the Web by Tim O’Reilly – about how trying to control your medium will lead to doom or big business (but most likely doom in the long run).
• Apple’s Mistake by Paul Graham – about Apple trying to be the middle man where no such was needed or even fits.

All in all, my conclusion is clear: Symbian/Droid/Palm/WinMo/TouchWiz/whatever open mobile operating system you can name, here I come!

UPDATE: PPK just posted his solution to the dilemma: create web apps, not native apps (but in a somewhat cryptic wording though) – and since that post he then posted an equally strangely worded “rebuttal”.

No related posts.

Well, that’s a lot more attention, that anyone else has ever given my blog – now what does that mean? Feel free to post non-spam-y comments ;)

No related posts.

After reading up on UK copyright law, my judgment is unclear – but mostly in favor of the NPG and thus against Wikimedia.

The CDPA of 1988 clearly states in the very first paragraph:

1. Copyright and copyright works
   1. Copyright is a property right which subsists in accordance with this Part in the following descriptions of work—
      1. original literary, dramatic, musical or artistic works,
      2. [...]

To me, one crucial part is the word “original”. This removes the possibility of claiming, that merely taking a photograph of another work is copyrightable in itself. The law goes on to specify what each of the four here-described works are, and it is clear, that taking a photo is an artistic work (defined in §4 sec. 1.a), but it is not an original work (in my opinion), when it is a photograph of a painting.

But the manipulation done by the conservationists when digitizing the photos includes a lot of work, which is definitely craftsmanship, but if deemed as artistic craftsmanship, it is in itself covered by the CDPA in §4 section 1.c:

4. Artistic works
   1. In this Part “artistic work” means—
      2. [...]
      3. a work of artistic craftsmanship.

My conclusion is thus, that it is irrelevant whether it is a derivative work or not (non-original works are never copyrightable) – the matter of the case is, whether the work of conservationists is artistic craftsmanship or not. And I’m leaning towards a “yes”, but I’m pretty sure the case cannot be settled without a court ruling, as it is clearly a case of interpretation. You can argue that a conservationist is merely a scientist, not an artist – but even scientific papers and articles are subject to copyright, and thus I believe, that the work of a conservationist is subject to copyright given that it is artistic craftsmanship and/or original scientific work.

No related posts.

After the install was complete however, all I saw was this:

Silverlight Smooth Streaming - not a very smooth start though

Probably not the result I was expected to see – it was quite a terrible first impression to give me. I might just need to reload, I might just need to restart my browser, but if any of that was needed, the installer should have done just that – and not give me this ridiculous result.

Better luck next time, Team Silverlight.

No related posts.

For this use, I was missing a good list of ccTLD‘s, that are available for direct second-level registration without local presence. Some countries don’t allow direct second-level registration like .uk (United Kingdom) or .tr (Turkey). All of the below-mentioned domain names are available for direct second-level registration and do not require local administrative contacts, trade licenses or similar restrictions. Most of them can be purchased online, but some require paper forms sent back and forth and might take some time to register.

Table not yet complete…

Notes:

• ^* 2 years minimum registration – listed priced here is half that
• ^** Price is based on the cheapest available hosting plan – not domain only

No related posts.