A Troubling Trend

Did you know that Carrizo is being made on a 20nm node, has six cores, and turbos to 4.4GHz?

Well, its flagship, the A10-8890K does.

…Except all of that is a load of crap. You might have seen that on some news sites that are generally considered reputable. There seems to be a trend. Fact-checking is no more, but click-baiting certainly is. It's not like it really matters in the end. After all, if you're first and the rumor was faked, you get just as much ad revenue as you do if it were real. I wanted to see if I could take advantage of it and start a rumor of my own.

My goal was to make the most convincing possible CPU-Z screenshot of a fabled Carrizo flagship, but with one exception: it would be 20nm and have six cores. In addition, I would hide some messages in the background of the picture that would be nearly impossible to spot but could be fairly easily revealed in Paint or some other program. I didn't want to start a rumor that people would believe per se; fooling readers was not my goal. I did want to start a rumor that would encourage click-baiting and would, hopefully, make the sites that published them look a bit foolish in the process.

On July 25th, I posted the following in the One Million Post Thread (OMPT):

Quote:

Can I request some CPU-Z screenshots of any Nehalem chips? Saved as uncompressed .PNGs, please. I need to start some rumors, and by start some rumors, I mean get WCCF some sweet, sweet ad dollars.

My goal was to look at a number of different chips and compile them. I had remembered Nehalem as being 40nm for some reason, realized my mistake, and corrected myself. Turns out that didn't matter. CPU-Z screenshots are incredibly easy to fake. Each character, regardless of what comes before or after or where it is located in the read-out, is identical to the pixel. All it took was some copying and pasting. I came up with the following just three hours later:

Quote:

@Alatar (since you've seen your fair-share of crap leaks) and others, how convincing is it?

You think a news site will pick this up? biggrin.gif

Well, apparently having something with six cores and an “ELITE QUAD-CORE” logo is a bit of a discrepancy. Whoops! I made three revisions in the next three hours, making everything as accurate as I possibly could. The fourth and final one I made looked like this:

Fantastic. It matches AMD's recent announcements and legitimate looking leaks and rumors, such as having 1MiB of L2 per module, 65W TDP, and staying on socket FM2+. The only parts that I intentionally incorrectly faked were the six cores and the 20nm node, as those had been confirmed or rumored to stay quad-core and 28nm. The “ELITE HEXA-CORE” logo is new and realistically wouldn't be in the CPU-Z database.

Well, I had gone elsewhere to get food or something and returned to find TheLAWNOOB's version:

I'll admit, I mentally facepalmed. How could anybody take this seriously? They just confirmed it was staying on FM2+ and the clockspeed was stupidly high! Well, at least it didn't come with R9 graphics or SMT (which somebody else had suggested I include earlier). He posted the following to OCN in the AMD CPUs section a half hour later (https://www.overclock.net/t/1504122/amd-carrizo-apu-spotted-features-fm3-and-20nm-afterall/0_100):

Quote:

A CPUZ screenshot was leaked by a reputable source from the semi-conductor industry earlier today.

The follow screenshot confirms a working, nearly finished sample of the next gen AMD Elite Hexa core APU, the 20nm Carrizo.

It features a bumped boost clock of 4.4Ghz. The stock clock is unknown at this time.

It features a modest TDP of 95W, which is decent considering it's an Hexa core APU with embedded AMD Radeon R7 Graphics on board.

As usual, AMD refused to comment on this leak.

The OMPT guys might have helped give this rumor some traction. I was amused, but I was skeptical it would take off. A lot of the inconsistencies had been confirmed, and quite recently at that. I checked news sites the next day, July 26, but found nothing. This morning, July 27, I don't think I could have timed it better. PunkX 1 linked to a (now deleted, but I assure you, we have it archived) Guru3D article in the news forum, and I caught it within a minute or two of it being posted at 11:51.

https://www.overclock.net/t/1504354/guru3d-20nm-amd-carrizo-apu-spotted-features-fm3/0_100

For hopefully obvious reasons, I then linked to it in OMPT. We had quite a bit of fun with it until Alatar, understandably and buzz-killingly, locked it. I did have the decency to point out the secret messages, for what it's worth, and thank God they didn't save the image as a .JPG – the compression would have rendered them unreadable:

And yes, that is the exact version G3D posted. This one just has the messages revealed.

The article and image were deleted at around 1:00 P.M., unfortunately, but we had saved copies (which will be discussed later) of the article and its two later revisions. I thought it was over. Well, no. That would be too easy.

WCCF Tech is notoriously, uh, unreliable and click-baity.

Naturally, they posted an article about it. When? About an hour after G3D deleted theirs. Um, okay. Well, they posted it, and as it turns out, it gained credibility because G3D posted it! (Article: http://wccftech.com/amds-carrizo-apu-a10-8890k-cpuz-hexacore-20nm/, Image: http://cdn3.wccftech.com/wp-content/uploads/2014/07/AMD-20nm-Carrizo-APU-A10-8890K.png)

Quote:

Now a CPUZ from a reliable source is pretty much what any news house worth their salt hopes for. And this appears to be just that.

Well, I hope that's why. Otherwise, it seems the default status of everything on the Internet is true. (As an aside, I am a Nigerian prince…) I will give the author some credit for tagging it a rumor, but he only did so because of one discrepancy:

Quote:

However, one thing does bug me about this CPUZ shot. Note the readings at the cache, level 2 to be precise. It says 3×1024 KBytes. Thats just 3MB of L3 Cache right there. For 6 Cores that is a very low amount of cache. So either this screen is fake after all, or, AMD might be working some mysterious magic where the memory is concerned.

I halved the cache from the Kaveri screenshot (ironically downloaded from G3D in the first place) because VR Zone (Source, not translated: http://chinese.vr-zone.com/121197/no-ondie-stack-ram-amd-carrizo-soc-structure-shown-07162014/) did show a legitimate-looking slide that indicated Carrizo would have 1MiB of L2 per module. That was what tipped the author. The likely/possibly decreased L2 capacity. Not the new socket (when AMD confirmed FM2+ is sticking around for a while), not the 95W TDP (even though roadmaps show 65W as the maximum), not the six cores (even though most rumors are pointing to more quad-cores), and not the 20nm node (even though it's basically been confirmed that 28nm is sticking around).

In the WCCF comments, somebody pointed out that the G3D article had been deleted. The author responded with the following:

Quote:

Yeah I noticed that just now. But you know what they say. Once on the Internet, always on the internet.

He also linked to a cached version of the original article. Several hours later, the author debunked his own article as fake, at around 5:45 P.M. Why? The G3D article had been deleted. Before that, he had reposted it to another site (http://www.technationnews.com/2014/07/27/amds-carrizo-apu-a10-8890k-cpu-z-leaked-six-excavator-cores-with-95w-tdp-on-the-20nm-node/), and several user-posted articles and links appeared elsewhere.

All of that demonstrates a very, very serious problem in the tech news industry.

Nobody seems to have any amount of integrity. It's all about the ad dollars. M.O.E., if you will, or Money Over Everything.

I will give the author of the WCCF article some credit, as he kept his original article entirely intact, even when he did discover it was faked. Still, he did not change the headline to reflect that, and I think it's safe to assume that the first half of the title, “AMD’s Carrizo APU A10-8890K CPU-Z Leaked – Six Excavator,” is enough to net a click from a curious user. Had it said “[FAKE] AMD’s Carrizo APU A10-8890K CPU-Z Leaked” instead, then I might be a bit less frustrated. Well, at least we know that they have this habit of posting first and fact-checking later. It's expected, at least.

And then, we have Guru3D.

They are, to my knowledge, reasonably well-respected. They were the first to post the article and the first to delete it. This situation is a mess, to say the least.

First, plagiarism.

If any of you are in high school or college, will you please ask your English teacher or any professor to look at the following? I'll just repeat TheLAWNOOB's forum post here, the source:

Quote:

A CPUZ screenshot was leaked by a reputable source from the semi-conductor industry earlier today.

The follow screenshot confirms a working, nearly finished sample of the next gen AMD Elite Hexa core APU, the 20nm Carrizo.

It features a bumped boost clock of 4.4Ghz. The stock clock is unknown at this time.

It features a modest TDP of 95W, which is decent considering it's an Hexa core APU with embedded AMD Radeon R7 Graphics on board.

As usual, AMD refused to comment on this leak.

Now, we have G3D's article, the “paper”:

Quote:

Interesting, A CPUZ screenshot was leaked by a source from the semi-conductor industry earlier today. The follow screenshot confirms a working, nearly finished sample of the next gen AMD Elite Hexa core APU, the 20nm Carrizo. It features a bumped boost clock of 4.4Ghz. The stock clock is unknown at this time.

Details are still sparse regarding Carrizo, but it features a TDP of 95W, which is decent considering it's an Hexa core APU with embedded AMD Radeon R7 Graphics on board.
The successor to Kaveri was expected to support DDR4 memory though this is unconfirmed. The normal 95W implies a reasonably efficient architecture on a more advanced manufacturing process. With Kaveri being built on the 28nm node, it is interesting to see that Carrizo will be manufactured using 20nm SOI process. As usual, AMD refused to comment on this leak.

Ask them if that would be considered plagiarism, and point out that no sources were cited.

That was their article for a short time as they had originally posted it. That might be among the most blatant examples of plagiarism I have seen. Even the typos TheLAWNOOB made are intact! Anything to get a story, right? All dat ad money, right? Not only that, they failed to cite their source initially, and they made ad revenue off this. Let me say that again: Guru3D copied and pasted a forum post and made money off it. That is disgusting, and I'm sure that this has happened before.

They updated this article twice, and each update was a backtrack to distance themselves from the rumor as much as they could. The updates essentially added two disclaimers (thanks to PunkX 1 for keeping track of these):

Quote:

Let me just add that this screenshot should be taken with a rather big grain of salt and could very well be a fake. There is no way to proof its validity at this time whatsoever.
Quote:

A CPU-Z screenshot was leaked by a 'supposably' source from the semi-conductor industry earlier today. At least that is what is claimed in the overclocked forums

Classy. Good job, guys.

geggeg reminded us (or at least me) of the issue with EVGA and PrecisionX and Rivatuner.

G3D had decided to take the moral high ground, as he put it, and condemn EVGA for copying features and code from the program. I decided that an apology and a calling-out were in order. After all, they plagiarized. They did exactly what they condemned EVGA for doing. So, because I'm either brave or a moron, I sent an email to them using an email address that contains my actual name (censored, of course):

Quote:

Don't think deleting the Carrizo story makes it go away. I have saved copies of it, and several other people have screenshots. You plagiarized. Shame on you. You criticize EVGA for copying features of another program (that whole PrecisionX/Rivatuner thing?), and then you do THIS? Try that in any remotely reputable university and see what happens.

*Their article*

*TheLAWNOOB's post*

The first quote comes from your article. The second quote comes from the forum post you took it from. You kept the exact same typos as well! You can apologize, or you can be outed as hypocrites who are only in it for the ad dollars.

The quotes were there, but I edited them out here to save space. I received the following reply:

Quote:

Hello [Uni],

We posted a news-submittion 1:1 that was submitted from one of our readers. We sourced the story (you probably did not see the source link?), and after 50 minutes we noticed this was a fake screenshot and thus removed the post again.

But yes you are right my apologies, we should have looked better at this story and quoted the original lines. But was this click-bait from EVGA users ? That's a little awkward.

Kind regards,

The Guru of 3D – http://Guru3D.com
The Hardware Guru – http://HardwareGuru.com

– “Always remember you're unique, just like everyone else.”

Oh, that makes it okay, doesn't it? It was user-submitted? And it was cited, was it? Well, no. I wasn't paying close enough attention to the revisions the page had gone through, and I sent them the text that contained the half-arsed citation. That was stupid. I'm not sure if his attempt to justify this is better or worse, actually, knowing that.

Now, let's go back to the plagiarism point I made previously. They initially cited absolutely nothing. I doubt they would have even bothered if they hadn't realized the rumor might be a load of crap. They don't want to look bad, do they? Remember, they added their “citation” with a caveat saying the screenshot could be fake, and it was the second update. Tell me, which site is “the overclocked forums”? Is it Overclocker3D? Extreme Overclocking? Maybe it's Overclock.net? Could it be Overclockers.com? Yeah, there's a bunch of those.

Let's be nice and assume you get lucky by picking OCN. Where is this forum post located? No title was given. Searching “Carrizo” or “Carrizo leak” is going to yield too many results to search. “Carrizo 20nm six core” does yield the correct thread, but that's not exactly the first search term somebody will think to use.

I suppose, since it's technically possible to find, that this is all justified, right? Yeah, no. Let's consult how MLA tells us to cite a web source. It requires, among other things: the name of the article, the name of the author, and the name of the website; an exact URL is no longer required. Was “AMD Carrizo APU Spotted Features FM3 and 20nm Afterall” cited? No. Was TheLAWNOOB cited? No. Was Overclock.net reference? Barely. I'll be generous and give them half a point. It's not like that matters. Any reputable university would have them kicked out by now.

Second, the administration of Guru3D decided to try and sweep this all under the rug.

In the forum thread (http://forums.guru3d.com/showpost.php?s=c83b707c64669307d229fb396a7e534c&p=4882745&postcount=9), also the source of the comments on the original article, we get this gem:

Quote:

This screenshot is a confirmed fake you guys. Moving on wink.gif

LOL apparently it's a new thing to spread and hoax false info on forums ? This is hidden on the screenshot:

That's right, I'm the bad guy for taking advantage of a trend and testing a hypothesis. Not the guy who posted the article to begin with. Not the guy whose article contained, initially, no disclaimer. Not the hypocrite who plagiarized a forum post. Not the hypocrite who proceeded to make money off it anyway. Not the coward who retracted the article instead of pointing out that it's fake. Not the coward who can't stand being seen as wrong.

If you're going to post rumors you found on some forum somewhere, would you please have the decency to not delete them later? Would you please clearly call them fake and admit you were wrong? Would you please leave a warning that this might not be true to begin with?

Guru3D has lost any credibility it once held with me.

I apologize if you were tricked into believing this, but let this serve as a warning. Any rumor you see could have been easily faked, and “news” sites don't seem to care about fact-checking anymore. Anything you see should be treated as click-bait, and every article you read should be questioned. Don't think this won't happen again.

28 thoughts on “A Troubling Trend”

  1. I was afraid they would put the blame on “EVGA users” for “baiting” them the minute they saw your reference to their Precision-X/Afterburner posts. No such thing happened in case anyone was wondering. I really had no idea myself about this being a hoax till I saw Alatar mentioning it as such. This is unfortunately the same everywhere in most fields where ad-based revenue affects journalism- anything for more views, we need to be first, let's fact check later/never!

  2. WOW….. The way of the world, it's all about the money! Who cares about anything else?
    Who needs truth when there are dollars to be made.
    Stand for this and fall for anything. The story of our modern society,… and our downfall as a nation.

  3. I was very surprised by Guru3D, Hillbert and Unwinder when it came to EVGA Precision X. After I did prove Unwinder's claims that EVGA had some remnants from a previous Precision build they had used, I also proved the program was vastly differently from the work Unwinder had done. Even the similarities of EVGAs menu is laughable as EVGA was the one who supposedly designed/built them.

    All EVGA had to do was change some marketing shtick, and apologized for the confusion. Instead Guru3d led a crusade against EVGA and I found the entire ordeal unprofessional.

  4. Great work! Although I have seen this happen one to many times sadly. So many of the the tech rumors today are so fake it's not even funny.
    I can spot most of them (and many of you), but normal people can't and get fooled.

    Real shame about G3D though, have used them since the beginning. Now I know that I can't take rumors from them seriously at all..

  5. this is all explained on how the Internet works now for “News” in all areas not just tech. See the book called “Trust me I'm Lying” explains it all. Sources are no longer wanted nor needed by internet publishers.

    Have a crap B movie you just made and want to get Sales? Get an Internet Promo firm to put up flyers about your move, then deface them, and “pay” a blog to notice it , of course send pictures of the defaced flyers to the blog. If a religious blog even better. And stand by for the typical Yahoo,MSNBC,CNN,HuntingtonPost “Outrage”. yet this will have publicized your no name movie and people will flock to go see it…

  6. Click baiting is however done by like almost all websites. Ghostery ensures they don't get paid for my visit. (in the long term as marketing is less effective that way) Not that I was inclined to be frail of mind enough to let annoying commercials actually make me buy something.

    Good job on putting Guru3D in their place it was unjustifiable saying they had a source on the inside when it was really a fake.

  7. Wouldn't the original source need to be factual for this to be plagiarism? It's not an excuse but the general public being misled about when the next APU is going to come out and how much it will cost is kinda not worth the time, effort or cost of a lawsuit.

  8. I'm shocked someone didn't smell a rat right away. I saw that “CPU-Z screenshot” the day it broke, and while I'd like it to be true, there were WAY too many things about it that were suspicious.

  9. I have to say that I am not sure why anyone would ever be surprised that nonsense travels fast on forum boards and or so called industry news sites, but it did make me wonder just how many people visit a site such as this one per day……

Leave a Reply

Your email address will not be published. Required fields are marked *

Recommended