Some time ago, during a job interview, I was asked at the end why in some version of my résumé I wrote in the soft skills section “attention to detail; good in finding correlations; good in data mining”*–especially the last two “skills” drove the curiosity of one of them. I don’t remember what I answered exactly, but with regards to the last one, I mentioned that sometimes I was able to find in seven minutes of googling something that people were struggling for hours to find–it’s a matter of identifying the relevant keywords that would enable you to find the needle in a haystack. To tell the truth, I believe all three skills are working hand in hand in real life; and they are partially talents or gifts, and less of something that can be learned.

Now that I made that introduction, I can tell you how pissed I am at those so-called professionals who are making ludicrous mistakes they never acknowledge and that most people don’t even notice. I forgot to mention that the aforementioned interview was about a Software Test Engineer position–hence my critical position about IT professionals whose errors are easily tolerated, but people like me–able to spot an error from a thousand miles sometimes–never get the appropriate recognition of their abilities. Yes, I’m somewhat bitter and resentful, and not only because of the bugs I reported over time on different open-source projects (you know, those bugs that are never resolved, and years later someone closes them because in the meantime GNOME or the Linux kernel or the respective distro are having totally different versions), and also not only because sometimes the recruiters are such asses.

In my previous post I’ve shown how many SSD reviewers fail to properly design their benchmarks so that the capacity of the internal cache (that can have as much as 12 GB) is exceeded, and the true writing speed of the device is determined. This might look a minor flaw, but when such a SSD has a really pathetic performance, yet it’s ranked as #3 by some specialized magazines, and it’s proposed at the same price as much better brands by a notebook manufacturer (let’s call it Schenker)…

I have to admit I’m not forgiving when it comes to reviewers. I take a lot of information from hundreds and hundreds of reviews, being them made by specialists or just end-user feedback in English, French and German. I don’t change my laptop and my smartphone often enough, and I don’t own a tablet–yet I’m well-enough informed through the power of corroborating data from so many sources.

❶ Let me show you an incompetent idiot. In January 2013, Daniel Schuster from NotebookCheck.net made a review of Sony Xperia J. The German version includes this assertion:

Ganz deutlich fällt ein Blaustich bei allen Bildern auf. Dieser trübt sehr stark die Farbwahrnehmung. Bei Außenaufnahmen rauschen die Bilder weniger, der Blaustich bleibt jedoch bestehen. Ebenso fehlt es allen Bildern an Schärfe.

The English version puts it this way:

bluish cast is very evident on all pictures. This affects the color perception massively. The pictures do not show as much noise outdoors, but the bluish cast remains. Furthermore, all pictures lack sharpness.

A reader made this comment two months later:

Regarding the bluish tint on the camera I own this phone and also noticed this so I took off the protective blue film and now it’s perfect. As a reviewer I would expect a closer inspection of the phone.

Another reader added:

“A bluish cast is very evident on all pictures”. Try to remove the little blue plastic over the objective.

Do you think the reviewer has amended his review? No, Sir. No way.

Let’s get clear: it’s not some retarded gibbon who trashes a phone on Amazon, on Saturn or MediaMarkt, etc. This is about a professional and about a reputable website. And he even got helpful feedback from the general public!

❷ Since we’re at this website, let me show you how they fail to adapt a script that adds colored symbols next to the RAM and internal storage: all their smartphone reviews have red or yellow icons even if they have great specs, simply because the retards are using a script meant for computers, and this script only gives a “green” rating for 4 GB or more of RAM, and 1000 GB or more of storage!

memoryicons

Yes, they’re that stupid. 2 GB of RAM is not good enough (not even 3 GB, they really need 4 GB to turn it green). Also, 20 GB of free internal storage is bad juju: red! It’s a fucking smartphone, guys!

❸ Let’s now get to another kind of blindness. I’ve read a few reviews of an entry-level smartphone that has however a 4000 mAh battery, which makes it potentially attractive to some: Archos 50 Power. Here’s a review on NotebookCheck.net: do you notice anything special about the the rear-facing primary camera, other than it has “a resolution of 13 megapixels”? Not really. Here’s a French review on 01net.com: “Le capteur de 13 mégapixels n’est pas vraiment bon.” Another one: “Archos offre un rendu photo très moyen malgré la présence d’un capteur de 13 Mpx. Le constructeur semble bien ne pas s’embarrasser d’algorithmes poussés pour affiner les prises de vues. Résultat, les clichés sont un peu ternes et bruités et surtout bancals sur le niveau de détails.” OK, what gives?

The real issue is that almost all the smartphones under the Archos brand are having fake back camera resolutions, i.e. they use software interpolation, and they write it on their website! (What they don’t write is the actual sensor resolutions, but the typical rule is that an interpolated 13 MP camera has a 8 MP sensor, and an interpolated 8 MP camera has a 5 MP sensor.)

In our case: “Back camera 13 MP* AF + Flash LED” and in small print, at the bottom: “*Software interpolation.”

Believe me, this is the case with most Archos smartphones, and most reviewers never notice this! Unbelievable.

To examine the pictures from the French review, let me inform you that what you see here is not noise, nor the normal texture of a material, but interpolation artifacts:

Archos50P

If your eye is not trained to identify software interpolation artifacts, let me show this phenomenon as manifested by my Huawei Ascend G700, a smartphone whose specs don’t mention the interpolation, but I discovered myself that 5 MP is the real sensor resolution, not 8 MP:

IMG_20160708_132246

Left: interpolation artifacts. Right: just noise (and a lack of detail).

Now, why are the bloody manufacturers forcing a software interpolation (I’d have called it “extrapolation” from where I see it) instead of using the native resolution… beats me. It’s not like the user of a cheap phone would print posters with such photos, and even if they did, any imaging software would obtain much better scaled-images should they be given the original, unaltered pictures taken by the sensor!

With interpolated photos, one has to scale them down to get rid of the artifacts, and the image that doesn’t show them has to be much smaller than the non-interpolated image! For instance, when talking “8 MP” (interpolated) pics with my phone, I have to scale them down to 50% (so to 2 MP) to get the visual quality of what I get when taking unaltered 5 MP pictures!

Side note: most phones come with the stupid camera default that says “16:9 aspect ratio,” although the sensor has a physical 4:3 aspect ratio. Photos have always been in the 4:3 aspect ratio, being them on film or digital. It’s up to the user to decide whether they need a different format, and to crop the result accordingly. But the poor user of a smartphone usually doesn’t understand why a camera advertised as 8 MP produces 6 MP photos, and a 5 MP camera outputs 3.7 MP images–that’s because some stupid ass thought that people bought a smartphone to use it as a movie player, hence the 16:9 display format, and that photos must fill the smartphone’s screen with no gaps. Hey, but the bloody sensor is 4:3, why not using two bands of it? It doesn’t make sense to forcefully crop the image, let the user do it when needed!

Back to software-interpolated photos, I just discovered that the entry-level Wiko Robby also has them, despite the fact that the official specs don’t mention it, and Wiko generally makes “honest phones,” without interpolation. Paying 129 € for a 5.5″ phone with an aluminium case shouldn’t entitle you to great expectations, but since when is honesty a luxury? OK, maybe they thought (just like Huawei) that 5 MP would not sell nowadays, so they had to up it to 8 MP.

The software interpolation can be seen quite unequivocally in this photo found in a German review. This is not noise:

wiko_robby_Tagaufnahme_Original

Similar interpolation artifacts can be found in the photos taken by Florian Wimmer from NotebookCheck.net in his review of Robby, yet he only notices that “the photo quality is not very satisfactory when compared with higher-quality lenses. Details look vague and surfaces spotty when zoomed.”

IMG_20160624_130001

Same thing, in another German review:

robby-IMG_20160529_140244

They might seem noise to you, but believe me, they’re interpolation artifacts. Just set the camera resolution to a lower resolution (8 MP instead of 13 MP, or 5 MP instead of 8 MP) and you’ll find a much, much better image, on which the real noise would appear quite different.

A funny thing I noticed in a German review of an even cheaper Marshmallow phone, Wiko Lenny 3: while the reviewer claims that the pictures are worse than those taken by Robby, I believe that the artifacts are by no means worse:

L3_IMG_20160609_110609

Disappointment note: it looks like Wiko loves to cheat on some other models too, although apparently not on Fever, U Feel and U Feel Lite: a reader comments on Wiko Pulp 4G: “Photos correctes si on paramètre en 8 méga pixel,” which suggests that the 13 MP camera is indeed using an interpolated 8 MP sensor.

❹ Getting now to another idiocy, let me tell you that I happened to have recently read (and watched) reviews of a few cheap phones by Wiko, and I’ve learned that the following cheap CPUs present the following range of scores in AnTuTu 6.x:

  • 4-core Cortex-A7 (32-bit ARMv7) MT6580 at 1.3 GHz (in Wiko Robby and Lenny3): ~25 k (distinct tests: 2423825150; 24860; 23897; 25078; 25237; 23921)
  • 4-core Cortex-A53 (64-bit ARMv8-A) MT6735 at 1.3 GHz (in Wiko U Feel): ~30 k (distinct tests: 2953930209; 30044; 30004; 29939; don’t count the MT6735M models running at 1 GHz in Archos phones)
  • 8-core Cortex-A53 (64-bit ARMv8-A) MT6753 at 1.3 GHz (in Wiko Fever): ~36 k (distinct tests: 36661; 36707; 34151; 37104; 35830; 37027; 36000; 3701535910 or 36061 in Acer 630S; and I ignored some tests made in AnTuTu 5.x)

Guess what? A genius from NotebookCheck that goes by the name of Florian Wimmer (hi again!) managed to get (in German too) for the CPU in Wiko Robby the “fabulous” AnTuTu 6.x score of 15797!

The review mentions that the very same CPU got 23997 in LG K7 on the same site, but the reviewer doesn’t seemed troubled by a 33% loss of performance! I’ll let you judge the professional qualities of this reviewer…

OK, let me help you: when you perform a benchmark, you should make sure there aren’t unwanted background apps and processes, downloads, APKs upgraded by Google Play, etc. This is even more important for such an entry-level phone that only has 1 GB of RAM! Better off, reboot the phone and wait a couple of minutes before starting the benchmark. Oh, and make sure the battery is charged (how about plugging the charger during the test?) and that the energy-saving features are not decreasing the CPU speed–we have here both Marshmallow’s and Wiko’s own power management features, but you have to have more than two neurons to realize that. Finally, you should know the general level of performance of a specific hardware–especially when you fucking tested it in a different device!–and have a bell ring when you’re facing suspicious values.

Don’t misunderstand me: NotebookCheck’s reviews are generally of high quality, and they consider aspects some other sites don’t even touch (PWM flickering, color accuracy, brightness and contrast, real-life GPS test, and more), and when I chose my laptop, my decision took into account one of their tests. But let’s call a spade a spade.

Well, I didn’t need such example to debunk the old myth of “German quality/precision/accuracy/etc.” I knew that today’s Germans are by no means up to the legend…

EDIT: I’ve found some more weird situation. A review made by a 15-year-old German found Robby to have an AnTuTu score of 16930, which simply can’t be with that CPU and GPU. The reviewer correctly assumes that the RAM of only 1 GB might be the culprit (“Als Grund des Übels vermute ich den sehr knapp bemessenen Arbeitsspeicher von nur einem Gigabyte.”), but he failed to make sure that no unwanted background apps exist, and that “Phone Assist” or other energy-saving features are properly configured for maximum performance. There is also a possibility that some builds of the firmware have problems, as an Amazon customer implies in an anger comment which first says that the device is bad, then on June 28 the devices works better and it’s also more responsive after a firmware update, only to find that after one more update (version 20), the device is bad again! Now, the French Wiko Robby update page only offers version 16, while the German page does the same, but it also lists as previous versions… version 20 and version 16, and that means they made a rollback, by retiring version 20 and keeping version 16–the good one. [UPDATE: Now the current update is V.25, and the previous are listed, in order: V.16, V.20, V.16.] Either way, a professional reviewer should investigate a 33% drop of performance, and alter the default settings in an attempt to measure the best performance of a phone. After all, several other people have measured better–and credible–values.

❺ As a final example, this time before being able to offer the URLs of the incriminated reviews, there is one thing that bothered me a couple of months ago, as I was reading reviews of Acer Aspire V3-371 and V3-372 (prior to decide for another model).

At least one reviewer explicitly stated that he’s unable to understand why Acer chose to eliminate the spacing between the ENTER key and the key that’s close to it on these laptop series. I’ll show you upfront two pictures that should tell it all:

aspirev3

The top picture shows the ENTER key as it is in the UK and the German layouts; the bottom one shows the US layout. The answer is simple: Acer tried to cut the costs by having a unique casing with a single perforation template regardless of the targeted keyboard layout, and this included an L-shaped hole for “ENTER and whatever key is close to it” no matter ENTER is supposed to be a horizontal line (e.g. US English, Canadian French) or an reversed L (e.g. UK English, German, French). This is cheap and ugly, but this is a perfect explanation–and the true reason they took this decision. (A couple of years ago, some car manufacturers started to make dashboards with a central positioning of the dials–I always thought this was made so that they don’t need to make partially mirrored dashboards for UK, Australia, Japan and the like. Maybe the public was angry, because they now reverted to left dials for “normal” driving, and right dials for left driving.)

All in all, I won’t eat my words: people are stupider than they should be to perform their jobs adequately. Or maybe, why not, they just lack some basic skills called “attention to detail; good in finding correlations; good in data mining.” (Most people also lack common sense and judgment. Remember all those fake alarmist news that spread through e-mail or on Facebook? That’s because most people, regardless of their education, chose not to think and not to use their judgment. Googling or specifically searching on sites such as Snopes.com would have reduce such hoaxes by at least 80%.)

❻ A last word to my… no, not sponsors, but recruiters. Some of them are unaware that their lies are visible from the Moon. When there is a job opening when they really hire another guy, I can accept a reply that reads “we have taken the decision to hire a different candidate who fulfills the position requirements even more comprehensively,” but when it’s about a job advertisement that runs and runs for months, so it’s clear that its purpose is not to fill a specific vacant position, but to act as a honeypot that enables them to hand-pick professionals that are both competent and cheap, it’s only sheer stupidity and gross disrespect to come with a text saying “there was a tremendously difficult task for us to take this decision, but eventually we resolved not to go forward with your application, as we opted for a different candidate for this position.” Such assholes simply ignore that it’s only too simple to notice that the ads go on unchanged, so it wasn’t that they found a better candidate and the position is no more available–they simply discarded an application without being able to use a less outrageous lie.

By the way, this includes people who recruit for their QA departments. They want competent, rigorous and observant people while on the job, but otherwise unobservant applicants who are stupid enough not to notice the degree of lying they receive. It’s easy to infer how much I trust recruiters and how much I love them. (The feeling seems to be mutual though.)
__
*my CV actually had it “minng” instead of “mining,” which is actually unacceptable