Tuesday, June 26th 2018

Revised NVIDIA Reviewers NDA Raises Eyebrows: Our Thoughts

An "attack on journalism" exclaims German tech publication Heise.de, on NVIDIA's latest non-disclosure agreement (NDA), a document tech journalists and reviewers have to sign in order to receive graphics card samples and information from NVIDIA. The language of this NDA, released verbatim to the web by Heise, provides a glimpse of what terms reviewers agree to, in order to write launch-day reviews of new products. NDAs are sort of like the EULA you agree to before installing software. There are NDAs for even little things like new thermal pastes, and reviewers end up signing dozens of them each year. Over time, it becomes second nature for reviewers to not publish before a date prescribed by the manufacturer, NDA or not.

The spirit of an NDA is: "we are giving you information/a sample in good faith, don't post your review before date/time/timezone." Such an NDA casts no aspersions on the credibility of the review since it doesn't dictate how the review should be, or what it should say. It doesn't say "don't post your review before we approve what you wrote." NVIDIA samples usually ship with a PDF titled "reviewer's guide," which only politely suggests to reviewers something along the lines of "here's our cool new graphics card that's capable of playing this game at that resolution with these settings, just don't test it on something like Linux with Nouveau drivers, because that either won't work or won't show what our card is truly capable of." Heise's close inspection of the latest NDA by NVIDIA suggests to them that NVIDIA is mandating positive reviews now. We disagree.
Over the past several launch cycles, NVIDIA and AMD have slated product launch and market availability on separate dates, resulting in reviewers being unable to buy graphics cards from friendly stores a few days in advance, to post launch-day reviews. Retailers that sell cards on market-availability day usually begin stocking up only a couple of days earlier, leaving reviewers with not enough time to write reviews with retailer-sourced cards, if they intend to post their reviews on launch-day (there are very few exceptions to this). This restricts reviewers to sampling directly from manufacturers; because publications get a lot more readership on launch-day than publishing their work weeks later, after getting cards from a retailer (by which time the public is generally aware about the product, and is less likely to read the review). Reviewers don't mind signing onto NDAs which tell them "you must not leak before NDA expiry time, or else no more samples."

On June 20th, Heise, along with several other publications (including us), received a notice from NVIDIA that they have revised their NDA, and that they must read and sign it before the 22nd of June. This new NDA needn't be a prelude to anything (a product launch or an event), but rather NVIDIA proactively collecting NDA signatures for future reference, so it could send future invitations/samples on short notice. This happens from time to time. Close inspection of the NDA reveals sentences such as: "the receiver uses confidential information exclusively in favor of NVIDIA," which Heise interprets as "you can't write a negative review."

Not all information shared by NVIDIA (or any hardware maker for that matter), is free to be disclosed at the expiry of review publication restrictions. NVIDIA's technical marketing people can sometimes put out off-the-record remarks or details to help reviewers better understand the product they're reviewing. These are usually 1-on-1 verbal communications between people who have built years of trust.

"Notwithstanding the expiration of this Agreement, the recipient's obligations with respect to any Confidential Information will expire five years after the date of their disclosure to the recipient," the NDA continues. Heise also interpreted the NDA survival clause (a standard component of most NDAs) as meaning that any information deemed a "trade secret" by NVIDIA (which if any technical marketing person is dumb enough to disclose to the press), remains embargoed forever under this NDA. "The protection of information, which is a trade secret, never goes out," it writes. Here is a crash-course on survival clause by a law firm.

A good example of a survival clause would be the NDA signed by The Coca Cola Company and a third-party company that manufactures its concentrate (so they need access to the top-secret recipe). This concentrate is shipped to bottling plants around the world, to make Coke as we know it. If Coca Cola stops sourcing concentrate from a particular supplier, the latter is still obligated under law to never disclose the top-secret recipe.

When Heise and c't protested with NVIDIA, they were told that "many journalists" have already signed up. TechPowerUp is among those "many journalists."

TechPowerUp did receive this NDA around the 20th, and promptly signed it, because we aren't reading too much into the controversial lines pointed out by Heise. I'm sure you won't spare us the criticism in the comments of this article. We've come across the phrase "in favor of" in many NDAs, not just from NVIDIA, and never once interpreted it as "favorable." This NDA is not going to stop TechPowerUp from pointing out any shortcomings of NVIDIA products, and none of NVIDIA's NDAs in the past ever have. During the review process, all NVIDIA does is check on progress, and whether we have encountered any abnormalities that they might be able to help with. Completely ignoring that inquiry is fine, and we've done so many times. Whenever we've come across bad products from NVIDIA, such as the GeForce GTX 480, or bad implementations of NVIDIA cards by its AIC partners, we've never hesitated to bring them to the attention of our readers, and will never stop doing so. One could easily argue that the drama after the GTX 480 launch was for the benefit of NVIDIA, because it pushed them in the right direction, to improve their product, which has led to their market dominance today.

Over the years, NVIDIA has tightened its grip over product launch cycle to ensure non-signatories or violators don't have access to samples, and so the NDA cannot be interpreted as a directive to only post positive reviews (lest NVIDIA ends up killing the credibility of every launch-day review, and jeopardizing its own product launch). Also NVIDIA doesn't need any NDA to cut off media that they don't like to work with for whatever reason. They can simply stop providing information or samples, it's not like NVIDIA has any obligation to work with everyone.

Public perception of NVIDIA has already taken a beating in the wake of the GPP controversy, and it's the duty of press to point out similar misadventures by the company, but maybe not based on misinterpretations of internal documents. We feel that Heise is overreacting and possibly looking to become a martyr, by just following the trend of bashing NVIDIA. Source: Heise
Add your own comment

160 Comments on Revised NVIDIA Reviewers NDA Raises Eyebrows: Our Thoughts

#1
close
@[USER=43587]btarunr[/USER]: Imagine we are in September 2014 when GTX 970 was launched. You find out from a leaker that Nvidia knows about the 3.5GB RAM issue and you plan to publish an article. But Nvidia preempts your move by officially disclosing this info under NDA (perhaps attached to a sob story of how they're working to fix it to make it look less "covering tracks" and more "working for you").

Are you still allowed to publish this?

Because if not I can see why people are worried. You could become Nvidia's secret keeper for 5 years. And by the time the NDA expires people don't really care much about old news, do they? Not the same impact, definitely not on sales.
Posted on Reply
#2
ajc9988
R-T-B said:
Seriously, you are correct if we can assume all data is gathered impartially and without bias.

Unfortunately, you know as well as I do that nothing could be further from the truth in the online journalism world.

Take it from me, this place is better than most at reporting the true facts of matters. And me backing up Dave is seriously not my style.
To be clear, I am not attacking TPU nor its journalists, nor its coverage or bias or lack thereof.

My point was more to discuss the fears related to the document and certain coverage, such as the GPP and Nvidia basically gobbling up partner's brand IP and trademarks and that coverage like that can now be pre-empted by giving confidential information to all outlets at the outset. After that, hearing rumors on the effect of the agreement is pre-empted in the NDA. This is not just about reviews.

But I did want to make clear that this is not making statements related to the journalists here or their integrity and is solely related to potential uses of the agreement.
Posted on Reply
#3
cadaveca
My name is Dave
close said:
@[USER=43587]btarunr[/USER]: Imagine we are in September 2014 when GTX 970 was launched. You find out from a leaker that Nvidia knows about the 3.5GB RAM issue and you plan to publish an article. But Nvidia preempts your move by officially disclosing this info under NDA (perhaps attached to a sob story of how they're working to fix it to make it look less "covering tracks" and more "working for you").

Are you still allowed to publish this?

Because if not I can see why people are worried. You could become Nvidia's secret keeper for 5 years. And by the time the NDA expires people don't really care much about old news, do they? Not the same impact, definitely not on sales.
NVidia could not prevent such info from coming out by NDA, because most NDA expire when the product launches.

ajc9988 said:
But I did want to make clear that this is not making statements related to the journalists here or their integrity and is solely related to potential uses of the agreement.
That was clear to me from that start, but perhaps you need to look at this situation with a wider angle.
Posted on Reply
#4
Captain_Tom
W1zzard said:
Not a single western company has ever tried to influence the outcome of my reviews in any way (including NVIDIA). Some Asia companies have tried (a little bit), but of course without success.

Why would you want to screw with reviewer results, and then get burnt by actual customers who encounter product flaws. Nowadays everybody can post their experience online. Results of hardware reviews can easily be reproduced, or are immediately noticed when comparing with other reviewers.
You are massively overestimating 2 things about Novice Consumers: 1) That they will do as much research as Enthusiasts, 2) That people will even notice they didn't get what they paid for.

People often spend money just to feel good after reading one review on a big website. Almost no one does in-depth research or fact checks their purchase after they have made it.

As long as Nvidia can get glowing day 1 reviews, they can be content that 90% of people will not notice anything that emerges after. I don't even have the time to give endless examples of this - but I will mention the fact that Nvidia STILL gets praise for "good drivers" even after it has been quantified by Microsoft crash logs that Nvidia cards crash more often than AMD on average. Oh, and after how many drivers from Nvidia brick cards or break youtube every year?


But let's see what happens here. [B]The article states TPU will stay honest, and generally speaking TPU has indeed been pretty trustworthy in their reviews over the years. I hope that remains true.[/B]

But drop this utter BS about Nvidia not having a motive to force stuff down reviewers throats. That's hilariously naive, and in fact makes me immediately question TPU's ability to even notice when they are being maliciously used for anti-consumer practices.
Posted on Reply
#5
R-T-B
close said:
Are you still allowed to publish this?
To quote a legal friend of mine:

"Yes, as long as you can prove you got the info another way besides from the NDA'd group"
Posted on Reply
#6
TheMailMan78
Big Member
R-T-B said:
I think I love this comment.

We should start a "Society for the Appreciation of Snowflakes" with the goal of ending the use of "snowflake" as a derogatory term. You can be president.



The only reason I can think is if you have confidence you can spam common review sites with "fake reviews" but even this is unlikely to work 100%.



Be TPU.

People say you bias towards nvidia. People say you biased towards AMD/ATI. Be bias towards nobody.

Cancel out.

PS: I become 4chan.

Kill me.
That Chan speak was pretty good man. If you were not so Anti-Trump I would think you were a KEK and not a normie lol.
Posted on Reply
#7
R-T-B
ajc9988 said:
My point was more to discuss the fears related to the document and certain coverage, such as the GPP and Nvidia basically gobbling up partner's brand IP and trademarks and that coverage like that can now be pre-empted by giving confidential information to all outlets at the outset. After that, hearing rumors on the effect of the agreement is pre-empted in the NDA.
That's not really true. It just means we know what to look for if they "pre-empt it" and then we have to find another source for the same info.

Seriously, that benefits nothing.

PS: I need to stop using the "we" here. I am speaking theoretically and I'm sure it pisses some people off. I don't work here anymore, to be clear.

TheMailMan78 said:
That Chan speak was pretty good man. If you were not so Anti-Trump I would think you were a KEK and not a normie lol.
Do not confuse me for anything normal. Frog god hates normal. I love frog god. I have to or he'll make drown in tadpole.
Posted on Reply
#8
Casecutter
I have a scenario... let's say a company says this card has 4Gb of memory, but while testing it doesn't access the last .5 Gb at the same speed as the first meaning the VRAM allocation and VRAM bandwidth is incorrectly stated by the release documentation provided.

A) Release my findings in the Review after the embargo.
B) Submit to the company my finding and wait till they respond.
C) Submit to the company the findings and just release without my finding after the embargo is lifted.
D) Submit to the company my finding and say you have up and until the embargo is lifted to respond.

I would say the biggest issue was the 5 year term and the amount of time that is provided to research or get counsel as to my possible risk. What are the terms if I later say I no longer want the be subjugated or participate in this what does it actual take in signatures, disclosures and time to not be under the stipulations of the NDA.
Posted on Reply
#9
ajc9988
R-T-B said:
That's not really true. It just means we know what to look for if they "pre-empt it" and then we have to find another source for the same info.

Seriously, that benefits nothing.

PS: I need to stop using the "we" here. I am speaking theoretically and I'm sure it pisses some people off. I don't work here anymore, to be clear.
You do understand what the clause saying "rightfully received" can mean in a court of law, correct? Such as someone else violating their NDA making the receipt wrongful potentially. That is in the mislabeled second 3, subsection b.
Posted on Reply
#10
R-T-B
ajc9988 said:
You do understand what the clause saying "rightfully received" can mean in a court of law, correct?
I'll confess that's actually not a phrase I'm familiar with.

And again, I'll fully admit though I do have some legal knowledge, I am an internet lawyer (AKA not a real one, so beware).

Anyone else want to chime in on that point? I don't see a way they could prevent the publication of something you learned entirely seperate from NDA sources, even if they clued you into the issue, but I could be wrong.

Captain_Tom said:
But drop this utter BS about Nvidia not having a motive to force stuff down reviewers throats.
I don't think anyone questions whether they have motive to do so. Every company has motive to do so. The question is whether it's even practical for them to really try and I'd argue it isn't practical for them to try to alter reviews directly.
Posted on Reply
#11
sneekypeet
Loads of reports coming in from this thread. Some posts have been "adjusted" and we ask that anyone who wants to attack, flame, enrage, poke fun of, or harass other members, move along and do not post. Anything that fits the above descriptions that follow this public warning will be removed and potentially points given depending on severity of the offenses.
Posted on Reply
#12
close
cadaveca said:
NVidia could not prevent such info from coming out by NDA, because most NDA expire when the product launches.
The NDA doesn't necessarily tie confidential info to a product launch. Confidential info is defined as "any and all technical and nontechnical info disclosed or made available to you by Nvidia". And it can be related to anything, including the terms of the agreement. As such the NDA expires 5 years after you receive the info from Nvidia, whatever it may be.

The obligation of confidentiality removes liability only if the info was "rightfully received without restriction of NDA". So I would say you will not be able to publish ANYTHING that comes from unofficial sources if you were also told by Nvidia under NDA. Unless they become "public domain" first, and someone else broke the news.

Also you are no longer allowed to post any editorials/opinions regarding Nvidia since anything you say would be based at least partly on confidential information received from them. So an article like "Why 2020 will be a hard year for Nvidia" or "What to expect from the GTX 2080" will be impossible without basing at least one statement on details mentioned under NDA by Nvidia. And I'm sure they have ways of sneaking everything you're told under this NDA.

Moreover, after signing this NDA you will not even be allowed to take part in discussions such as this one on the forum unless everything you say is unrelated to info you received from Nvidia.

So do you think you will be able to post an official answer to my original question after maybe consulting TPU's legal counsel?
Posted on Reply
#13
cadaveca
My name is Dave
close said:
So do you think you will be able to post an official answer to my original question after maybe consulting TPU's legal counsel?
No need for that... I didn't sign any NDA from NVidia. You can't hold me to something I personally didn't sign. you might want to consult a lawyer on your own to understand how it is possible.
Posted on Reply
#14
mat9v
I don't care much about this "for the benefit of Nvidia" stuff, BUT since it is a blanket NDA stating that any Confidential Information provided to Recipient are under it for aperiod of 5 years or indefinitely is classed as trade secrets, it MAY be used as a gag for any journalist that try to write about any information it aquired, be it positive or negative for Nvidia.
Let's say for example that Nvidia provides a journalist with information that some announced product will have 8 GB of memory but due to construction limitations it will have 6GB of fast memory and 2GB of slower memory - in that case said journalist will be unable to mention such a nugget of information regardless of the time of product release - it is a secret and if Nvidia does not specifically release such an information to the public, reviewer is unable to mention it in any way.
Sure, such a situation is not very likely, but blanket NDA without any expiration date is hard to explain. Just my 2 cents.
Posted on Reply
#15
close
cadaveca said:
No need for that... I didn't sign any NDA from NVidia. You can't hold me to something I personally didn't sign. you might want to consult a lawyer on your own to understand how it is possible.
I understand that but my question didn't point at you as a person but at TPU. The point is "any and all information" received from Nvidia by someone who signed the NDA can no longer leave that person's head without prior authorization from Nvidia for 5 years. Not legally at least.

As journalists that should scare you. Yet TPU's official position as indicated by this article is that "all's good". Even though you may be forced to sit on information that could be of public interest and not be able to report it in any way.

I find it fair to be told if this is indeed the official position of TPU. While I don't really doubt the integrity of people writing for TPU, after signing an NDA all bets are off and legality will probably precede integrity. Want it or not your hands are tied after signing. And this has to have a bearing on my views related to future Nvidia related articles.
Posted on Reply
#16
mat9v
cadaveca said:
No need for that... I didn't sign any NDA from NVidia. You can't hold me to something I personally didn't sign. you might want to consult a lawyer on your own to understand how it is possible.
If you are a staff member of TPU and your superior (chief editor, owner of TPU) signed NDA you are forced by terms of such NDA just as he is. If that did not work like that every worker in any company would have to personally sign every NDA that is presented to his/her company.
Posted on Reply
#17
cadaveca
My name is Dave
close said:
As journalists that should scare you. Yet TPU's official position as indicated by this article is that "all's good". Even though you may be forced to sit on information that could be of public interest and not be able to report it in any way.
I am not aware of any reviewer at ANY site that actually holds journalistic credentials. So please do not give us that label. you might want to question anyone that suggests they are a "tech journalist". I do know such people, good friends with one local, in fact. It's be pretty low for me to discount the time he invested in that career.

mat9v said:
If you are a staff member of TPU and your superior (chief editor, owner of TPU) signed NDA you are forced by terms of such NDA just as he is. If that did not work like that every worker in any company would have to personally sign every NDA that is presented to his/her company.
I am NOT employed by TPU.
Posted on Reply
#18
ajc9988
mat9v said:
I don't care much about this "for the benefit of Nvidia" stuff, BUT since it is a blanket NDA stating that any Confidential Information provided to Recipient are under it for aperiod of 5 years or indefinitely is classed as trade secrets, it MAY be used as a gag for any journalist that try to write about any information it aquired, be it positive or negative for Nvidia.
Let's say for example that Nvidia provides a journalist with information that some announced product will have 8 GB of memory but due to construction limitations it will have 6GB of fast memory and 2GB of slower memory - in that case said journalist will be unable to mention such a nugget of information regardless of the time of product release - it is a secret and if Nvidia does not specifically release such an information to the public, reviewer is unable to mention it in any way.
Sure, such a situation is not very likely, but blanket NDA without any expiration date is hard to explain. Just my 2 cents.
The reason I do not like that example is because testing showed that the last half GB of mem on the 970 was slow. The story was broken through forensics. That is not forbidden under this agreement, but if disclosed to the journalists, they have to wait for independent discovery/verification, which means the sales at release are already done. Nvidia already got their money. We can look at the results of the class actions lawsuit to see whether Nvidia was actually harmed or benefited (that is rhetorical, they made more money than loss on the 970).

The concern is gagging the CEO's statement that no new cards will be coming for a long time at a press engagement without express permission from Nvidia to report such with permission being in writing, or the GPP, which was discovered due to a violation of another person's NDA, which actually was leaked to AMD which then farmed out the story, which AMD in that situation may have partially sterilized it to allow for reporting, but getting other sources to speak off the record about confidential information, with Nvidia having already given confidential information to the journalist, to fill out the story on the GPP to get it to the public, that would be more difficult when complying with this agreement, potentially. That is why I used the GPP as an example.
Posted on Reply
#19
jabbadap
close said:
Imagine we are in September 2014 when GTX 970 was launched. You find out from a leaker that Nvidia knows about the 3.5GB RAM issue and you plan to publish an article. But Nvidia preempts your move by officially disclosing this info under NDA (perhaps attached to a sob story of how they're working to fix it to make it look less "covering tracks" and more "working for you").

Are you still allowed to publish this?

Because if not I can see why people are worried. You could become Nvidia's secret keeper for 5 years. And by the time the NDA expires people don't really care much about old news, do they? Not the same impact, definitely not on sales.
They really can't hide that kind of flaw. If tests shows that there's slower part of memory, it's easy to prove with showing test data that the flaw exists and you don't have to use nvidia given "Confidential information" as source. It's even actually said on that nda in chapter 3. Recipient shall not be liable for disclosure of Confidential Information that... c) is or was independently developed by employees of recipients.
Posted on Reply
#20
cadaveca
My name is Dave
jabbadap said:
They really can't hide that kind of flaw. If tests shows that there's slower part of memory, it's easy to prove with showing test data that the flaw exists and you don't have to use nvidia given "Confidential information" as source. It's even actually said on that nda in chapter 3. Recipient shall not be liable for disclosure of Confidential Information that... c) is or was independently developed by employees of recipients.
Thanks! You've hit the nail on the head here.

If there is something that might affect the end user, it is testable, and obvious, and as such, as soon as the card is released, is not subject to NDA clauses. I can run my own tests, since I didn't sign that NDA, and post my results, with no recourse for NVidia against TPU, by the NDA itself. I love how people will comment about stuff , but clearly didn't read what they are commenting on. Clearly, you read it.
Posted on Reply
#21
mat9v
cadaveca said:
I am not aware of any reviewer at ANY site that actually holds journalistic credentials. So please do not give us that label. you might want to question anyone that suggests they are a "tech journalist". I do know such people, good friends with one local, in fact. It's be pretty low for me to discount the time he invested in that career.


I am NOT employed by TPU.
I'm sorry, I just read "Staff member" - my bad :)
Posted on Reply
#22
cadaveca
My name is Dave
mat9v said:
I'm sorry, I just read "Staff member" - my bad :)
Yeah, that can give the wrong impression for sure, but that is something new under my name, but for many years was not present. I am a self-employed individual. If you do business stuff, how that all works should be pretty obvious.

Anyway, as posted above, even if I was, it wouldn't matter. All that matters is that I do my own thing, independent from what the NDA signatory does, without info provided by said party.
Posted on Reply
#23
close
cadaveca said:
I am not aware of any reviewer at ANY site that actually holds journalistic credentials. So please do not give us that label. you might want to question anyone that suggests they are a "tech journalist". I do know such people, good friends with one local, in fact. It's be pretty low for me to discount the time he invested in that career.
I used the "popular" definition as "a person who writes for newspapers, magazines, or news websites or prepares news to be broadcast". You may not have the credentials officially but it makes little difference to the topic at hand. You [or whoever that may be at TPU] are reporting on tech news and your readers may rely on the accuracy of your reporting. It's reasonable to expect that TPU would flag a conflict of interest such as not being able to report accurately due to signing that NDA.

So then take my question as "to whom it may concern". I initially addressed it to @btarunr since his title is "Editor".

Guess my point is since I can expect TPU reviewers to sign the NDA, I should also expect any and all Nvidia related articles to be censored in one way or another in order to eliminate any possible info declared confidential because it does not benefit Nvidia.
Posted on Reply
#24
Casecutter
cadaveca said:
NVidia could not prevent such info from coming out by NDA, because most NDA expire when the product launches.
I guess that what this is about are sites and their reviews subjugate by this 5 year umbrella NDA and then a NDA for the product and specifications. Or are you beholden by this "confidential information must be used solely for the benefit of NVIDIA". I suppose if a reviewer finds a discrepancy during a test that is not covered by confidential information they are good to report without any apprehension of running afoul of this 5 year subjugation.
Posted on Reply
#25
cadaveca
My name is Dave
close said:
Guess my point is since I can expect TPU reviewers to sign the NDA, I should also expect any and all Nvidia related articles to be censored in one way or another in order to eliminate any possible info declared confidential because it does not benefit Nvidia.
The NDA only hold the signatory individual as responsible. Just read it... and the only person who would sign an NDA is the person getting the hardware/information. The rest of our organization is not affected as long as there is no discussion between us about it. And that rarely happens. I don't post my own NDA-covered reviews usually until just moments before it should be published, just so that there is no problem with such things here. It's very easy to overcome situations like this.

Casecutter said:
I guess that what this is about are sites and their reviews subjugate by this 5 year umbrella NDA and then a NDA for the product and specifications. Or are you beholden by this "confidential information must be used solely for the benefit of NVIDIA". I suppose if a reviewer finds a discrepancy during a test that is not covered by confidential information they are good to report without any apprehension of running afoul of this 5 year subjugation.
Even if it is covered by NDA, there are ways around that. NDAs of this sort in this industry really do nothing but provide necessary protections for the stock market. They have a very specific purpose. I'm also not in the US, so not subject to US law directly. Of course, I mean myself, not TPU.

I'll give you an example:

So, I post the launch day of the next VGA. I guess this date, based on the past (ie, my birthday, which is when they usually launch cards), and I get this date right. Maybe this date is covered by NDA< but since it was not disclosed to me directly, should I decide to post it on the front page, they can't be held responsible for what I do, because IT SPECIFICALLY SAYS THAT in the NDA.

This ensures that the worries you have mentioned here are not a real worry... it just means that whoever signed the NDA will not be the person to explore such subjects. Fortunately, there are many people that contribute here.
Posted on Reply
Add your own comment