TechDistortion

Cooking with Tech Articles

As part of a marriage one is expected to get involved with your partners interests (up to a point) and for me that included watching MasterChef this season. The frustration is hard to bear as one watches these people, some of them with potential to become amazing chefs, be knocked out of the competition due to a simple mistake. The assessment is on a series of individual tasks with the decision to keep them or put them out of the competition based solely on the quality of what they’ve produced, against their quality of those their peers produced, at that moment in the allotted time.

Imagine if companies managed their employees in this way. Each employee in a group is given the same task to complete and the one that produces the worst quality result is fired. No consideration is given to past performance, specific talents that may be valuable to the company or overall experience. Companies that are run in this way are rare if not, non-existent as they should be. Such a concept simply doesn’t work.

The parallels with Technology opinion articles may not be immediately obvious but there is a striking similarity with the way in which these TV Shows progress and how many tech-blog-news sites have been producing content lately. Company X releases a new product: it is judged on its own: “A triumph” or “A failure: Company X is doomed” without taking past direction into account, understanding the company’s strengths, how long the company has been in the business they are producing devices for or strategic moves the company may be making.

Judgement is quick.

Pageviews spike (then fade) quickly.

Analysis is scarce.

The fact that TV cooking shows like MasterChef exist and are structured in this way demonstrates that people in the general populace aren’t interested in truly finding the best chef, but rather watching a series of manufactured challenges that rarely find the best but instead provide “more exciting” viewing. The reality is that reality shows have developed this way because the market (people in general that watch TV) responds to it.

Now consider the reality of Tech Blogs. They too have evolved to supply the market with sensationalised pageview-driven drivel. They too aren’t interested in finding the truth or writing the best possible article with proper research and context. They are however a product of the desires of their readers favouring the “more exciting” over the well researched.

That said it seems that this is unlikely to change any time soon and tech opinion-blogs or “news” sites will continue to produce this kind of content.

It’s sad.

It’s depressing.

Alas, it seems to be what WE want.

If Only I Had More Klout

In 2008 a small company called Klout launched with a product of the same name which now allows users to opt-in and link their Twitter, Facebook, LinkedIn, Google+, Instagram, Foursquare accounts and more into a common user profile and then grades people based on their “influence” on each social network and then overall. The concept is to put a number on someone’s ability to influence through social media on a scale of 1 to 100 where higher numbers are better.

To focus on Twitter many people have noted that Klout score has more relevance than follower count. This article is essentially an extension of The Social Numbers-Driven Pissing Contest written nearly two months ago. If you haven’t read that article I highly recommend you do before continuing.

Read it now? Great! Let’s continue. What follows are the problems with Klout as I see them:

1) Interaction Quality is not easily Measureable: Klout can only calculate its figure based on interaction¹ however the type of interaction is non-deterministic. What if I’m being a troll to hundreds of other Twitter users eliciting hostile responses - to Klout this is no different to a pleasant response from the same number of users for good reasons. Of course one might expect the trolling approach long-term would inevitably lead to less responses however my observations are that for some users it can be easier to get a response by trolling than by gentle, good-mannered discourse.

2) Areas of “Influence” Aren’t Specific: If Person X has 100 followers interact with them and they are all going to University together socially, then Person X would have an equivalent score to someone who had 100 followers evenly spread around the world in all areas of expertise, across all demographics and age groups. It’s not hard to see that the latter counts in the real world as being of more value in terms of “Influence” however Klout can not differentiate between the two. Klout uses Topics to show areas of focus but these can not possibly tell much of the overall picture. Famously early on, the US President was rated with a lower score than several prominent bloggers. Clearly “Influence²” isn’t terribly specific.

3) Microcosms of Enthusiasts Drive the Numbers: Social networks allow people to find each other across the world with areas of similar interest. Famous people in the real world offer an opportunity/insight into their lives by participating as well. By and large it seems that groups tend to flock together whereby groups of like-minded people tend to follow each other in small social-webs that interact amongst themselves quite prolifically. A topical tweet amongst that group could have dozens nodding in agreement, favouriting and retweeting the original tweeter and driving up their score. If exposed to the greater Twitter audience the reception would likely be exceptionally less involving. Hence the microcosm and its overly enthusiastic members tend to drive each others numbers higher than if they were graded against a more complete Twitter audience. Essentially then all scores are highly biased.

There’s more but let’s leave it there. The final problem though isn’t that there are  things wrong with Klout as just highlighted but rather it’s what’s wrong with us. What is it that we find such a scoring system to be of any use or attraction at all? Certainly there are psychological needs that people want to feel validated but in the end what does it really amount to that’s tangible? In the real world scores and marks in examinations are a measure of potential competence in a given subject area. Given Klouts broadness and flaws based on its input data this information is simply far too vague and soft to be of any objective use to anybody.

Still the champions of scoring themselves as some measure of success will pound the table insisting that their Klout score is an excellent measure of social media success. Too many others will see this behaviour and wish they had more Klout and therein lies the problem with the system - for the vast majority of users the score has no meaning unless it is used to compare oneself with others. The behaviours that this drives are similar to those where everyone knows everyone else’s salary in a company. Generally it drives hostility and jealously in the under-remunerated and worse it feeds superiority and arrogance in those over-remunerated.

I’m happy with my score - I enjoy Twitter and my blog even if it doesn’t do stupid volumes of traffic but like anyone I wouldn’t say no to an increase. If I don’t get one it’s no great loss but if you’re the sort of person that only cares about out-scoring everyone else then eventually you will learn like many others before you - it’s an end unto itself and there is nothing waiting for you at the end of that road.

Saving Power Without Giving It Up

Part of engineering design and product development is compromise. The difficulty is choosing which compromises to make and why. To understand which are the right ones to make we consider end user use cases. How will they use the product? When and why will they use it? The answers to these questions should drive the design of the product in question. If they don’t, chances are that most people won’t be satisfied by it and the product will ultimately fail. As designs evolve over time, good companies keep re-asking those very questions as technology changes, peoples needs change and thus so should products change with them.

Much has been discussed about Moore’s Law and David Houses observation that computer power doubles every 18 months and how silicon is reaching its limit beyond which the process simply can not continue. There are competing technologies vying for a place in the silicon dominated computing world but most are many years away from being mass-producable or even financially viable. Anticipating this approaching road-block, operating system and software developers are moving towards more parallel processing with multi-core CPUs now standard on desktop, laptop, tablet and smartphone platforms around the world. In order to keep reducing the size (and hence increasing the density) of CPUs, smaller and smaller processes must be used (45nm, 32nm and Intel Tick-Tock), but in order to use them their average clock speeds must be reduced. At this stage it has become more about increasing/compressing/multi-processing tasks rather than striving for additional computational speed.

The paradigm shift recently is that it is no longer about CPU “power” previously equated with speed, but rather CPU “power” as it relates to CPU power consumption. The smaller we shrink, the less power we use and with architectural improvements we maintain the same level of computational performance overall. Within the boundaries of the laws of physics this can only be a good thing as these devices can now power an advanced smartphone and still have a battery life of up to a day¹.

As the mobile computing revolution has pushed for lower and lower powered mobile CPUs, the desktop/laptop CPUs have also been advancing in a similar direction and we are now rapidly approaching an inflection point in the computing industry when powerful CPUs are no longer the driving force but rather, less power consuming CPUs.

There was a time when each new version of popular software (like Microsoft Office and web browsers) needed a little more computing power than the previous version. Operating systems as well started out with outline dragging windows, then progressed to full frame rate window and content dragging, and then to shrink/spin/genie 3D window effects that each required still more hardware power with each operating system upgrade. Recently the move has been to offload the graphical components to the GPU leaving the CPU for running software and computational tasks and there are now many good low-power GPUs available that provide satisfactory performance as well.

Perhaps it is impatience but looking around the PC market place for the last 5 years, operating system upgrades have not required hardware upgrades for performance reasons when using the major operating systems²

This situation has many people asking themselves, if it’s not about CPU “power” in the traditional sense, what actually matters in a computer now? Interface usability? Screen size? Power consumption? Perhaps it’s all of these or some others but let’s focus on power consumption. Around the world there is an increasing energy shortage. Driven by market forces and companies wishing to maximise profits the keepers of the energy are the holders of the keys in our modern world and that energy is no longer cheap. In some cases the cheaper resources have long since been used and the resources in use now are more costly to extract. There is also a mounting series of evidence giving credence to the theory regarding global warming and even if that energy is created from Nuclear Fission, polluting the world with concentrated radioactive waste seems to also be a questionable approach.

Thus energy usage is increasingly on people minds for one of two reasons³: 1) Energy costs money and the current world economy has pushed peoples budgets hard so every cent counts; 2) Saving energy reduces the amount of pollution released into the environment as we simultaneously switch to renewable energy sources.

The MBA this article is being written on has a 2.0GHz Core i7 CPU which consumes 17W at Idle⁴ and 48W at full load⁵ which we compare with my Mac Pro⁶ at 130W at Idle and 240W at full load. At 23c/kWh that difference is staggering when you consider that apart from the occasional video encoding, the MBA does just fine - even compiling in Xcode it keeps enough pace with the Mac Pro that it’s good enough for my needs.

With Apple introducing Timer Coalescing in 10.9 of OSX “Mavericks” and the rollout of Intels Haswell CPUs across their machines⁷ it’s clear that Apple sees the future in low power computing. They in particular seem to be reassessing customers needs and focussing on lower powered devices across their product lines. With all of this in mind, before you purchase your next computer ask yourself a serious question: How much “power” do you really need?

In Memory of Mick Devlin

My last interaction with Mick was 17 days ago on Twitter, with this delightful little interchange: @johnchidgey “When the new Mac Pro is released, how long before a pile of scrap paper builds on top as it is mistaken for a desktop rubbish bin?” @persistentgecko “A mallet will be placed beside the machine to allow the owner to use on people who treat it like a bin.” Light-hearted and always with a humerous twist, Mick couldn’t help himself and I loved it.

Like many other people I have met on Twitter from time to time a few days pass without a word and I realised that Mick hadn’t commented on anything I’d said. As always I shrugged it off because sometimes people have internet problems, take a break from the internet (sometimes just social networks) or get very busy at work; but then a Tweet from my good Twitter friend Clinton Philips (@clinton1550) pointed me towards this, and I then knew that Mick was dead.

Mick and I first crossed paths online at the Australian MacWorld forums when I was trading under the pseudonym “AppleConvert” but when I left the forums and moved to Twitter we followed each other and kept in touch. Apart from our love of all things Apple, Mick lived in Townsville - a city that I had spent a bit of time working in and I had an affinity for, and we shared a love for the Strand and its views of Magnetic Island. He would post The Strand view of Magnetic Island regularly on Twitter sometimes I thought to make me jealous as I was based in Brisbane some 1,350+ kilometres (850+ miles) away.

He was an avid listener of the Exastential podcast and was always interacting with myself and Clinton after every show. When we put the podcast on hiatus after twenty episodes (which it is still on) Mick would beg us to start podcasting again. We didn’t have a big audience but we had a huge fan in Mick Devlin.

When I put TechDistortion on hiatus nearly four months ago, Mick commented that he enjoyed reading it and wanted me to keep writing some day. I did return and Mick once again was a regular reader of my work, piping up with feedback and on some occasions helped me to prune my work when I’d made mistakes.

I’d never met him in person but that hasn’t made his loss any less real to me. I value my friends on Twitter that interact with me and who enjoy what I do. Mick was one of my best Twitter friends and I will sorely miss our conversations. Rest in peace my friend.