Texas Instruments sent a DMCA takedown to a site archiving data sheets
twitter.com
twitter.com
p = B709D3A0CD2FEC08EAFCCF540D8A100BB38E5E091D646ADB7B14D021096FFCD
q = B7207BD184E0B5A0B89832AA68849B29EDFB03FBA2E8917B176504F08A96246CB
d = 4D0534BA8BB2BFA0740BFB6562E843C7EC7A58AE351CE11D43438CA239DD9927 6CD125FEBAEE5D2696579FA3A3958FF4FC54C685EAA91723BC8888F292947BA1
which can be used to sign firmware for the TI-83+. This lead to the creation of the Wikipedia "black lock" on articles, meaning they can't be edited because of internal Wikipedia corporate policy, which didn't previously exist.
Wikipedia eventually won the dispute, and they host the numbers here: https://en.wikipedia.org/wiki/Texas_Instruments_signing_key_...
In this context I can see that being another pressure which assures that large models are only able to be built by FAANG style companies.
If you don't put copyrighted works into the NN but then feed a copyrighted work to it as an input the result is still copyrighted.
I really don't find this that complicated. NN's having a license that is a composite of the software license and the licenses of the training data feels super natural.
See: What color are your bits? https://ansuz.sooke.bc.ca/entry/23
This seems reasonable at first, but not the more I think about it. Copyright is not like an AGPL software license that claims/poisons everything it touches.
The NN is not an artistic work and probably should not be considered a derivative work any more than your brain or memory of a copyrighted work should be considered a derivative work. Likewise, the outputted work of a NN is not automatically a derivative work any more than the work of an artist that has been inspired by other artists' copyrighted content is. That's just not how copyright works.
Obviously, it is, since the only mechanism copyleft licenses have to claim/poison things the licensed material “touches” is the copyright doctrine of derivative works, and the licensable exclusive rights the original work’s copyright owner has to create derivative works.
This is why one company I worked for refused to let us use Webpack 1.x- a dependency of a dependency was an ancient npm module from the time that adding licenses to modules wasn't very common. It wasnt until the 2.x line came out that they finally updated to a version that updated to a version that didn't have that problem.
Yet my comparison of the human brain still stands, I believe.
Just because you learned about something in art class from looking at a copyrighted work doesn’t make everything you produce after that a derivative work.
(For examples of what I mean, see any copyright dispute involving YouTube, Dropbox, etc. or the repercussions of illegal content on pornhub vs. Facebook.)
If any data is ever illegal then by association certain numbers are also illegal.
EDIT: The Wikipedia article points this out.
8.3.4. "How will privacy and anonymity be attacked?"
[...]
like so many other "computer hacker" items, as a tool for the "Four Horsemen": drug-dealers, money-launderers, terrorists, and pedophiles.
https://en.wikipedia.org/wiki/Four_Horsemen_of_the_Infocalyp...
The legal system is not like a computer system and tech people who think they can fool it with simple tricks like it was a computer system are in for an unpleasant surprise.
It is clear that TI no longer has an interest in developing new products, or figuring out better ways of making existing products, and would instead like to rent-seek and make the world a smaller place.
TI is now just rentseeking on their ananlog dominance and whatever sinking boat Intel is, TI is way worse.
I don’t think “rentseeking on their analog dominance” is a fair characterization.
TI is more than happy to work with you on their most cutting edge designs if you're shipping in appreciable volume. Talking a quarter million chips or more per year. That's just the way the wind blows in their business. Gotta recoup as much R&D expenditure as possible up front. I've worked with them on these sorts of designs. They employ tons of really sharp engineers and still make lots of high quality, very useful chips. You just have to buy enough of them for it to be worth it for TI to work with you.
No OSHW application I know of has ever moved enough units to cross this bar. Not even close.
They get way more ROI by heavily supporting bigCos who will give them millions of dollars of business YOY rather than helping a zillion mom and pop shops who can barely wire a resistor the right way.
Think of it as a tacit strategy of "firing your worst customers". Having worked in bigCo vs small Co, there are a lot of small Co and OSHW EEs who barely have a clue what they're doing. It's not worth TI's time or money to help these people out. Upleveling lousy engineers thru support only to net them a 10 piece part order, is a really lousy ROI. (Unless of course we're talking an ASIC or some other astronomically high ROI custom piece. But, I'm assuming that we're not, because I don't work for a military prime contractor.)
What I don't understand about the hardware industry is the extent to which they actively restrict access to information on how their product works unless you're buying in volume. Why is it necessary to have layers of NDAs on a product datasheet? Many of the people who might write software to integrate with some hardware are not the people manufacturing high-volume products based on it! (See: Linux driver developers.)
I also understand. But in some ways I disagree: set a higher price instead. Let me, the customer, decide if the price is too high.
That makes a lot of sense - if you're building a software platform.
If you're building a chip, it makes sense to be the first mover on the cutting edge. By obscuring specs, your competitors won't know what specs are the most valuable to your customers. That sort of thing is critically important to semiconductor business.
Because obscuring the use case and the customer of a cutting edge chip is a semiconductor company's most reasonable moat these days. If you telegraph what big customers are buying by freely sharing your most cutting edge product portfolio, someone else will copy it. (Probably someone in China with a hefty dose of subsidy from the gov't.) Classic case of "Your margin is my opportunity."
ADI or On Semi can, and will, catch TI after 18-36 months of development effort. The longer TI can keep them off their tail by making it hard to see what customers are buying, the more profit they can make without a reasonable competitor on the market.
Software is a commodity that TI distributes as a compliment to the thing that actually makes them money, which is selling chips. For analog circuits, this is completely irrelevant, and doesn't affect their operating procedure. For processors and peripherals that require drivers, it's a cost of doing business for them - not an enabling feature.
It's a cost of doing business that would be ameliorated by telling people how to use the product, because then they're not the only ones able to do the work. I'll use Linux drivers again as an example: I can only imagine the money it must cost for these SoC manufacturers to keep their kernel forks with binary blob drivers up to date with mainline Linux, and the security vulnerabilities that go unpatched in their products when they don't. That's money that Intel and AMD just don't have to spend; they contribute to the kernel, yes, but they're far from the only ones doing the work.
...if you're talking about processors, sure. But only talking about processors gives a really narrow view of TI's business, and doesn't add any context into why they sell the way they do. They make a ton of money on analog applications that require no OSSW support at all. They know how to sell chips that way - they've been doing it for fifty plus years. They are an analog company that didn't get into the processor business until comparatively recently. I'm sure a bunch of this cloak and dagger around datasheets is legacy holdover from an analog sales world. They know what they are doing in that department, without doubt.
As to why they're DMCA'ing someone for a 50 year old datasheet - I'd guess it's as simple an explanation as someone in Legal carrying out their boss's mandate. Nothing more, nothing less.
Alternatively, some customer pays an absurd amount of money to add Function X to the chip, and pays a premium to be the only customer who can use Function X.
Burr Brown opamps and DSPs are enough to keep the legacy of TI alive. OMAPs are just DSPs with processor cores. They were not designed to be fast first.
Since then, some mfgs. have been restricting datasheets to avoid potential patent litigation.
This might be the case here. Possibly another example of death by a thousand cuts as our society gets more regimented.
If only. In the past it was common to provide schematics of your hardware in order that people could repair it. This was common for consumer products, industrial, instrumentation, test gear, almost everything. For example, the manual for the Amiga 500 had the schematics in it.
While I was at University I made some money by repairing DJ-hardware. Controllers, CD-Players and Turntables. There things are often expensive, but they are quite easy to repair because there are detailed service manuals available. Not on the official sites, though.
Example: in the past my friend bought Moog Prodigy synthesizer and along with everything it also came with all the electronic circuitry diagrams. Good luck finding anything like this with the modern hardware.
You wish they were stuck in the past. People like Steve Wozniak often designed new stuff only based on the manuals that back then had complete schematics in them. They didn’t have enough money first to buy anything but that information later on resulted in sales.
The electronics and computers industries have been built initially based on free detailed documentation from all the manufacturers of components.
When I was a student in electronics, I learned more from datasheets and application notes than from most university manuals.
All companies continued to provide good technical documentation until close to the year 2000.
With the growth of the Internet I believed that it will become easier than ever to get technical documentation, as it could be now downloaded instead of having to get printed copies.
Unfortunately, at the end of the nineties a lot of negative tendencies have appeared. Many large electronics companies with decades of successful histories were split (e.g. Motorola, Siemens, Philips and others) and their successors seem to have lost most of their previous experience.
A large number of other electronics companies have started a descending evolution and, sooner or later, between 2000 and 2020, they were bought by competitors, so now only a handful of US non-fabless manufacturers of semiconductor devices have survived.
At the same time the incomprehensible fashion of requiring Non-Disclosure Agreements for getting the complete datasheets or even any datasheets has become more and more widespread.
This policy of the NDA has been conceived by morons who have not the slightest idea about how an electronics product is designed.
Whenever I design a new product, I need first to be able to read all the datasheets of all the products that might have even an extremely remote chance to be useful, to decide and select what I could use.
I will not bother to sign an NDA to learn about products that might be not useful at all for me.
The only positive effect that the NDA may have for a producer is that it might prevent the already existing customers to redesign their products, because they might not bother to obtain similar NDA from other vendors. On the other hand requesting NDA's will deter many potential new customers, especially due to the insistence on providing an estimated sales volume, before signing the NDA.
That is stupid, because before seeing the information under NDA I have no idea if I would ever want to buy 1 sample, much less 1 million pieces per year.
Managers seem much happier about NDA's than designers, because they frequently feel like they are some sort of special bond with the vendor and the information under NDA provides some sort of competitive advantage, but that is extremely far from the truth.
In most cases the only valuable information contained in the datasheets under NDA was about horrible bugs that required complex workarounds. If I had known about the bugs before the NDA, I might have never chosen those components, so maybe that was a desirable feature of the NDA for the vendor.
The end result is that when I was young and I designed anything, I could choose any component from a large number and I could easily select the best for my specific needs. Moreover, in many cases I could find a method to use in a novel way something that was not intended for my particular application, but due to the good documentation I could understand whether it would be good for different uses.
Now there is frequently just one remaining producer for any component, or if there are more of them, you might need to choose one by lottery, because without the information under NDA you cannot know who has the product more suited for you, and you can sign the NDA only if you are already committed to buy that product and not another.
That's a positive from their competitors requiring an NDA, not for they requiring one.
Anyway, large software sellers benefit a lot from stupid procedures from their clients that create a huge cost to start buying from a new supplier (like passing all suppliers through legal or having a management vote for them). If all hardware manufacturers require NDAs, they can force their buyers into creating similar procedures, closing the market for new entrants.
The legal departments within companies have very, very different measures of success than that of just about anyone else. If you have weak company leadership, legal will start to drive policy more than it should.
That's a really interesting dynamic I have never experienced or heard of. Explains a lot! Do you know other examples?
(ed. clarify which part I wanted examples for)
It's a very different game than Intel's business model, which depended on a pure performance play and being 18-36 months ahead of all competitors tech wise. TI's strategy requires way more customer contact and development, because analog applications are so much more specific.
This is exactly why: to stop small developers from buying them. There are three types of components in the electronics world to my observation.
1. Components for the general public (e.g. opamps, microcontrollers, power supply controllers) - the datasheets are publicly available. This is probably 70% of the marketplace.
2. Components for the OEM - Some components are only meant to be sold to OEMs in huge quantity. For example, a company may sell $1 LCD controllers, if such a company is only interested in selling to huge manufacturers for big money, they often classify their datasheets as "confidential" - it's a joke, the only purpose is stop small developers from wasting their time. Whether it's a Type-2 part heavily depends on the company. For example, a big U.S. semiconductor company may sell it to the general public, but a cheap Taiwanese vendor often classifies all the information as confidential.
3. Security through obscurity. If the chip contains unique or advanced technology, for example, the latest generation of NAND flash or SoC, the datasheet is classified to make it more difficult for other people to do market research. Similarly, the payment card and consumer DRM vendors essentially own the security chip industry, selling channels of the chip and availability of the datasheets are tightly controlled.
Type-2 and Type-3 are responsible for the most frustrating experience. If you are an independent kernel hacker who wants to port Linux to a new gadget, not being able to get the datasheet of the $1 LCD controller because it's "confidential" is a real headache.
But as I mentioned previously, Type-2 chips are not really secret - the confidentiality is only meant to stop small developers from bothering them. Thus, the chips themselves can usually be found for sale on the Shenzhen electronics marketplace, and it's often possible to find numerous leaked datasheets on the web. You can also reverse engineer consumer gadgets as a reference design. For example, Realtek Ethernet controllers, 100% of the public datasheets are leaked by insiders. If it's not, sometimes getting the datasheet may be possible you do some social engineering, "we are prototyping a new product based on the LCD controller chip, now our problem is register xxh..."
On the other hand, if it's Type-3, don't even think about it.
The vast majority of products by TI are Type-I: Usually all technical documentation is provided on the website, with schematics, PCB layouts, and reference designs (unless it's a specialized chip for a particular application). I recently used a TI chip in my design just because of the availability of documentation.
> Here is TI DMCAing a bunch of random datasheets. Completely jellybean stuff like 555s, 74xx logic, and op-amps. Nope nope nope nope. This is crazy.
These chips don't even belong to the Type-2 category. It's unreasonable to send DMCA takedown notices in terms of security-through-obscurity. The only explanation I can see is TI's overzealous corporate policies.
TI is great to work with on Type 1 chips. They're nowhere near the worst in the Type 3 category. (Looking at you, Qualcomm and Marvell.)
I totally agree that the DMCA notices issued don't make much sense, given this context. I just get irritated by OSHW folks acting like their business means something to TI. They are ants crying in the footprints of elephants.
posting the link here when its done, then someone else can submit this to libgen and other places.
zip file (8gb) here: https://mirrors.deadops.de/ti.zip
They have a few global distributors remaining, and if you talk to any of them for a new design, you'll notice they try to avoid promoting TI parts because of TI's new pricing and commission policies.
Whats up at TI? They used to be great for datasheets and Appnotes.
I'm not sure that would explain them attempting to limit the availability of their datasheets though. That doesn't make much sense to anyone.
For example, DMCA notice: https://www.lumendatabase.org/notices/18049758?access_token=...
First link from "ALLEGEDLY INFRINGING URLS": https://www.distrelec.de/en/multivibrator-monostable-pdip-te...
They are pretty much all the same - product links. Did google simply obey this and delist the URLs? How is this even legal.
TI recently dropped some distributers - maybe they are doing this as a way of removing 3rd party retailers under guise of copyright violations - DMCA abuse, pure and simple. Should be HEAVY penalties for this kind of crap.
http://digital-law-online.info/lpdi1.0/treatise15.html
The courts could (and should) render TI’s underlying copyrights unenforceable.
Sadly, the cited precedent is from 1990, and I don’t know of any more recent cases.
edit: I also just posted this as a top-level article. People in the tech community should have a broader understanding of this line of legal defense. https://news.ycombinator.com/item?id=25686276
The only explanation I can come up with is that TI wants a closer relationship with their customers. They don't want to be a commodity chip producer.
When distributors host the datasheets, then customers would probably never visit TIs website, and just compare chips from multiple vendors on the distributors website. Customers are just going to pick the cheapest part that matches the specs.
Maybe TI wants to be seen as a premium vendor. They don't want customers to look for a LED driver on an electronics distributors website, they want you to look for a TI LED driver on ti.com. They can explain the benefits of their chips much better on their own website.
As someone who occasionally tinkers with electronics, I think this decision by TI sucks. I like that Mouser and Reichelt host all their datasheets. But I guess TI doesn't care if I'm going to buy two LM3886s or not.
When I hear these kind of stories I always wonder this too. Like, who was allowed to turn this into a company policy and why on earth did they think it worth their time? Usually something that would seem to harm the public image and goodwill of the firm.
This speculation in this comment seems pretty convincing imo:
https://news.ycombinator.com/item?id=25683856
(mainly the part about legal getting into the driver seat in a semi-fossilized sort of company like TI).
I suppose a simple URL would be OK though!
The main reason I put it on my site was because they made the document hard to get, otherwise I would have just put up a link to it.
It was clearly a Bot. A human would have seen it as a promotion for a line of obscure TI products.
You'd think that for all the cases in which data sheets get taken down. Unfortunately TI doesn't seem to.
and a zip file (8gb) here to download in bulk (or for others to make their own mirrors): https://mirrors.deadops.de/ti.zip
A better remedy would be to publicly shame them about what they just did. TI will still carry on with their business but others will just know better than to engage with TI in the future!
The only way to reach most companies is to turn up on their PR radar. If something is seen as potentially damaging to their brand your concern will enter the company at a place in the organization which will have a far more direct channel to someone who can make decisions.
If you go through their investor feedback channel, unless you own a fair chunk of shares in the company or represent major shareholders, they will ignore you. Of course, if you are a major investor you wouldn't contact them through that channel anyway. You would already have a relationship to someone in, or close to, top management who is responsible for handling major stakeholders.
This archive is distributing out of date documentation without making it clear that it’s out of date. This could cause a major problem for someone that uses it.
The documentation for the same chip is perfectly available on TIs website, although to get older datasheets you’ll need to dig further (by design).
There’s a ton of sites mirroring datasheets, and they weren’t DMCA. It looks like someone at TIs legal got creative about how control this sites to update their info.
Once an IC is designed, the datasheet doesn't substantially change. It isn't a computer program that gets continuous updates every couple of months. They might add some errata for some corner cases or clarification. What happens when the IC is discontinued? I still need the datasheet, so I can understand the circuit when finding a replacement IC.
It’s cases like the one you mention when you want to know that the new parts have different parameters and you might need to change other components to compensate.
Copyright must be reworked,and reduced. back to AT MOST the original 14 years plus 1 extension of 14 year to the original human creator (not a company)
Failure to do this will continue to see humanity limited in our growth
Why would they bother doing this?
I just refuse to buy parts without datasheet except for parts that don't need explaining.