Bij deze woorden gaan de haren op mijn armen al een klein beetje rechtstaan. Vooral als het artikel eindigt met het venijnige zinnetje: “de foutenmarge bedraagt…”

Wat is immers het probleem? De peiling van La Libre Belgique is een flutpeiling. Voila. Ten eerste worden er appelen met peren vergeleken (federale verkiezingen t.o.v. regionale)¹, en zijn er de laatste jaren maar weinig partijen die in een en dezelfde gedaante aan twee opeenvolgende verkiezingen hebben deelgenomen. (Door het vormen en terug uiteenvallen van kartels.)
Ten tweede vallen de “verschuivingen” in 99% van de gevallen² binnen de foutenmarge.

De zoveelmaandelijkse peiling van La Libre Belgique wordt steeds uitgebreid herkauwd in de Standaard, en wordt ook door de VRT veelvuldig geciteerd. Helaas houdt dit citeren dikwijls in dat de foutenmarge niet mee vermeld wordt.

Bekijk bijvoorbeeld dit Teletekst-bericht van vandaag:

Ziet u daar nog iets staan over de foutenmarge?

Waarom is die foutenmarge zo belangrijk? Omdat ze al die veelbesproken “verschuivingen” er voor hetzelfde geld geen zijn. Ze vallen immers steevast binnen de margin of error.

Kijk even mee naar de peiling van december:

De blauwe balk geeft het resultaat weer van juni 2009 (de jongste regionale verkiezingen), de groene balk het hypothetische resultaat mochten er in december verkiezingen gehouden zijn. De rode balkjes tonen de foutenmarge, of de onbetrouwbaarheidsfactor zo u wil. Met andere woorden, trek gerust de horizontale lijntjes door naar links. Als het blauwe balkje tussen de twee lijntjes valt is het statistisch gezien waanzin om hier een conclusie aan vast te knopen.

Ook dit weekend was het weer van dat. Een nieuwe peiling, groot nieuws. Onterecht. Bekijk de grafiek en oordeel zelf:

Het enige nieuws is dat N-VA voor het eerst een peilbare groei doormaakt. Het verschil (+4,8%) is immers groter dan de foutenmarge (+-3,57%).

Mischaël Modrikamen kan zich tussen de processen door verheugen op 4,3% van de fictieve hypothetische stemmen voor zijn PersonenPartij. Of 0,73%. Of 7,87%. Een foutenmarge werkt immers zowel positief als negatief. Met een foutenmarge van 3,57% is er bijgevolg een whopping gat van 7,14%. Ruimschoots de kiesdrempel van 5% dus.

Liefste journalisten. Het vertrouwen van de mensen in de journalistiek is bijna even laag als dat in de politiek. (Enfin, dat blijkt uit – jawel – allerlei peilingen.) Misschien is het tijd om dit soort nonderzoeken³ een beetje kritischer te bekijken? Het zou mij in ieder geval meer vertrouwen kunnen geven in de pers.

Willen we dat afspreken?

Kleine update: Vandaag wordt de peiling ook in De Morgen herkauwd. Zonder vermelding van de foutenmarge, maar met spectaculaire verklaringen. Wat het helemaal grappig maakt is de laatste paragraaf:

Toeval dat de foutenmarge in Brussel 4,38% bedraagt?

Disclaimer: Ik ben geen statisticus, nog journalist. (Al heb ik in een ver verleden wel een tijdje journalistiek gestudeerd, maar dat is een ander verhaal.) Voor hetzelfde geld bevat mijn uitleg enkele grove fouten. Hou u niet in om uw aanvullingen te posten via onderstaand commentaarformulier.

With that caveat out of the way, let’s see what the fuzz is all about.

SureBackup
Just to reiterate once more, Veeam SureBackup is not a stand-alone product, nor one feature of the upcoming Veeam Backup & Replication version 5. It is a collection of features that build on the existing functionality of VB&R.
The most significant feature is called Recovery Verification. With this feature, backup operators can opt to automatically verify the integrity of each and every backup.
To enable this, VB&R includes an NFS daemon that presents the images on the backup pool as a regular NFS datastore to the ESX farm. VB&R automatically registers the virtual machines (in a fenced-off network) and tests if the backup is bootable. (And doesn’t get stuck in Safe Mode, for instance.) This mechanism is OS- and application independent.

Recovery Verification is done in a dedicated resource pool. This can be a resource pool on the production environment, but it can also be a separate pool. This way, cross-site or remote backup is possible, with verification at the disaster site. The verification process takes application dependencies into account. (For example, it can know the relationship between and the correct boot order of a typical mail environment which includes DNS – AD – Mail – Blackberry – … servers.)

The verification wizard can also be used for application-item recovery (boot from a specific backup and then enter the application to recover the lost data), and for on-demand production replicas. (Eg. to test the impact of a patch in clone of the complete environment rather than a cloned individual VM.)

What about SRM?
With all this Virtualization-Powered Protection, the overlap with VMware’s own Site Recovery Manager increases even further. Veeam doesn’t openly position Backup & Replication against SRM, but it will surely be used to cover a lot of the SRM use cases. (As it is today.)
Furthermore, Veeam B&R is quite a bit cheaper than SRM, and doesn’t rely on – expensive – storage array replication. If you require near-zero RPO and RTO, then of course you’ll need storage based replication. (Preferably synchronous.) If ‘a couple of minutes’ is acceptable, you can run the backup or replication job as often as possible, thereby achieving a similar (but not equal) level of protection at a fraction of the cost. (Veeam calls this SmartCDP.
Of course Veeam B&R doesn’t do IP renumbering and other advanced recovery functionality present in SRM, but these kind of features will undoubtedly be something the active PowerShell community will get to work on. The on-demand production replica functionality even covers some Lab Manager territory.

Pricing and Availability
As mentioned before, Veeam Backup & Replication will be available in Q3 2010. There is no sign of a public beta yet. Pricing and packaging is not known at this time, but Veeam is expected to introduce two versions, at two price points. Hyper-V support, first expected at the end of 2009, is something that’s still being worked on.

UPDATE: More info can be found in the just-released Press Release.

“Welk nummer van Milk Inc. zou niet misstaan als naam voor een meubelzaak?”

Klik hierrrrr voor het antwoord. I LOL’d.

Ga uw gang met deze afbeelding, en post het resultaat van uw werk, bv. via e-mail (lodev [a] deze website) of op Twitter (tag even met #lolonard).

De eerste resultaten na de ‘jump’.

(by @krisbuytaert)

(by @roelvangils)

Nu is het aan U!

Wat doe je dan in zo’n geval, als je deze prachtige doelkans ziet?:

“Gaat zo eens kijken wat de buurt hier bied op culinair gebied.”

Welaan, dan denk je:

1. Thee is ook een soort drank, dus dat valt onder “culinair gebied”.
2. Er mankeert daar een “T”.
3. “T” klinkt als “Thee”
4. Binnenkoppen die handel:

“@de-persoon-die-de-vraag-stelde: T misschien?”

Om dan volgende reactie te krijgen:

“@lode Als je klaarkomt op mensen verbeteren op twitter is dat jou ding! Ik heb andere hobby´s! Maar val mij daar aub niet mee lastig.”

En dit komt dan van een persoon die op zijn blog proclameert:

“Het zei duidelijk, Twitter zit zonder zelfspot en humor! Op eieren moet je lopen voordat je een grap maakt. Zelfspot is voor mij nog steeds de beste humor die er is. Geen relativeringsvermogen: unfollow”

We zullen zijn goede raad dan maar opvolgen zeker? #unfollow.

It’s a bit unclear what features one should expect, besides speed, that OpenDNS doesn’t already offer.

Quick comparison:

• Google DNS servers are a bit easier to remember: the nameservers are 8.8.8.8 and 8.8.4.4. OpenDNS uses 208.67.222.222 and 208.67.220.220
• uses the global load balanced infrastructure by Google, which usually works very well and fast (unless they accidentially blackhole a part of the net) – OpenDNS has a somewhat similar distributed, albeit a bit smaller infrastructure.
• Google explicitly does not provide wildcard results to redirect users in case of typo’s. RFC purists will agree with Google that “Sometimes, in the case of a query for a mistyped or non-existent domain name, the right answer means no answer”. OpenDNS takes the opposite approach by providing users with search results when a page is offline.
• Google doesn’t seem to offer a feature like SmartCache that can show you cached versions of a page after it gets knocked offline due to Slashdotting/Fireballing

I’m going to try out GPDNS on a few systems and check the performance, but I’m not sure if I will use it as my primary DNS service any time soon.

Google Public DNS is another step in the way of Google becoming synonymous with “the web” or even “the net”.
As Laurie put it:
“Dear Google: you are supposed to be a site on the Internet, not be the entire Internet all by yourselves. Public trust only goes so far.”

Others agree that this seems a bit creepy. (Of course, OpenDNS has got its criticisms as well. - Gruber likes Google better.)

One thing to remember – take out your tinfoil hat:

Whilst Google’s motto is “don’t be evil” and their mission is to “organize the world’s information and make it universally accessible and useful”, they’re primarily an ad company.

They make money selling ads. Any tool that better helps them detect user patterns helps them to deliver even more targeted advertising. When you’re using the Google DNS servers, it’s trivially easy for them to ‘watch’ you even when you’ve cleared all your cookies and aren’t surfing on one of the many top web properties where they’re present, be it through an acquisition (Youtube), an ad deal (Myspace), the advertising network used (Doubleclick / Adsense) or one of the many free tools theyprovide that provide useful information, but also feed directly into their giant database. (Google Analytics).

Jason Kottke and John Gruber believe Google’s motives are speed, speed, speed. Their privacy policy regarding Google Public DNS sounds OK, so this might be true. Their track record however is not completely flawless. I don’t want to sound like a crazy conspiracy theorist. I’m just imagining the possibilities… (Man, now I even sound more like Glenn Beck )

One step further to EPIC?

Cougar Town, S02E09 – Here Comes My Girl (2009):
Jules (Courteney Cox, again) “saves” thanksgiving for all her friends who can’t make it home or whose families have fallen apart.
Andy – whilst eating a hamburger shortly before the big turkey dinner: “Thanksgiving is a food marathon. You have to warm up. This is like stretching.”

Really, ABC?

Last year I was in the market for new earbuds, after my second pair of Sony In-Ear headphones broke down.
My buddy Denny (whom I see IRL about as often as he blogs, which is way too seldom) recommended the Bose® In-Ear Headphones. I took his recommendation (even though the little f*ckrs are quite expensive) and have been very happy with the choice.

Shortly before my last USA trip, Bose released a version of the same headphones with an inline microphone, the Bose® Mobile In-Ear Headset. It’s compatible with most smartphones with standard jack connectors (and supplies adapters for brands with non-standard diameters).

It works great on the iPhone. The inline microphone also works as a button to pick up and hang up calls. It also functions as a remote control for the iPod application. ¹

The best thing about it though is that on the new-ish unibody MacBook, the headset actually works as a headset, which is great for Skype.

I’ve been wanting to do this ever since the iPhone first came out. At the time MacBooks still had a separate line-in and line-out, meaning the microphone part of iPhone headsets was ignored.

The 15″ and 17″ Unibody models still have separate line-in and line-out sockets, but apparently the line-out has a dual function on this models too. Yet another reason to treat yourself to a new laptop or at least a new headset for the holidays.
Update: The 13″ MacBook and 13″ MacBook Pro are at an all time low price now at Amazon (including free shipping), just in time for a happy thanksgiving!

The Los Angeles based clothing company is a favourite among designers for its high quality t-shirts that can be bought in bulk for screen printing. They pride themselves in being ‘locally produced’ (locally meaning the US) by well paid workers, as opposed to in a sweatshop in the far east. One well-known company that uses AA as a wholesale source is Threadless.

Cameron Moll is a renowned designer and author, known for amongst others his personal weblog Authentic Boredom, his book on Mobile Web Design, and the job site Authentic Jobs. I admire his work and count myself lucky for having viewed one of his talks in person. (Way back at @media 2006 in London.) Cameron takes offense at the advertising, which is “suggestive at best and explicit at worst”. He feels offended by this, and has decided to “put his money where is mouth is”. He will no longer use American Apparel as his source for t-shirts, nor buy any screened tee which is printed on AA materials.

American Apparel billboard by Flickr user @MSG

There are indeed some quite explicit ads, which I understand might be offensive to some, like Cameron. I completely respect that.

There’s more to this discussion though, as is evident in the comments that follow Cameron’s article. As you might know, Cameron was previously employed by one of the driving forces behind the campaign to amend the California Constitution to ban same-sex marriage, or as they call it “Preserving the Divine Institution of Marriage“.

While Cameron tries to “keep Church and State separate” in his blog posts and comments, I do feel this is relevant here.

On the one hand, if you look a bit further than the posters that seem to have been created by hormonally supercharged 16-year olds, you’ll see a company that, whilst being an “evil corporate chain“, cares for its workers and generally does the right thing.

Compare this to the writings, videos and ads you find in the LDS Newsroom devoted to Prop 8.

I’ll take over the top sleazy advertising over hateful half-truths and whole lies any day.