Instagram: Smart and Non-Smart Filtering

Iranian media have reported that smart filtering of the Instagram social network has been initiated since Wednesday night, meaning that only certain Instagram pages deemed "immoral" or "criminal" by Iranian officials will be filtered. Previously, the Iranian Minister of Communications had repeatedly emphasized the necessity of implementing "smart filtering" and stated that the government and authorities agree on this matter. However, it appears that among all websites and social networks, Instagram was the first to be targeted for "smart filtering." Radio Farda conducted an interview with Arash Abadpour, an internet activist and computer specialist, regarding the dimensions of this event. Mr. Abadpour, it has been reported that smart filtering of the Instagram social network has begun. How feasible do you think the implementation of this program is? Yes, we have news that this has happened on Instagram. If we look back at the news, similar discussions about Instagram existed last year. At that time, it was denied whether filtering had occurred or not. We have seen specific pages being filtered before. In the filtering system, which operates based on the page address, i.e., the URL, this is entirely possible. In Instagram links, it is also clear which page belongs to which account. Therefore, for example, if the Instagram account "Rich Kids of Tehran" is identified as problematic by that flawed structure for any reason, it can be filtered. However, when this issue was discussed in the media, it was raised that this is part of implementing smart filtering. It should be noted that this method is actually a precursor to smart filtering because smart filtering was supposed to perform a different function. A memorable event in the Iranian blogosphere was when filtering agents searched for specific words, and any website containing those words was automatically filtered. Is something similar going to happen with social media smart filtering, or is a different model being pursued? We have heard contradictory explanations about smart filtering so far. It has been repeatedly stated that smart filtering will start at a certain time, but then we saw that this did not happen. In any case, it seems that this has been implemented in Qom province, and the smart filtering system is operational. The explanation we hear is that smart filtering aims to bring that blockage from earlier points in the network to the access level. This means what the government describes as a family-friendly internet. In this model, a page opens, but before it is shown to the user, there is a code that determines whether, for example, a photo or text is appropriate or problematic. Or they can tell from the URL that this text and photo should be removed. Regardless of whether that structure has the right to prevent citizens from accessing specific pages, this is the logic that smart filtering is supposed to have. Non-smart filtering is exactly what you mentioned. It is very blind and simply removes pages with specific URLs. Or if a specific word exists, it removes it, and it is not actually smart. In the explanation given by the Minister of Communications about smart filtering, he says if 90 or 95 percent of the content of websites is appropriate, we filter 5 to 10 percent of inappropriate content. However, it seems that this is very different from what has been done because it still follows the previous method, which is the complete removal of a website or an account. Many times, our discussion is about the technical aspects of filtering and exactly how it is happening. But we need to understand the issues at a higher level and ask to whom the structure that performs filtering is accountable, under what laws it operates, and how it can be protested against. Do you think there is any guarantee that at least those pages that are not considered criminal under the laws of the Islamic Republic will not be filtered? We have seen similar events in the Persian blogosphere where some websites were filtered despite adhering to general frameworks. Here, we are moving beyond the technical discussion and looking at the political and social framework of the issue. This discussion is very important. Many times, our discussion is about the technical aspects of filtering and exactly how it is happening. But we need to understand the issues at a higher level and ask to whom the structure that performs filtering is accountable, under what laws it operates, and how it can be protested against. This structure has so far operated in a way that it decides which pages to filter and sometimes provides us with explanations, for example, saying that it filtered a website because it was pornographic or promoted violence. However, everything remains at this stage, and it is not accountable beyond that. That is, we cannot say that this page does not fall under these definitions and should not have been filtered. When we do not have this possibility, that structure also has complete freedom. We have no guarantees, and there is no place for protest or oversight. In fact, there is no democratic structure. Citizens cannot enjoy the right to access information. Therefore, we need to rise above that level of technology and ask what legitimacy the institution that filters has. What happens that the Iranian government adopts such a policy, whereas previously, it preferred to completely filter a network like Facebook to erase the issue? Here, we need to look at the cost-benefit analysis of that structure. That is, we know that structure wants to have complete oversight over all communication channels of citizens. For this reason, it has absolute control over newspapers, books, magazines, etc. The internet is a problematic matter. Individuals can easily speak on the internet, and oversight is very minimal, making it difficult to find them. So the government tries to impose this oversight. However, costs also become important here. Completely cutting off a tool like Facebook, which has popularity, imposes a high cost on the structure. That structure has gradually come to realize that there may be solutions that can reduce this cost. Well, reducing internet speed is one of these methods, or using the Chinese method, i.e., creating parallel social networks. These parallel social networks are within Iran and are under supervision. We see that the Minister of Communications has recently talked about this and said that now that social networks like Viber have been filtered, we encourage users to use Iranian tools that are free. These are the methods of that structure to reduce its costs because it does not want to bear the cost of completely filtering Facebook or Instagram. It can approach this selectively. This way, it can at least superficially claim that the internet is free and only problematic pages have been blocked. This is a well-known method in the world, and Iran's filtering structure is trying to do this. Completely cutting off a tool like Facebook, which has popularity, imposes a high cost on the structure. That structure has gradually come to realize that there may be solutions that can reduce this cost. So we should expect to hear news about lifting the filter on Facebook as well. Is that correct? It depends on how much of what is said about smart filtering is realized. Personally, I do not think that smart filtering can meet the expectations placed on it. This is because if internet restrictions are lifted, speed restrictions and access limitations are removed, and geographical access improves, then the volume of internet traffic is so high that complete oversight on it becomes very difficult. Moreover, smart filtering is essentially a method of using artificial intelligence algorithms to detect content. You provided a good example, using keywords. Suppose the names of political leaders are placed on a blacklist, and therefore if there is a page on Facebook that mentions, for example, Ayatollah Khamenei in the comments, that page is filtered. Well, we know there are various methods to use alternative words. In the same blogosphere you mentioned, people started intentionally misspelling words. Well, the system will face significant challenges in countering this method. For this reason, I do not think smart filtering will be practical for 70 million people who do not want to be filtered. There are alternative methods that people use, and the structure quickly realizes that its costs remain very high and it has to revert to the more traditional and older methods it used. On one hand, we are witnessing President Rouhani's government's enthusiasm for a national internet, and on the other hand, we hear about smart filtering. How do these relate to each other? The national internet is an alternative solution that can, at least on paper, solve many of these problems. It is an idea that the Chinese have implemented for a long time and have received good results from it. You see in the Chinese social network, a keyword is completely censored. People also know that all the logs of the network are in the hands of security forces, and therefore self-censorship occurs. If it were possible for Iran's traffic to be cut off from global traffic and, for example, instead of Facebook, we had Fesnema, instead of Twitter, we had its internal version, and instead of Gmail, we had a national email, then the issue would be somewhat erased. In this case, users would be working on servers located within Iran, and there would be complete oversight over them, all encryption algorithm keys would be in the hands of the government, and the government would no longer have concerns about this because it knows users are aware that everything they do is being stored. However, executing this project has a fundamental problem, which is the technical and financial investment at the level of several large global companies. Something like Google with its tools is necessary. Something like Facebook with its tools is needed. Something like Twitter is probably needed. Something like Amazon is needed. This technological and financial investment is so high that it calls into question whether national internet is really possible. On the other hand, individuals and users must be attracted to the internet. The internet must be attractive enough for them to do this. In any case, several parallel paths are progressing in the project of internet oversight in Iran. The national internet project is one of them. It seems that there are numerous problems ahead of it. Smart filtering is another one of these. Other similar actions are being taken. Ultimately, the fundamental goal is that the structure can impose complete oversight and, on the other hand, individuals feel that they are under complete oversight when they are on the internet, and therefore engage in less activity against that structure.