Elsevier Launching Rival To Wikipedia By Extracting Scientific Definitions Automatically From Authors' Texts

from the don't-do-as-we-do,-do-as-we-say dept

Elsevier is at it again. It has launched a new (free) service that is likely to undermine open access alternatives by providing Wikipedia-like definitions generated automatically from texts it publishes. As an article on the Times Higher Education site explains, the aim is to stop users of the publishing giant's ScienceDirect platform from leaving Elsevier's walled garden and visiting sites like Wikipedia in order to look up definitions of key terms:

Elsevier is hoping to keep researchers on its platform with the launch of a free layer of content called ScienceDirect Topics, offering an initial 80,000 pages of material relating to the life sciences, biomedical sciences and neuroscience. Each offers a quick definition of a key term or topic, details of related terms and relevant excerpts from Elsevier books.

Significantly, this content is not written to order but is extracted from Elsevier's books, in a process that Sumita Singh, managing director of Elsevier Reference Solutions, described as "completely automated, algorithmically generated and machine-learning based".

It's typical of Elsevier's unbridled ambition that instead of supporting a digital commons like Wikipedia, it wants to compete with it by creating its own redundant versions of the same information, which are proprietary. Even worse, it is drawing that information from books written by academics who have given Elsevier a license -- perhaps unwittingly -- that allows it to do that. The fact that a commercial outfit mines what are often publicly-funded texts in this way is deeply hypocritical, since Elsevier's own policy on text and data mining forbids other companies from doing the same. It's another example of how Elsevier uses its near-monopolistic stranglehold over academic publishing for further competitive advantage. Maybe it's time anti-trust authorities around the world took a look at what is going on here.

Follow me @glynmoody on Twitter or identi.ca, and +glynmoody on Google+

Filed Under: definitions, knowledge, open access
Companies: elsevier

Scientific Publishers Want Upload Filter To Stop Academics Sharing Their Own Papers Without Permission

from the where-there's-a-gate,-there's-got-to-be-a-gatekeeper dept

Back in March of this year, Techdirt wrote about ResearchGate, a site that allows its members to upload and share academic papers. Although the site says it is the responsibility of the uploaders to make sure that they have the necessary rights to post and share material, it's clear that millions of articles on ResearchGate are unauthorized copies according to the restrictive agreements that publishers typically impose on their authors. As we wrote back then, it was interesting that academic publishers were fine with that, but not with Sci-Hub posting and sharing more or less the same number of unauthorized papers.

Somewhat belatedly, the International Association of Scientific Technical and Medical Publishers (STM) has now announced that it is not fine with authors sharing copies of their own papers on ResearchGate without asking permission. In a letter to the site from its lawyers (pdf), the STM is proposing what it calls "a sustainable way to grow and to continue the important role you play in the research ecosystem". Here's what it wants ResearchGate ("RG") to do:

RG's users could continue "claiming”, i.e. agreeing to make public or uploading documents in the way they may have become accustomed to with RG's site. An automated system, utilizing existing technologies and ready to be implemented by STM members, would indicate if the version of the article could be shared publicly or privately. If publicly, then the content could be posted widely. If privately, then the article would remain available only to the co-authors or other private research groups consistent with the STM Voluntary Principles. In addition, a message could be sent to the author showing how to obtain rights to post the article more widely. This system could be implemented within 30-60 days and could then handle this "processing" well within 24 hours.

In other words, an upload filter, of exactly the kind proposed by the European Commission in its new Copyright Directive. There appears to be a concerted push by the copyright industry to bring in upload filters where it can, either through legislation, as in the EU, or through "voluntary" agreements, as with ResearchGate. Although the lawyer's letter is couched in the politest terms, it leaves no doubt that if ResearchGate refuses to implement STM's helpful suggestion, things might become less pleasant. It concludes:

On behalf of STM, I urge you therefore to consider this proposal. If you fail to accede to this proposal by 22 September 2017, then STM will be leaving the path open for its individual members to follow up with you separately, whether individually or in groups sharing a similar interest and approach, as they may see fit.

What this latest move shows is that publishers aren't prepared to allow academics to share even their own papers without permission. It underlines that, along with fat profits, what the industry is most concerned about in this struggle is control. Academic publishers will graciously allow ResearchGate to exist, but only if they are recognized unequivocally as the gatekeeper.

Follow me @glynmoody on Twitter or identi.ca, and +glynmoody on Google+

Filed Under: academia, copyright, knowledge, papers, sharing
Companies: researchgate, sci-hub

More Thoughts On The Senate's SESTA Hearing

from the what-is-the-plan-here dept

So far this week, we've explained why SESTA is such a bad bill and how it will create a massive chilling effect that could impact nearly every online service. And while the Senate hearing on the bill wasn't as bad as we feared it would be, it certainly had its problematic moments -- such as when the bill's co-author, Senator Richard Blumenthal, argued that companies that couldn't afford to moderate/filter everything should be prosecuted. However, in the days since the hearing, I've had a few thoughts about aspects of the debate that are, to say the very least, troubling.

"Do Something Now!": While it was good to see many Senators at least pay lip service to the idea that the bill needed changes to not have massive unintended consequences for the entire internet, it was troubling the way they approached this process. Prior to the hearing, Emma Llanso wrote up a list of four important questions Senators should ask during the hearing -- and none of them were asked. The key questions were fairly fundamental ones: what actual gap is there in the law and will this fix it? It was disturbing that no one seemed to discuss that at all.

After all, under CDA 230 today, nothing stops federal law enforcement from going after Backpage (other than if the DOJ doesn't think Backpage broke the law). CDA 230 does not cover federal government law enforcement. Similarly, CDA 230 does not cover content "developed" by a company itself. So, if as many people claim, Backpage develops illegal content itself, it's still liable. Finally, just a few years ago, Congress passed the SAVE Act, with the exact same stated intent: to carve a hole in CDA 230 to go after Backpage by making it a federal crime to advertise sex trafficking. That law hasn't been used -- and none of the Senators seem to be asking why.

Instead, there's very much the traditional politician's syllogism of "we must do something, this is something, we will do it." A few times during the hearing, Senators demanded from the two opponents to SESTA that they provide better language if they're concerned about this language. Notice the problem here: they were admitting that this language is problematic, but seemed to have no interest in understanding why or how to fix it -- instead, demanding that others give them new language, with the implicit threat that if they don't, this language will stay because this is "something."

That may be all too common, but it seems like a dreadful way to make policy.

The knowledge standard is a mess: This is important, and got some discussion during the hearing, but not nearly enough. The "knowledge" standard in the bill is a complete and total mess. The supporters of the bill brushed it off as no big deal, often by misstating what the bill actually says. California Attorney General Xavier Becerra focused solely on the criminal standards for knowledge, saying he needed to prove "beyond a reasonable doubt" that there was intent, while NCMEC's Yiota Souras insisted the knowledge standard was very narrowly tailored.

Both of them are wrong -- in somewhat staggering and dangerous ways. Again, the actual text of the bill says the following:

The term ‘participation in a venture’ means knowing conduct by an individual or entity, by any means, that assists, supports, or facilitates a violation [of sex trafficking laws.]

This is problematic on multiple levels. First "knowing conduct" by itself is vague and not clearly defined. As Professor Eric Goldman suggested during the hearing, knowing conduct could just mean that a platform knows it's doing something, but has no idea of the outcome. For example, if we knowingly allow users to post comments on our site -- and someone uses that to post sex trafficking ads -- even though we didn't know they were posting sex trafficking ads, our conduct in enabling comments was "knowing." If that sounds like a crazy scenario to you, fair enough -- but shouldn't the law state that much more clearly? Make it clear that the knowledge is not just of its own conduct, but that the conduct is specifically targeted at breaking the law.

The second problem with the knowledge standard is the claim that you can violate the law if your "knowing conduct" "by any means"... "assists, supports or facilitates a violation" of sex trafficking laws. As we explained earlier, assists, supports or facilitates is super broad. It's possible to read this to mean that hosting a website where someone sets up a blog to advertise sex trafficking makes you automatically liable -- even if you knew nothing of the actual content. After all, you did "knowing conduct" (hosting a website) and that website was used to "assist, support or facilitate" a violation of sex trafficking laws. We would not read the law this way, but as the language is currently written, it could very well be read that way.

And that means it will be read that way by someone. We've seen tons of civil lawsuits filed against deep-pocketed companies (and not so deep-pocketed companies) on a much more flimsy basis. If you don't think a bunch of lawyers won't be searching for such cases to bring, you haven't paid much attention.

And these lawsuits can be very costly and time-consuming. Remember the Viacom v. YouTube case? That was an intermediary liability case that was almost entirely focused on the question of whether YouTube had the requisite knowledge to be liable (sound familiar?). And it went on for more than seven years before it was finally settled, rather than having a court issue a final ruling. With such a weak knowledge standard (even weaker than the DMCA's that was fought over in the Viacom case), you can bet there will be long and costly lawsuits over just what the hell Congress meant by "knowledge."

It's fairly stunning and concerning that those pushing for the bill insist there's no problem with the knowledge language. It's as if they have no idea of the fairly recent litigation history over this very issue.

A lack of enforcement isn't fixed by blaming intermediaries: As mentioned, nothing in CDA 230 stops the Justice Department from going after intermediaries if they break the law. When this was (barely) brought up during the hearing, one response was that the DOJ is either overwhelmed or isn't doing its job -- and by opening up prosecutions to state law enforcement, it would allow more of these kinds of cases to be brought. And while that may be true, that seems like a really weird way to solve the problem. If the true problem is a lack of willingness or resources by the DOJ to take on these cases, then why isn't the discussion and legislation directly targeting that problem? Why isn't there a discussion of allocating more resources, or finding out why the DOJ isn't bringing these cases, and offering legislation or resources to solve whatever may be blocking action.

Instead, it seems like a very strange approach to say the answer to a lack of will or resource is... to make more companies liable.

A profound misunderstanding of CDA 230. This one bothers me quite a bit. If we're talking about amending CDA 230, at least get the facts on CDA 230 right. Unfortunately, many Senators and some of the witnesses did not. There's this incorrect belief that CDA 230 leaves no recourse for victims of sex trafficking. That is not even remotely true. The actual perpetrators of sex trafficking are still very much in violation of the law. And, again, if platforms actually develop illegal content themselves, then CDA 230 protections don't apply.

Finally, so many of the Senators and commentators act as though because of CDA 230, no company does anything to moderate their platforms. This is laughable. Basic public pressure has lead most platforms to moderate their platforms thoroughly -- and that's one of the features of CDA 230, in that it says that any effort to moderate/filter your platform to remove content you don't want to see (even if legal) does not attach any liability. SESTA changes that standard (even as its authors claim it doesn't). Because of the "knowledge" standard, moderating content may now be seen as evidence of "knowing conduct." So this effectively wipes out one of the most important tools that platforms have used to stop their platforms from being used for sex trafficking -- and no one seems to want to acknowledge that.

Unless the supporters of this bill are willing to face up to these basic facts and problems with the bill, their headlong rush into "doing something" seems likely to cause a lot more harm than good.

Filed Under: cda 230, intermediary liability, knowledge, sesta

Is There A Single Online Service Not Put At Risk By SESTA?

from the the-risks-are-big dept

Earlier today, I wrote up a list of the many problems with SESTA and how it will be abused. Over and over again, we've seen defenders of the bill -- almost none of whom have much, if any, experience in managing services on the internet -- insist that the bill is "narrowly targeted" and wouldn't create any problems at all for smaller internet services. However, with the way the bill is worded, that seems unlikely. As stated in the last post, by opening up sites to facing both lawsuits from state Attorneys General and civil lawsuits, SESTA puts almost any site that offers services to the public at risk. The problematic language in the bill is that this is the "standard" for liability:

"The term 'participation in a venture' means knowing conduct by an individual or entity, by any means, that assists, supports, or facilitates a violation...."

And that could apply to just about anyone offering services online. So, let's dig into a few examples of companies and services potentially facing liability thanks to this nuclear bomb-sized hole in CDA 230.

• Airbnb: I did a whole post on this earlier. But there are multiple reports of how some people have used Airbnb for prostitution. I'm quite sure that Airbnb doesn't want its users to use the platform in this way, but under SESTA, it will face criminal and civil penalties -- and it has no way to prevent it. Prosecutors or civil litigants can easily point to these articles, and note that this demonstrates "knowledge" that Airbnb is "facilitating" sex trafficking, and, voila, no CDA 230 protections.
• Square: The popular payments processor that has made it easy for small businesses to accept credit cards... can also be used for prostitution/sex trafficking. Square can't specifically watch over what each of its customers is selling, but clearly, it wouldn't be difficult for prosecutors and/or civil litigants to argue that Square is knowingly facilitating trafficking by allowing traffickers to accept payments.
• Facebook: Obviously, tons of sex traffickers use Facebook to advertise their wares. There have been news stories about this. So clearly Facebook has "knowledge." If it can't magically eradicate it, it may also now have to deal with criminal and civil lawsuits lobbed its way.
• Snapchat: Last year there was a story of a teenager lured into a sex trafficking ring via Snapchat. So, clearly, lawyers might argue that Snapchat has "facilitated" sex trafficking.
• Wordpress/Automattic: Automattic hosts a significant portion of the entire internet with its powerful Wordpress.com platform. It's pretty damn likely that at least some sex traffickers use Wordpress to set up sites promoting what they're offering. Thus, it's "facilitating" sex trafficking.
• Wikipedia: You wouldn't think of Wikipedia as being a hub for sex trafficking, but the site gets hit with spam all the time, and people trying to promote stuff via its webpages -- and even when spam is edited out, it remains viewable in the history tabs. So, if a few links to trafficking advertisements show up, even if edited out, Wikipedia could face liability for facilitating such ads.
• Google Docs: Did sex traffickers use Google docs to create fliers or manage a spreadsheet? Is that "facilitating" sex trafficking? Well, we might not know until after a court goes through a long and involved process to figure it out.
• Cloudflare: Tons of websites use Cloudflare as a CDN to provide better uptime. What if one of them involves advertisements for sex trafficking. Is Cloudflare liable? After all, its CDN and anti-DDoS technology helped "facilitate" the service...
• Rackspace: Popular hosting company has millions of websites. If one of them hosts advertising for trafficking, it can be liable for facilitating sex trafficking.
• Amazon: Through its web services arm, Amazon hosts a large portion of the internet. Can you say for certain that Amazon S3 isn't used somewhere by some sex trafficking parties?
• YouTube: These days, almost anything can be found in videos, and while YouTube has a system to "notify" the company of abuse, that may be used against the company, claiming "knowledge." After all, in a case in Italy where Google execs were found criminally liable, the fact that some users clicked the "notify" button was used as evidence against the execs.
• Namecheap: Has Namecheap ever registered a domain that was then used by sex traffickers? Well, then it can be argued that it facilitated sex trafficking, and is not protected by CDA 230.
• Indeed: Basically any "job board" is at risk. There are stories of sex traffickers seeking victims via promises of summer jobs -- and while these stories are highly dubious at best, similar accusations against online job boards under SESTA would be easy to make.
• Reddit: As an open forum, certainly sex trafficking gets discussed in some corners of the site. Is it possible that some have used it to help facilitate trafficking? Absolutely.
• Any site that has comments: And that includes sites like Techdirt. Want to get a site in trouble? Why not spam its comments with links to sex trafficking ads? Voila, you've now put those sites at risk of criminal and civil liability...

The point of this is that this list can go on and on and on. Almost any internet service can be used in some way to "facilitate" sex trafficking. And rather than recognizing that the problem is those engaging in sex trafficking, SESTA now lets everyone go after the tools they use. But nearly all those tools are mostly used for perfectly legitimate, non-illegal activity. Yet, under SESTA, all face massive liability and the potential for criminal charges.

Filed Under: cda 230, knowledge, section 230, sesta

Why SESTA Is Such A Bad Bill

from the so-much-damage dept

We've been talking quite a bit about SESTA -- the Stop Enabling Sex Traffickers Act -- and why it's so problematic, but with hearings today, I wanted to dig in a bit more closely with the text to explain why it's so problematic. There are a large number of problems with the bill, so let's discuss them one by one.

Undermines the incentives to moderate content and to work with law enforcement:

This remains the biggest issue for me: the fact that the bill is clearly counterproductive to its own stated goals. When people talk about CDA 230, they often (mistakenly) only talk about CDA 230(c)(1) -- which is the part that says sites are immune from liability. This leads many people to (again, mistakenly) claim that the only thing CDA 230 is good for is absolving platforms from doing any moderation at all. But this actually ignores the equally important part of the same section: CDA 230(c)(2) which explicitly encourages platforms to moderate "objectionable" content, by noting that good faith efforts to moderate and police that content have no impact on your protection from liability in part (1).

In other words: as currently stated, CDA 230 says that you're encouraged to moderate your platform and takedown bad content, because there's no increase in legal liability if you do so. Indeed, it's difficult to find a single internet platform that does zero moderation. Most platforms do quite a bit of moderation, because otherwise their platforms would be overrun by spam. And, if they want people to actually use their platforms, nearly every site (even those like 4chan) tend to do significant moderation out of public pressure to keep certain content off. Yet, under SESTA you now face liability if you are shown to have any "knowledge" of violations of federal sex trafficking laws. But what do they mean by "knowledge"? It's not at all clear, as it just says "knowledge." Thus, now if a site, for example, discovers someone using its platform for trafficking and alerts authorities, that's evidence of "knowledge" and can be used against them both in criminal charges and in civil lawsuits.

In other words, somewhat incredibly, the incentive here is for platforms to stop looking for any illegal activity on their sites, out of fear of creating knowledge which would make them liable. How does that help? Indeed, platforms will be incentivized not to do any moderation at all, and that will create a mess on many sites.

The vague "knowledge" standard will be abused:

This is sort of a corollary to the first point. The problematic language in the bill is this:

The term ‘participation in a venture’ means knowing conduct by an individual or entity, by any means, that assists, supports, or facilitates a violation...

But what do they mean by "knowing conduct"? Who the hell knows. We already know that this is going to get litigated probably for decades in court. We have some similar problems in the DMCA's safe harbors, where there have been legal battles going on many years over whether the standard is "general knowledge" v. "specific knowledge" and what is meant by "red flag knowledge." And in SESTA the language is less clear. When people have attempted to pin down SESTA's sponsors on what the standard is for knowledge, they've received wildly varying answers, which just means there is no standard, and we'll be talking about lawsuits for probably decades before it's established what is meant by "knowledge." For companies, again, the best way to deal with this is to not even bother doing any moderation of your platform whatsoever, so you can avoid any claim of knowledge. That doesn't help at all.

The even vaguer "facilitation" language will be massively abused:

In that same definition of "participation in a venture" what may be even more problematic than the vague "knowledge" standard, is the vaguer claim that an entity "by any means, that assists, supports or facilitates a violation..." of sex trafficking laws, meets the standard of "participation in a venture." All three of those terms have potential problems. Assisting sounds like it requires proactive action -- but how do you define it here. Is correcting typos "assisting"? Is having an automated system suggesting keywords "assisting"? Is autocompleting search "assisting"? Because lots of sites do things like that, and it doesn't give them any actual knowledge of legal violations. How about "supporting"? Again, perfectly benign activities can be seen as "supporting" criminal behavior without the platform being aware of it. Maybe certain features are used in a way that can be seen as supporting. We've pointed out that Airbnb could be a target under SESTA if someone uses an Airbnb for sex trafficking. Would the fact that Airbnb handles payment and reviews be seen as "supporting"?

But the broadest of all is the term "facilitating." That covers basically anything. That's flat out saying "blame the tool for how it's used." Almost any service online can be used to "facilitate" sex trafficking in the hands of sex traffickers. I already discussed Airbnb above, but what about if someone uses Dropbox to host sex trafficking flyers? Or what if a sex trafficker creates advertisements in Google Docs? Or what if a pimp creates a blog on Wordpress? What if they use Skype for phone calls? What if they use Stripe or Square for payments? All of those things can be facilitation under this law, and the companies would have no actual knowledge of what's going on, but would face not only criminal liability but the ability of victims to sue them rather than the actual traffickers.

This is the core problem: this bill targets the tools rather than the law breakers.

Punching a hole in CDA 230 will be abused:

This is one that seems to confuse people who don't spend much time looking at intermediary liability protections, how they work and how they'll be abused. It's completely normal for people in that situation to not recognize how widely intermediary liability is used to stifle perfectly legitimate speech and activity. However, we know damn well from looking at the DMCA, in particular, that when you set up a process by which there might be liability on a platform, it's regularly abused by people angry about content online to demand censorship. Indeed, we've seen people regularly admit that if they see content they dislike, even if there's no legitimate copyright claim, they'll "DMCA it" to get it taken down.

Here, the potential problems are much, much worse. Because at least within the DMCA context, you have relatively limited damages (compared to SESTA at least -- the monetary damages in the DMCA can add up quickly, but at least its only monetary and it's limited to a ceiling of $150,000 per work infringed). With SESTA, criminal penalties are much more stringent (obviously) which will create massive incentives for platforms to cave immediately, rather than face the risk of criminal prosecution. Similarly, the civil penalties show no upper bound under the law -- meaning the potential monetary penalty may be significantly higher.

The chilling effects of criminal charges:

Combine all of this and you create massive chilling effects for any online platforms -- big or small. I already explained earlier why the new incentives will not be to help law enforcement or to moderate content at all, for fear of creating "knowledge" but it's even worse than that. Because, for many platforms, the massive potential liability from SESTA will mean they don't create any kind of platform at all. A comment feature on a website would become a huge liability. Any service that might conceivably be used by anyone to "facilitate" sex trafficking creates the potential for serious criminal and civil liability, which should be of great concern. It would likely lead to many platforms not being created at all, just because of the potential liability. For ones that already exist, some may shutter, and others may greatly curtail what the platform allows.

State Attorneys General have a terrible track record on these issues:

In response to the previous point, some may point out (correctly!) that the existing federal law already exempts federal criminal charges -- meaning that the DOJ can go after platforms if it finds that they're actively participating in sex trafficking. But, for as much as we rag on the DOJ, they tend not to be in the business of going after platforms just for the headlines. State AGs, on the other hand, have a fairly long history of doing exactly that -- including directly at the behest of companies looking to strangle competitors.

Back in 2010 we wrote about a fairly stunning and eye-opening account by Topix CEO Chris Tolles about what happened when a group of State Attorneys General decided that Topix was behaving badly. Despite the fact they had no legal basis for doing so, they completely ran Topix through the ringer, because it got them good headlines. Here's just a snippet:

The call with these guys was actually pretty cordial. We walked them through how we ran feedback at Topix, that how in January 2010, we posted 3.6M comments, had our Artificial Intelligence systems remove 390k worth before they were ever even put up, and how we had over 28k feedback emails and 210k user flags, resulting in over 45k posts being removed from the system. When we went through the various issues with them, we ended up coming to what I thought was a set of offers to resolve the issues at hand. The folks on the phone indicated that these were good steps, and that they would circle back with their respective Attorneys’ General and get back to us.

No good deed goes unpunished

So, after opening the kimono and giving these guys a whole lot of info on how we ran things, how big we were and that we dedicated 20% of our staff on these issues, what was the response. (You could probably see this one coming.)

That’s right. Another press release. This time from 23 states’ Attorney’s General.

This pile-on took much of what we had told them, and turned it against us. We had mentioned that we required three separate people to flag something before we would take action (mainly to prevent individuals from easily spiking things that they didn’t like). That was called out as a particular sin to be cleansed from our site. They also asked us to drop the priority review program in its entirety, drop the time it takes us to review posts from 7 days to 3 and “immediately revamp our AI technology to block more violative posts” amongst other things.

And, remember, this was done when the AGs had no legal leverage against Topix. Imagine what they would do if they could hold the threat of criminal and civil penalties over the company?

Similarly, remember how leaked Sony emails revealed that the MPAA deliberately set up Mississippi Attorney General Jim Hood with the plan to attack Google (with the letter Hood sent actually being written by MPAA outside lawyers?). If you don't recall, Hood used claims that, because he was able to find illegal stuff via Google, it meant he could go on a total fishing expedition into how it handled much of its business.

In the Sony leak, it was revealed that the MPAA viewed a NY Times article about the value of lobbying state AGs as a sort of playbook to cultivate "anti-Google" Attorneys General, who it could then use to target and take down companies the MPAA didn't like (remember, this was what the MPAA referred to, unsubtly, as "Project Goliath").

Do we really want to empower that same group of AGs with the ability to drag down lots of other platforms with crazy fishing expeditions, just because some angry Hollywood (or other) companies say so?

Opening up civil lawsuits will be abused over and over again:

One of the big problems with SESTA is that it will open up internet companies to getting sued a lot. We already see a bunch of cases every year where people who are upset about certain content online, target lawsuits at those sites just out of anger. The lawsuits tend to get thrown out, thanks to CDA 230, but lawyers keep trying creative ideas to get around CDA 230, adding in all sorts of frivolous attempts. So, for example, after the decision in the Roommates case -- in which Roommates.com got dinged for activity not protected by CDA 230 (specifically its own actions that violated fair housing laws) -- lots of people cite the Roommates case as an example of why their own argument isn't killed off by CDA 230.

In other words, if you give private litigants a small loophole to get around CDA 230, they try to jump in and expand it to cover everything. So if SESTA becomes law, you can expect lots of these lawsuits where people will go to great lengths to argue just about any lawsuit is not protected by 230, because of supposed sex trafficking occuring via the site.

Small companies will be hurt most of all:

There's this weird talking point making the rounds, that the only one really resisting SESTA is Google. We've discussed a few times why this is wrong, but let's face it: of all the companies out there, Google is probably best positioned (along with Facebook) to weather any of this. Both Google and Facebook are used to massive moderation on their platforms. Both companies have built very expensive tools for moderating and filtering content, and both have built strong relationships with politicians and law enforcement. That's not true for just about everyone else. That means, SESTA would do the most damage to smaller companies and startups, who simply cannot invest the resources to deal with constant monitoring and/or threats from how people use their platform.

Given all of these reasons, it's immensely troubling that SESTA supporters keep running around insisting that the bill is narrowly tailored and won't really impact many sites at all. It suggests either a willful blindness to the actual way the internet works (and how people abuse these systems for censorship) or a fairly scary ignorance level, with little interest in getting educated.

Filed Under: cda 230, congress, grandstanding, intermediary liability, knowledge, moderation, sesta

China's New Online Encyclopedia Aims To Surpass Wikipedia, And To 'Guide And Lead' The Public

from the great-wall-of-culture dept

China has a long history of producing encyclopedias that goes back thousands of years. One of the most famous works is the fifteenth-century Yongle encyclopedia, which had over 15,000 volumes, and is still the largest paper-based general encyclopedia ever created. More recently, the main publication in this field was the Encyclopedia of China, whose first edition had 74 volumes. Later, CD-ROM and online versions were added. The third edition has just been announced, and although it is not quite on the scale of the Yongle encyclopedia, it is ambitious in its scope:

The third edition of the Chinese Encyclopaedia is currently China's largest publication project, with more than 20,000 authors from universities and research institutes contributing to articles in more than 100 disciplines.

Designed to be the nation's first digital book of "everything", it will feature more than 300,000 entries, each about 1,000 words long, making it twice as large as the Encyclopaedia Britannica, and about the same size as the Chinese-language version of Wikipedia.

As the article in the South China Morning Post notes, access to Wikipedia is patchy in China. Most of the uncontroversial articles can be read, but searches for sensitive keywords such as "Dalai Lama" and even "Xi Jinping," have a habit of timing out. The new project is clearly designed to steer people towards safer opinions:

"The Chinese Encyclopaedia is not a book, but a Great Wall of culture," Yang Muzhi, the editor-in-chief of the project and the chairman of the Book and Periodicals Distribution Association of China, told senior scientists at a meeting at the headquarters of the Chinese Academy of Sciences (CAS) in Beijing on April 12, according to a report on the academy’s website the next day.

...

Yang told the meeting China was under international pressure and felt an urgent need to produce its own encyclopaedia to "guide and lead the public and society".

Speaking of Wikipedia, Yang went on:

"The readers regarded it to be authoritative, accurate, and it branded itself as a 'free encyclopaedia that anyone can edit', which is quite bewitching," he wrote. "But we have the biggest, most high-quality author team in the world ... our goal is not to catch up, but overtake."

China certainly has the resources to complete this huge project by 2018, its planned launch date. And once those 300,000 entries are available to "guide and lead the public," it's hard not to think that accessing the rival Wikipedia will be made so hard that most people will give up trying, and stick with the new Chinese Encyclopedia. At that point, the Chinese authorities will indeed have created a "Great Wall of culture" to complement that Great Firewall of China, both designed to keep out all those inconvenient ideas.

Follow me @glynmoody on Twitter or identi.ca, and +glynmoody on Google+

Filed Under: china, encyclopedia, encyclopedia of china, knowledge

Initiative for Open Citations Takes Alternative Approach To Freeing Up Knowledge

from the structured,-separable,-and-open dept

We've just written about widespread frustration at the slow pace of the shift to open access publishing of academic papers, and about how some major funding organizations are trying to address that. Open access aims to make entire publications publicly available, and that is meeting considerable resistance from traditional publishers who derive their healthy profits from charging for subscriptions. Rather than continue to tackle publishers head-on, an interesting new project seeks instead to liberate only a particular part of each article, albeit an important one. The new Initiative for Open Citations (I4OC) seeks to promote the unrestricted availability of the list of citations that form a key part of most academic articles:

Citations are the links that knit together our scientific and cultural knowledge. They are primary data that provide both provenance and an explanation for how we know facts. They allow us to attribute and credit scientific contributions, and they enable the evaluation of research and its impacts. In sum, citations are the most important vehicle for the discovery, dissemination, and evaluation of all scholarly knowledge.

As the number of scholarly publications is estimated to double every nine years, citations -- and the computational systems that track them -- enable researchers and the public to keep abreast of significant developments in any given field. For this to be possible, it is essential to have unrestricted access to bibliographic and citation data in machine-readable form.

The present scholarly communication system inadequately exposes the knowledge networks that already exist within our literature. Citation data are not usually freely available to access, they are often subject to inconsistent, hard-to-parse licenses, and they are usually not machine-readable.

The I4OC aims to address those problems by encouraging all publishers, whether open access or otherwise, to provide the data on citations found in their journals in a form that is structured, separable, and open. "Structured" here means that it is held in a common data form that is machine readable. "Separable" refers to the fact that even for non-open access materials, the citation data is nonetheless freely available. And "open" means that it is released as raw facts, and thus without a license, or uses a CC0 public domain dedication that makes it quite clear that the citation data can be used for any purpose without needing permission.

As the I4OC home page explains, a key benefit from this new approach is increased discoverability of published articles, since even if they are not freely available, their citation data will be out in the open. Another is citation data can be analyzed in new and complex ways thanks to its machine-readable nature. Finally, it may be possible to create new services and even new businesses based around the new data resource.

All of that is highly welcome, but the fact that a separate initiative was required to make it happen underlines that fact that too much of humanity's knowledge remains locked up behind paywalls, where its full potential is hard to realize. The correct solution to that is not making one element available, but liberating the full texts as open access. And that means real open access, not the subverted kind that Richard Poynder analyzed in his compelling and troubling post "Copyright: the immoveable barrier that open access advocates underestimated".

Follow me @glynmoody on Twitter or identi.ca, and +glynmoody on Google+s

Filed Under: knowledge, open access, open citations

How To Improve Online Comments: Test Whether People Have Read The Article Before Allowing Them To Respond

from the probably-asking-too-much dept

For a while now, Techdirt has been writing about the decision by some sites to stop allowing readers to make comments on articles. We've pointed out that's pretty regrettable, especially when it's couched in insulting terms of "valuing conversations" or building "better relationships." Dropping comments is a lazy response to a real and challenging problem: how to encourage readers to engage in meaningful ways.

As well as a natural tendency for people to write hurtful or insultings things that they probably wouldn't say to each other face-to-face, there's another problem: the rise of Internet troll factories whose entire purpose is to flood sites with propaganda in the form of comments that espouse a particular viewpoint. As we noted recently, Google is looking to use machine learning technology to help identify and then deal with toxic comments:

a publisher could flag comments for its own moderators to review and decide whether to include them in a conversation. Or a publisher could provide tools to help their community understand the impact of what they are writing -- by, for example, letting the commenter see the potential toxicity of their comment as they write it. Publishers could even just allow readers to sort comments by toxicity themselves, making it easier to find great discussions hidden under toxic ones.

As Google itself admits, the issue is "about more than just improving comments. We hope we can help improve conversations online." A rather clever way to do that has been devised by NRKbeta, the technology site of the Norwegian government-owned radio and television public broadcasting company, NRK. Here's the basic idea (via Google Translate):

a small [on-screen] module is presented to you as a reader with three questions from the article that you must answer in order to be able to contribute to the discussion.

Actually reading the article before you comment on it -- pretty revolutionary, no? NRKbeta realizes that it's not a perfect solution:

We know of course that it is possible to "cheat" with these questions by searching the text above [the on-screen module], and that using this approach it cannot be guaranteed that everyone actually read the article, but we still think it's worth the experiment.

It's hard not to agree, because it tries to tackle one of the root causes of comments that add nothing to the conversation -- a failure to read what the article said -- by making it a pre-requisite before you can add your own thoughts. It also has the virtue of being extensible in various ways. For example, there could be more than three questions in the pop-up box, and your comment's place and prominence in the conversation could be determined by how many you get right. This might allow the thoughts of more engaged readers to bubble naturally to the top of the conversation. The fact that the code for the feature has been released as free software makes experimentation even easier. NRKbeta's idea certainly seems a better approach than simply giving up and removing comments altogether.

Follow me @glynmoody on Twitter or identi.ca, and +glynmoody on Google+

Filed Under: comments, knowledge, tests, trolling
Companies: nrkbeta

Ridiculous German Court Ruling Means Linking Online Is Now A Liability

from the never-link dept

Earlier this year, we wrote about a crazy decision in the EU Court of Justice, that determined mere links could be direct infringement on commercial websites (with "commercial" being not well defined). Now as various courts in the EU try to put this ruling into practice, they're already making a mess of it. In particular, a German case has set an impossible standard for a site, finding a site to have infringed on the copyrights of a photographer for merely linking to a photograph. And the backstory here is even crazier.

The image in question was originally uploaded to Wikimedia Commons by the photographer and put under a Creative Commons license (unfortunately, no one seems to name which CC license). And then this happened:

This picture was then modified by an unknown third person, who added UFOs to the picture that appear to be flying above the building. This new picture was uploaded by the third person on his website.

The defendant operates a website where he publishes and sells educational material that he creates. In the imprint of this website, the defendant posted a sentence that included the word “UFO”, which linked to the modified “UFO-Version” of the claimant’s picture. Clicking on the word “UFO” a new browser window in which the “UFO-Version” of the picture was shown would be opened.

The photographer deemed that this constituted an infringing use of his work and brought the defendant to court.

Now, let's just review this because we're already stretching all sorts of concepts to the point of breaking. The photographer uploaded an image under a CC license to Wikimedia Commons. Someone else modified it. Someone else -- operating a site that sells educational materials -- merely linked to the modified image from the word "UFO" -- and that's who gets sued.

This seems like a classic "Steve Dallas Lawsuit" in which the target is a corporate entity because "that's where the money is" even if it makes no sense at all for the liability to be applied there.

And here's the really nutty part: the German court agreed because of that insane CJEU ruling that says a "commercial site" should be assumed to know the copyright status of everything they link to. So merely linking to this silly modified photograph, the site itself becomes liable for direct infringement. That's insane. Yes, (depending on the CC license in question), it's entirely possible that the modified version was infringing, but that's no reason to make a site that merely links to that image liable for direct infringement over it.

At this point, if you run a "commercial" website in Europe that links off site to anything, you have a tremendous liability hanging over your head due to the insanities of copyright law.

Filed Under: cjeu, commercial use, eu, germany, knowledge, liability, linking

This Is Huge: New Project Releases All Current (Non-Confidential) Congressional Research Service Reports

from the big-big-news dept

Going back nearly a decade, we've been talking about the ridiculousness of Congress refusing to publicly release reports from the Congressional Research Service (CRS). As we've discussed many times, CRS is an in-house think tank for Congress that is both famously non-partisan and actually really good at what they do. CRS reports tend to be really useful and highly credible (which is part of the reason why Congress isn't a fan of letting them out into the public). Of course, as works of the federal government, CRS reports are in the public domain, but the way it's always worked is that the reports are released only to members of Congress. These include both general reports on topics that are released to every member of Congress, or specific research tasked by a member for the CRS to investigate and create a new report. The members who receive the reports are able to release them to the public, and some do, but the vast majority of CRS work remains hidden from public view. For the most part, both CRS and Congress have resisted any attempt to change this. Going back decades, they've put together a mostly ridiculous list of reasons opposing plans to more widely distribute CRS reports.

Some members of Congress keep introducing bills to make these public domain CRS reports actually available to the public. We've written about such attempts in 2011, 2012, 2015 and earlier this year. And each time they get shot down, often for completely ridiculous reasons, including the belief that making these reports public will somehow hurt CRS's ability to continue to do good, non-partisan research.

At times, different organizations and groups have taken up the cause themselves. Back in 2009, Wikileaks hit the jackpot and released nearly 7,000 such CRS reports. Steve Aftergood at the Federation of American Scientists has been posting CRS reports to a public archive for quite some time. There's also Antoine McGrath's CRSReports.com and some other sites that all create archives of CRS reports that they've been able to collect from various sources.

But earlier this week, there was a new entrant: EveryCRSReport.com. Unlike basically all of the other aggregators of CRS reports that collect released reports and aggregate them, it appears that EveryCRSReport basically has teamed up with members of Congress who have access to a massive stash of CRS reports loaded onto the Congressional intranet, all of which have been released via the site -- and it appears that the site is automatically updated, suggesting that the still nameless Congressional partners have set up a way to continually feed in new reports. To avoid public pressure or harassment (one of the core reasons used by Congress and CRS to reject proposals to open up the content), the site removes the names and contact info of the CRS staffers who create the reports. The reports that are available are not just in unsearchable PDFs, but they're fully HTML and fully searchable.

Here are a few reports that folks around here might find interesting: an analysis of ACTA and a recent deep dive into the net neutrality debate. Here's an interesting one on promoting internet freedom globally. Since the peaceful transition of presidential administrations has suddenly become a hot topic (not for good reasons), here's a CRS report on that from just last month. It's also good to see that they have a recently updated list of cybersecurity reports and research for Congressional staffers to dig into (though it's unclear how many actually do so).

And, yes, of course, there's one on the "going dark" encryption debate, in which the CRS report rightly notes that backdoors are a bad idea, according to basically all experts:

In considering future legislation on or regulation of encrypted systems and communications, the issue of exceptional access has been raised: is it possible to create a system with sufficiently narrow and protected access points that these points can only be entered by authorized entities and not exploited by others? Experts have generally responded, no. For instance, one group of computer scientists and security experts contends that requiring exceptional access "will open doors through which criminals and malicious nation-states can attack the very individuals law enforcement seeks to defend." As was the case during the crypto wars of the 1990s, new technology (the Clipper Chip) was introduced that was intended to only allow access to certain communications under specified conditions. Researchers were soon able to expose vulnerabilities in the proposed system, thus halting the implementation of the Clipper Chip.

This is a really awesome resource -- it's a goldmine of useful information, and very thorough, careful research. I've only just started digging in.

The whole thing was put together by Demand Progress* and the Congressional Data Coalition, which is a project created by Demand Progress and R Street (which our think tank, the Copia Institute, is a member of). It will be interesting to see how (if?) Congress and the CRS react to this. Hopefully, they don't freak out, and seek to shut down the various sources of this material. This really is a fantastic resource of carefully done, thorough research on a variety of topics, all technically in the public domain. Check it out.

Hopefully it will help both the rest of Congress and CRS to recognize that actually making publicly funded research public is not such a bad thing. The site itself was put together by Dan Schuman, who used to work for CRS, and he's actually written up a fascinating blog post about why he did it and why the internal culture at CRS, against such public releases, is wrong, but endemic to the organization (he didn't begin questioning it himself until after he left):

Over time, I came to realize that the policy concerning public access to CRS reports was counterproductive. Members of Congress could get the reports. Lobbyists and special interests could get the reports from Congress or from private vendors for a fee. Former congressional staff could ask their friends on the hill for a copy. But the general public, unless they knew a report existed, really did not have access.

And that’s too bad. CRS reports are written for intelligent people who are not necessarily policy experts. In a world that’s awash with 5 second YouTube ads, horse race political coverage, and the endless screaming and preening of political figures, these reports are a good way to start to understand an issue.

But he also notes that there are problems with CRS -- some of which CRS blames on the fact that reports are being released to the public -- including the fact that the reports have become "even-handed to a fault" to avoid pissing off Congress itself in talking down a bad idea. While some of this may also be attributed to worries about reports going public, this seems kind of silly. This is good and credible taxpayer funded research that's in the public domain. If Congress can learn from it, so can the public:

CRS used to be a very different agency. It used to provide unvarnished advice for members of Congress on the crucial issues of the day. But over time, and especially during the 1990s, the mode of analysis changed to a description of issues, moving away from an evaluation of the strengths and weaknesses of various courses of action. I don’t mean to overstate this, and there are many examples still of prescient analysis, but there was a real change in the way CRS did its work, in large part because of existential concerns. In short, CRS was concerned about irritating its congressional masters by attacking a pet project or cherished belief. The old-timers still had great latitude, but the agency became sclerotic.

Part of this calcification included a fear of public access to the reports. At one time, CRS had published a newsletter about its latest research. And now, while its employees still testify before Congress, they were discouraged and then generally prohibited from sharing their work even with their academic peers. Agency staff grew more insulated and isolated.

But on top of that, recognizing that there are benefits to this research being public, hopefully means that CRS can get beyond just giving out "even handed to a fault" research, and can actually get back to making real recommendations. Over the years, we've discussed the ridiculous move by Newt Gingrich a couple decades ago to kill off the Office of Technology Assessment, which actually helped Congress understand complex technological issues in a non-partisan way. A functioning CRS could do the same thing and help put an end to stupid technology debates that often feature clueless arguments on all sides. CRS shouldn't fear this role, nor should it fear its research being public. It's a great resource and having it public is great for everyone.

* I'm on the board of Demand Progress, but had no idea about this particular project from them, and, in fact, heard about it from someone else entirely...

Filed Under: congress, congressional research service, education, knowledge, public domain, resources