GMDS for surface pairs comparison

S natural succession to the previous experiments, I compromise speed and increase the number of points in GMDS from 10 to 50, hoping to see performance improved noticeably. So far, based on results that are coming out, no classification mistakes are being made, but more pairs need to be tested overnight before meaningful conclusions can be drawn.

As expected, increasing the number of points in GMDS helped improve performance, at the expense of speed of course.

They will now be increased even further to see how good an improvement can be made.

I have begun running GMDS-based recognition experiments with 300 points rather than the originally/already-tested 50. So far, based on 18 pairs, the classification is perfect (albeit the process is slow). It will be interesting to see how long it runs before the first error occurs. There was a talk back in December about a new server potentially arriving, with a lot more x86 cores in it.

Following the first mistake of the classifier, the ROC curve looks like this.

The next step doubles the density of the triangulation to see how major an improvement in performance it can entail (we never tried as many as 8,000+ vertices in such experiments).

With ~8,050 vertices the results are a bit baffling. Both computational servers got rebooted midway (memory upgrade), so there is not much data — to say the least — to base these results on.

By summing up all the GMDS stress (multiple iterations) we get iffy results.

By taking the optimal fit (over all iterations) we get something better.

In summary, despite the small size of the sample, it does seem to suggest that using a triangulation too dense leads to poor results or does not lead to optimal performance. I’ll explore other route towards improvement; adding density just slows things down considerably.

INYURL used to be one among few services that shortened long URLs for people (snipurl was another). These services would typically be used only when a URL was exceptionally (and perhaps irregularly) long — to the point of being incompatible with some mail/news readers. The landscape has changed somewhat and nowadays, with people becoming lazy and phone-tied, folks assume that original URLs don’t matter, even though they indicate their source, their date, and perhaps title as well, at least most of the time (if properly designed at CMS level). URLs are not merely barcodes, their visibility contributes a lot toward expectation and even the ability to recover broken URLs in the future (based on their source or title). Now that every John and Jane creates his or her own shortener, there is less incentive to keep those services up in the long run, especially once they drive a lot of ‘legacy’ traffic and do nothing for revenue (mere redirections). As the creator of the World Wide Web put it about a year ago, there is also the danger that authorities of TLDs will sooner or later be misusing their power to censor some popular shorteners and immediately break a whole lot of linkage, upon which much of the Web is based. In a sense, short URLs break the Web, but when there are limitations on the number of characters in services like Twitter, people are pressured to use them, even by default.

Best practice is, avoid URL shorteners unless there is absolutely no way around it. And if you must use a shortener, choose one which you believe is most likely to survive the next 5, 10, or maybe 50 years (very unlikely). Otherwise, the URLs will be as useful as a random string, i.e. disposable garbage. As wary as I am of Google, I recently started using Google’s shortener when no other choice exists (identi.ca/Twitter have length limitations). Being a tracker, it’s one service that Google is unlikely to axe any time soon.

HE Mr. Fitness 2012 competition kicks off tomorrow and I’ve signed up to participate in it. I am in a good position (or rather in good shape) to win it this year, having run a lot recently and also dropped some weight. I shall write about the competition regularly because it spans a period of 2.5 months. In my early twenties I won quite easily, but now it’s a lot harder to keep up.

OOGLE decided to hop on that whole “social” and “Web 2.0″ bandwagon by further destroying channel views in YouTube, replacing with arguably “simple” layout what people were already accustomed to. There is a parallel migration permitting people to stick to the old layout, but that won’t last for long. It’s merely a grace period, intended to calm down resentful users of YouTube.

For those who have not seen the new layout, here is my channel as an example. In the past there were a lot more options, both for customisation and for function. Google decided that because “less is more” or whatever we should just accept the removal of perfectly useful (and usable) features.

On the positive side, Google is phasing in free codecs and as more browsers (and more users) adopt support for those codecs we are going to live in a Web without Adobe Trash and the MPEG cartel.

F YOU are not worried about what you search for and how you surf the Web, perhaps you should. By letting us know that we are watched as we search and surf, we are led into a state of self-censorship. Ideas can be suppressed this way, so for the more progressive among us everything becomes riskier.

I have not much against Google. They tried to hire me several times and I support Android wholeheartedly. After Microsoft had hijacked Yahoo, the only viable crawlers-based search engine (no meta-searches and such) which was not operated by criminals seemed to be Google. On a relative basis, Google is not evil, but it is easy to pick on.

If you wish to use Google but not leave a trail for the oppressive US regime (the one set in place by Bush et al. after 9/11), then use Scoogle. It gives more search results per page and it offers a lot more privacy. I’ve set Scoogle as my default in Firefox address bar (instructions here) and in Konqueror, using the search URI “http://www.scroogle.org/cgi-bin/nbbw.cgi?Gw=\{@}” (where \{@} is a placeholder for the search query).

Enjoy your invisible (to oppressive authorities) searching.

The answer may surprise you

ordPress is a leading (probably “the” leading) blogging platform and it is also used extensively as a CMS, even in my business. I run about 10 WordPress sites and I take pleasure in the fact that it is GPL-licensed, community-supported, and managed by very cool and Free software-friendly people.

But who created WordPress?

Crediting just Automattic or Matt for WordPress would be unfair. GPL encourages the laying of one’s work upon another and WordPress actually comes from this piece of software, known to many as b2. Several years ago I personally met the person who liaised with Matt to create WordPress, which helped the succession of the mostly abandoned/forzen b2 (at that time). Matt is currently the only one who receives credit for WordPress, just like Jimmy Wales gets too much credit for Wikipedia (he was merely a co-founder).

In short, WordPress is the fruit of labour of many developers, not just extensions producers who brought appeal to the platform/framework. But originally, WordPress was merely a succession of b2. The nature of Free software is, credit is preserved to a degree, so we can always go back and check who did what. One might say that Michel created WordPress, not Matt.

LOT of people may not know this, but most of my blogging I actually do from a proprietary operating system, Palm OS. I find little reason to write from home, so taking the text out with jpilot and my Palm Tungsten is what inspires me to write long posts such as this one. Much of Techrights is also being composed in Palm OS.

Why Palm Tungsten?

Well, it’s simple really. In one word: keyboard. As devices get smaller and smaller they often neglect to accommodate for productivity, so they resort to gimmicks like touch (which Palm had over a decade ago) and not a good, affordable foldable keyboard.

Is someone else blogging from a PDA?