Carl Miller of Demos Still Misfires on ‘Islamophobia’

Carl Miller (@carljackmiller) of Demos, a ‘Cross Party Think Tank’  has produced some research that claims to show spikes in ‘Islamophobic’ tweets around incidents of Islamic terrorism.

There are problems with this research, as ponted out very well in this piece, by Benjamin Jones: Conflating abuse with criticism of Islam risks a return to a UK blasphemy law [1], from the National Secular Society (NSS). One big problem is the word ‘Islamophobia’ and how it is used; which in turn leads to a subsequent problem, the selection and analysis of the tweets used to produce the results.

Carl Miller responded to Benjamin here: Measuring Islamophobia on Twitter [2]. He acknowledges the problem, but then goes on to compound it.

In addition to the above we have the BBC Victoria Derbyshire programme on this [3], in which Carl Miller hightlights the conflation problem, attributes the fault incorrectly.

And then there’s Demos report [4], which contains an explanation of the data, but which has its own problems.

Islamophobia in Context

The article by Benjamin Jones [1] covers pretty much all aspects of the ‘Islamophobia’ problem, so do read that first.

But bear in mind these points as we go through all this:

  • Muslims conflate Islam and Muslims ‘as a monolith’ when it suits them, but not when it doesn’t. All Muslims are brothers and sisters – especially when debating of in conflict with non-Muslims. But they are also quick to denounce each other when it’s convenient. This is an essential component to the confusion about the term ‘Islamophobia’ and why many Muslims will take criticism of Islam to be criticism of all Muslims, and as anti-Muslim bigotry.
  • Non-Muslim pseudo-liberals that are more intent on defending what they see as a persecuted minority, are regularly confused by all this, to the extent that they will defend Muslims that are explicit in their own bigotry, about homosexuals or women, rather than defend the freedom of speech to criticise those bigoted beliefs that are backed up by religious texts. They will also go lightly on the persecution of minorities within minorities. Oh, yes, of course they oppose FGM, honour killings, forced marriages, homophobia … but try and do a Demos type analysis on their Twitter stream and you will recognise them by the absence of such denouncements and the abundance of ‘Islamophobia’ spotting.

Just so you’re clear how this goes, here’s an example:

  • Women in Iran start the #NoHijabDay trending on Twitter to oppose the forced modesty rules in Iran.
  • Western liberals retweet support with #NoHijabDay – but hardly a mention from many of the well known pseudo-liberals.
  • Some privileged western Muslims object to this ‘persecution’ of Hijabis, as if the trend is demanding they don’t wear it. They start #HijabDay.
  • Western pseudo-liberals go ape about the ‘Islamophobia’ they too have witnessed with #NoHijabDay and not only echo #HijabDay, but don the hijab in solidarity with their now persecuted Hijabi Muslim friends.

Question: How many of these liberated Muslim Hijabis will remove their Hijabs in solidarity with their Muslim sisters in Iran?

Question: How many of these liberated Muslim Hijabis wear some Christian paraphinalia in support of persecuted Chritians in ‘Muslim lands’?

On that second question in particular one can only imagine the crap they’d take online from fellow Muslims if they did. And though we’re often told that the hijab, the niqab and other garments are a ‘choice’, what crap do you suppose the average niqab wearing woman in East London would get from her husband if she went out in public without it in solidarity with those forced to wear it?

This is how dumb this whole ‘Islamophobia’ mess has become.

Again it’s worth remembering that many Muslims wriggle out of any criticsm of Islam with nuance, scholarship, metaphor [seriously, the explicit prescription for lashes for sex outside marriage in Q24:2 has been described to me as a ‘metaphor’ by some Muslims]. But this playing with words, that are supposed to be inerrant but which don’t mean what they say, is a problem for Muslims, that should not be set at the door of critics of Islam that point out the barbarity in the Quran.

This is significant: The Quran is not like the Humanist Manifesto, which includes no prescription to wage war or be violent towards non-Humanists. It’s not like the US Constitution, which includes no hateful remarks towards non-Americans. Even Marxism is quite limited in it’s prescriptions in this regard. But Islam has very specific requirements to do violence against unbelievers, and it asserts this is a duty – and the only point in question among Muslims is when this applies. Even liberal and reformist Muslims accept these passages for what they are, but reject their applicability now in the 21st Century. But many Muslims will assert that there are conditions when this injunction to kill unbelievers applies.

So, with that in mind, on with Carl Miller’s response to Benjamin.

Carl Miller’s Response to Benjamin Jones

In Carl’s response [2], he deals with the definition of ‘Islamophobia’.

The first two categorise of Islamophobia were, I think, uncontroversial:

  • The first was hateful abuse directed towards Muslims themselves, and especially those using derogatory slurs and descriptors;
  • The second was abuse that conflated Muslims generally (as people) with sexual violence

No, this IS the controversy!

The ‘first’ hateful abuse towards Muslims need have only minor relationship to Islam itself, since that category of hate can be based on a very limited understanding of Islam, and little interaction with Muslims. It might well be based on an irrational fear. But in that case the best ‘phobia’ term for it would be Muslimphobia. You can tell a Muslimphobe: a) they think Sihks, Hindus, ex-Muslims (i.e. all brownish people) are Muslims; b) their suspicion of brownish people being terrorist is raised up a few notches.

The ‘second’ hateful abuse is little different, in that it adds only a cursory knowledge of the association of sexual violence with Muslims. But having said that it is only the generalisation of that view that is problematic.

There are plenty of non-Muslim sexual abusers, but it is very clear from Islamic texts and from the words of many (I’m not being hyperbolic with ‘many’, but I’m not saying ‘most’) Islamic preachers and self-appointed scholars that their opinions on sexual slavery, child marriage, beating wives, is explicitly supportive.  These are real problems that some Muslims acknowledge – so unless Carl wants to engage in the ‘Uncle Tom’ slurs against these liberal Muslims then tweets relating to these issues are going to be very difficult to parse. I wonder if their tweets ended up in Carl’s data.

The key point here is that this is not ‘Islamophobia’ – it is not an irrational fear of Islam. Benjamin Jones makes this clear. The term ‘Islamophobia’ is a misapplied word that has been put to good service in discrediting criticism of Islam, by conflating it with hate speech.

Demos could quite easily have avoided the term altogether, and so completely removed this as an issue. They could have used ‘Anti-Muslim Bigotry’; or if that’s not catchy enough they could have used ‘Muslimphobia’, which would at least have made it clear that they were not conflating such bigotry with a criticism of Islam.

For all Carl Miller claims to be aware of the problematic use of the term, his report, and especially his appearence on the BBC programme with Victoria Derbyshire [3] and others only contributed to the abuse of the term. In fact from that public appearences from him and others involved you’d think there was no problem with the word at all. When Carl does mention a problem of distinguishing hate from fair criticism, he says the fault, the conflation, is in the tweets of the hateful Muslim bashers (cited later). He totally ignored the way in which Muslims and pseudo-liberals conflate the two.

Let’s have a look at the example tweets Carl provides that are supposed to come under the two examples definitions above (I’ve added a-h):

Here are some more examples of both. I’ve changed them slightly to stop the senders being retrospectively identified. The meaning is unchanged.

a) @xxx go eat curry you bastardised pack

b) It is very very hard to get lower than muslim scum, somehow the leftys managed to do it – <LINK>

c) @XXX @XXX fuck ur sand nigger parents #hack

d) How did that paki cunt end up mayor?

e) @xxx we have sand nigger and spics here — You guys all stick together – Animals stick together

f) Trump: My temporary ban on Ragheads entering the U.S. is a “suggestion” – <LINK> (NO, DO NOT BACK DOWN FROM THIS)

g) @XXX @XXX @XXX @XXX @XXX u will see when a muzzie rapes your daughter or wife u spineless weasel

h) American Women! How to deter male Muslim attack. In ur purse carry a spray bottle Pigs blood! Announce its pigs blood! Also draw ur Glock!

I’ve tracked down some of these tweets – by searching for phrases used, from April to August 2016.

Note especially in these examples that there was no reference to Islam as a religion or a political ideology. Nothing! Islam-ophobic? It’s clear that some are simply racist and there’s no more than a passing connection with religion.

Note also that some of these are from the US, and yet in the report [4] this claim is made:

The report also analysed Islamophobic Tweets that were likely sent from the UK. …  On tests of this method of geolocating Tweets, it has been found to be between 80 and 90 per cent accurate for those Tweets it could locate, and be able to locate between 40 and 70 per cent of Tweets.

The report is clearly targeted at UK use, where possible. The report does show maps of Europe, but they emphasise the focus on the UK. And in the original Demos post [5], and in the BBC programme [4], the active map of tweets lighting up is very clearly aimed at Britain. Further more, the BBC programme makes it seem like it’s all about British anti-Muslim bigotry.

And yet the example tweets have a US location? Why is that? Has Carl no better examples, from Britain?

(a), (b), (e), (g) – Not found.

(c) USA right wing trump supporter. Ethnicity unlcear (avatar may not be representative). Follows and followed by some military people. Why is this included?

(d) Yorksire UK. Lover of some black music. Occasional racist tweets, but implied support for black guys shot by cops in US. More abusive about Manchester & Liverpool than Muslims or asians. “Fuckin southern student cunts” Searched for ‘Muslim’ & ‘paki’ on this account and only the above tweet was found. Mostly sounds off about personal thoughts, not exceptionally racist or anti-Muslim. So, given this person’s affinity for abuse generally, this is explicitly racist, but not ‘Islamophobic’ in the sense that the term is generally (mistakenly) used. And it has nothing to do with Islamic terrorism, and yet much about the Demos report and the BBC appearance focus on the spikes in ‘Islamophobic’ tweets around terror events. Why is this an example in this study at all? Because the methodology – particular the bias of the analysts that train the system, cannot distinguish between general racism and ‘Islamophobia’?

(f) Geolocation South America, but misleading. Lots of Trump support. Clear opposition to Islam, but the racist element isn’t present. Is this anti-Muslim bigotry, or an over-generalisation – he refers to incidents that anyone opposed to Islam might object to. Slightly bigoted. Why is this given as an example in what is sold as a British problem?

(h) American right, anti-democrats, anti-Obama. Not racist as he shows support for black republicans. Anti-Islam. References to attacks from Muslims accur around rape and assault attacks by Muslims, in Europe for example. This is a case of generalisation, but it’s not clear that he would inists ALL Muslims are guilty by implication of particular acts by Muslims. This sort of generalisation is fairly classed as anti-Muslim bigotry, but it’s not clear he would hold hateful views about all Muslims, if he knew some. Why is this given as an example in what is sold as a British problem? Why is this associated with the problematic word ‘Islamophobia’ when it’s really about the criticism of Islam?

Note that (e), for example, could even be aimed at a non-Muslim who appears to come from the Middle East – such as, many of the atheist non-Muslims from the Middle East that defend Muslims against racist abuse. Note also that the inclusion of (e) in the data is tantamount to racism itself – the expectation that ‘sand nigger’ addresses only Muslims, implies that all people associated with dessert lands of North Africa and the Middle East are Muslims. This is the nature of Twitter that needs to be accounted for, and there’s no clue in the Demos report that it is.

So in these example tweets there are two  references to ‘Muslim’ and one use of the term ‘muzzie’. There’s the reference to pigs blood, which does refer to the dietry habits of Muslims. But, what is very clear is that these are overtly racist tweets and not anti-Islamic. They are not ‘Islamo-phobic’ but ‘Muslim-phobic’, ‘Arab-phobic’ (dessert reference) and ‘Pakistani-phobic’ (paki reference).

Next, Carl moves on to ‘Islamophobia’ definition categories he sees as more (i.e. actually) problematic, as if the above isn’t enough.

It was the final two categories that, I think, some found particularly problematic. These were:

  • A category we called ‘Islam is the enemy’, which contained claims (quoting the original paper) that ‘it is a fundamental injunction of Islam for all of its followers to be engaged in a violent struggle against non-Muslims and the west’;
  • and (similar) especially in the wake of a terrorist attack, the apportioning of blame for the attack not on the terrorists themselves, or on Islamist militancy, but on the Muslim population generally.

These are indeed problematic too.

The first category itself conflates two distinct issues. It’s quite possible and reasonable to claim ‘Islam is the enemy’, as one might for ‘Communism’ or ‘Fascism’, for it is a matter of fact (often ignored by many Muslims themselves, sometimes conveniently so) that Islam is a political-judicial-religious ideology. The texts of Islam, including the core document, the Quran, make very specific prescriptions and proscriptions relating to political and judicial affairs, as well as theological and moral.

This aspect of Carl’s ‘first category’ (in this second set) is quite distinct from the next part of it: ‘it is a fundamental injunction of Islam for all of its followers to be engaged in a violent struggle against non-Muslims and the west’. But that too is a case that can be made, directly from the texts of Islam. This is not ‘Islamophobia’!

Remember – and this is important – to most Muslims the Quran is the inerrant word of God. It means what it says. Oh, except it doesn’t. This is the contradictory nature of Islam that critics have to deal with. It is totally unjustified to accuse critics of ‘Islamophobia’ – even when their criticism is laced with abusive terms, if the abuse isn’t especially racist (remembering it’s the users of ‘Islamophobia’ that conflate it with racism). And the problem there is that you need a wider context of any tweet and its author – and the Demos study does not deal with this, at all. It is flawed.

So, ‘that it is a fundamental injunction of Islam for all of its followers to be engaged in a violent struggle against non-Muslims and the west‘ is a claim that can be argued, for or against; it is one understood as part of Islam by many Muslims and non-Muslims, and does not count as a phobia of any kind, and most certainly does not count as anti-Muslim bigotry.

The second category [i.e. that beginning “and (similar) especially”] is also flawed in the way Carl puts it, though that might be a simply syntactic error. I have never come across anyone that thinks the blame for terrorist acts lies ONLY with the ‘Muslim population generally’ and NOT ‘ the terrorists themselves, or on Islamist militancy’ – but let’s suppose the category was meant to read as inclusive – i.e. ‘not ONLY on the terrorists themselves, or on Islamist militancy’…

There is still a non-bigoted sense in which it is a legitimate claim – in fact the claim could be expanded to include apologist pseudo-liberal non-Muslims.

The term ‘generally’ – i.e. over generalisation – is not just a problem for people blaming ‘Muslims’ (implied generality), but for people like Carl reporting on the use of the term ‘Muslims’, as if the writer meant ALL Muslims. This is a well known ploy among critics of atheists that Carl ought to be aware of…

If someone says “ALL Muslims do X …“, then the statement clearly applies to all Muslims. But often, to avoid lengthy prefixes and caveats (it is 140 character Twitter after all) a claim will be made that “Muslims do X …“, but when some context is accounted for, such as surrounding tweets in a debate, it means “Those Muslims that do X (but not those Muslims that don’t, and not non-Muslims) …

This is also a well trodden path online, with characters like Glenn Greenwald, Reza Alslan, C J Werleman, Cenk Uygur, and many others, making false assertions about what some atheist (i.e. New Atheist; i.e. Sam Harris) has said. They regular interpret an implied ‘some‘ to mean ‘all‘, and the Demos study does this too.

Want a real example of how factual speech can be perceived as ‘Islamophobia’? Look no further than Catrin Nye’s report of her interview with Ruqaiya Haris [see BBC programme] that was rudely interrupted by some guy. He was out of order interrupting an interview – but what he said wasn’t anti-Muslim bigotry but an expression (however limitedly informed) of his dislike of Islam’s Sharia law. Attempts are made to implement Sharia in the UK, and this is well documented, just not universally applied or desired among all Muslims, since many don’t want it. And, while we’re on that example, look how aggreessively Nye and Haris go for the guy – more of this over-sensitivity and misinterpretation in another post.

There are countless examples of ‘Islamophobia’ as hate speech being presumed when disagreement about Islam is expressed. I’ll remind you again, EVEN HATING a political-judicial-religious ideology is NOT HATE SPEECH aimed at Muslims.


Back to Carl’s response. Let’s look at his examples for this definition of ‘Islamophobia’ (my addition of a-g labels):

Examples here include:

a) @xxx @xxx Islam allows you to do anything to infidels. In fact it encouraged jihad to pay for your sins

b) 1.5 billion Muslims Running 50 countries. They only do jihad terror becs they are marginalised. Give them another country.

c) @xxx There are 6 million muzzies in France. If only 10% of them are animals (we know its more) That’s 600,000 savage muzzie killers.

d) I feel sorry for the true people of London the Muslim take over has begun we must not let it go any further dame scum @XXX

e) Send these fucking muzzie scumbags back to sandholes they belong it and then bury them 4 good, those murdering pricks!

f) There are no moderate Muslims – Every Muslim plays a role in civilizational jihad from infiltrating government to crashing welfare programs.

g) There are no peaceful Muslims: there are only ones that have not been called to jihad yet

I searched for these tweets again.

(a) Not found. However, that is a claim about Islam that could be disputed by reference to the Islamic texts. It is not blaming all Muslims ‘generally’ for anything. As a supporter of reformist and liberal Muslims I would be comfortable making this statement if I thought it to be true – I actually think it’s partly true.

(b) Not found.  This looks like a sarcastic remark made in response to either comments made about the persecution of  Muslims by the lone state of Israel, or persecution of minority status of Muslims. A remark like this is often used to emphasise the fact that in parts of the world there are states so dominated by Islam, such that they have actually or nearly driven out other religious sects, or oppress them, that for Muslims to claim to be persecuted misses the point.

This is similar to the Hijab debate mentiond at the beginning of this post.

This is precisely how crazy this whole issue is, and why the Demos data is very likely to be so far skewed, given Carl’s examples, that one has to wonder how many tweets in support of Iranian Muslims are actually being classified as ‘Islamophobia’.

c) Not found. This isn’t made in specifically nice terms, but I’d dispute that this falls under ‘anti-Muslim bigotry’ (and hence not under the misapplied ‘Islamophobia’). Pew polls are very clear on the degree to which ‘moderate’ Muslims support death for apostasy and homosexuality, so the main point of this tweet is a fact claim about the opinions of Muslims. And given that many Muslims are prepared to call ISIS ‘animals’, and given that the ‘animals’ in this tweet refers to the 10% that believe in some of the darker demands of Islam, this tweet isn’t expressing an overly generalised view of Muslims. The reference to France is probably tied to the fact that there are parts of towns like Paris where there is enough sympathy for Islamism at least, and terrorists even, to give some grounds to this tweet in principle if not in the detail of the claim. Why is this tweet an example of ‘Islamophobia’ if Carl is not conflating fair criticism with hate?

Bear in mind our (UK) very own Anjem Choudary has only just been jailed, and the fears of him sustaining his external network of radicalised Muslims, and his capacity to radicalise more on the inside, are real probems and not some ‘Islamophobic’ fantasy.

d, e) Not found. These I can accept are anti-Muslim, specifically and applied generally.

f, g) Not found. These are claims about Muslims based on an understanding of Islam from a perspective that includes some interpretation of Taqiyya. Now the tweets may be incorrect, but they are truth claims about Islam, and the consequence for what could be expected of ANY Muslim – the point being that there is no way of knowing what any individual truly believes. Add to that the factual examples of preachers in mosques having been recorded prescribing such behaviour as entryism, this tweet amounts to no more than an exaggeration – more like a conspiracy theory. This could actually be classed as ‘Islamophobia’ – as an irrational fear that every follower of Islam is intent on seeing Islam implemented everywhere. It’s as much delusional as bigoted.

Other Definitions of Islamophobia

Carl gives some background, which includes the foresight of having identified the problems with the term early on. And yet Demos intentionally used it.

For our research here, the most important distinctions the Commission made between an Islamophobic expression from a legitimate criticism of Islam are:

  • Seeing Islam as a monolith. This is the view that Islam is undifferentiated and static; the use of sweeping generalisations that ignore the tensions, disagreements, debates and profound differences that exist within community (or rather group of communities). This is especially important for the research that we did, because one of the kinds of Islamophobia we identified was moving from the general to the particular, where a terrorist attack is seen as either an illustrative example, or proof, to condemn all Muslims generally (under the assumption that they’re all the same).
  • See Islam, as a faith, as an aggressive enemy. This would constitute not the view that opposes theological militancy, Islamism, or particular brands of Islamism, or ideologues within particular brands of Islamism. However, it would include the worldview that Islam, writ-large and across the board, regardless of the different interpretations and traditions it contains, is inherently committed to terrorism, violence, and an implacable hostility to the Western world. Expressions included here were those that claimed that Muslims supported terrorism and terrorists not because of anything that they had said or done, supported or condemned, but because they were Muslim.

Monolithic Islam

The problem with seeing Islam as a monolith isn’t one of the making of critics of Islam but of Muslims, and as such it is unreasonable to classify this as ‘Islamophobic’. See the earlier comments on ‘Islamophobia’ at the start of this post.

The monolith of Islam is expressed most forcefully in the way so many Muslims from around the world are radicalised and have joined ISIS. These are very disparate people, from many nations, using different languages, many born in those countries – but Islam binds them together as Muslims. Their monolith may not be the same as that of other Muslims, because of course Takfir comes into play. So we end up with the rediculous state of affairs where all Muslims are Muslims – oh, except those other guys that might als claim to be Muslims but aren’t really proper Muslims, so they’re apostates.

The monolith of Islam pervades articles that oppose any intervention by non-Muslims states in ‘Muslim lands’; and ‘Muslim lands’ is a monolith of Islam term used by Muslims. While most western non-Muslims see Saudi, Qatar, Syria, Iraq, Egypt and other places in the Middle East as lands where Islam dominates many of us are aware of the variations in governments. But fair enough, many non-Muslim people that don’t know much about the wider world think of these lands as ‘Muslim lands’, but then that’s not surprising, since many Muslims make the point that these are ‘Muslim lands’.

Any impression of a monolithic Islam is entirely the doing of Muslims. Look to the vocal ones that self-identify as spokespeople for Islam, the various groups like CAIR, 5Pillars and others – check their twitter accounts and you’ll see no end of the Islamic monolith – every time some issue arises about Muslims or Islam, they’ll tweet it.

As such, I don’t think you can treat tweets that respond to such claims to the monolith of Islam as ‘Islamophobic’, for buying what’s on sale. It’s not the critic that’s declaring the monolithic nature of Islam, but the people he’s responding to.

Does the Demos study deal with this complexity? No it does not.

Islam as an Aggressive Enemy

This is another problem definition, if it’s being used by Demos. The distinction it makes is too fine:

“This would constitute not the view that opposes theological militancy, Islamism, or particular brands of Islamism, or ideologues within particular brands of Islamism. However, it would include the worldview that Islam, writ-large and across the board, regardless of the different interpretations and traditions …”

I really can’t see how you’re going to determine that distinction from the content of a tweet that it is applying the tweeted statement globally or specifically. This is a massive problem with this study if Carl thinks it can determine this sort of distinction.

To be clear, in our research a criticism of Islam was not, by definition, taken to be Islamophobic, and was not understood by the researchers of our report to be. However, one that treated Islam as a single, hermetically sealed-off block, and one that, en bloc, is violently hostile to the West would be.

This is nonsense. Twitter is very cryptic. I’ve engaged in so many debates on Twitter only to find three or four or more tweets in you and you’re interlocutor are both ‘on the same side’ – you’ve talked right past each other. Other occasions I’ve seen the very same worlds used in these contexts:

  • For X
  • Opposed to X
  • Sarcastic opposition to X
  • Sarcastic opposition to opposers of X

THIS IS TWITTER! Unless the Demos study analyses the account of the tweet and the interaction with surrounding tweets, none of this complexity will be accounted for and tweets that are not anti-Muslim bogotry but which are criticism of Islam will be included.

Ideas v People

Finally, what about the mixing together of the criticism of ideas with the criticism of people? First, this is what concepts do; they bring together things that can be disparate in order to make sense of them.

You’ve got to be kidding me. What on earth do you think ‘Islamophobia’ does as a term? That’s precisely what it does, conflates criticism of a religion with persecution of people.

And you know who makes the most of that conflation? Muslims (Ooops! Pseudo-liberal alert: SOME Muslims). Because so many of them are so offended personally when their ideology – sorry, their religion – is criticised.

And they are aided and abeted in this misapprehension by people like Carl Miller, who are quite happy to acknowledge the complexity of the meaning of words, but then appear on a BBC News programme and make the most of the misapplication of a word:

There’s obviously a big moral gulf between a racist Tweet, and a racist murder, yet both fit within the concept of racism. For exactly this reason, we weren’t trying to draw any kind of moral equivalence between the different kinds of things that we analysed. How could we?

Well you do. By omission. You failed totally in your TV interview and tweets to give any of the caveats that pervade your web site and report [oh, until the end of the interview, where you basically said the study was worthless – Jeez].

Carl DOES Conflate Anti-Muslim Hate And Criticism of Islam on the BBC

First, from Carl’s latest post:

The research used structures of algorithms to filter tens of millions of Tweets into hundreds of thousands. Within this data, there will be Tweets of radically different kinds of seriousness and concern. We weren’t trying to say that all Islamophobia is equal, and that all these Tweets have some kind of equal moral standing.

But you did. That’s exactly what you did do, with the co-operation of Catrin Nye, Victoria Derbyshire, Ruqaiya Haris  and Ajmal Masroor.

So, now from the BBC programme, Carl said:

We’ve been seeing five thousand tweets every single day which are anti-Islamic, seriously derogatory and hateful being sent in the English language across the world. And it’s been building month on month since May, with July as the wrost so far with just under seven thousand tweets, judged as anti-Islamic every single day.

Note that this is now anti-Islamic. So, criticism of Islam is deemed Islamophobic. Clearly Carl is conflating criticism of Islam with hate speech toward Muslims. Note also the overlaid text on screen: ‘Islamophobic’ – not Carl’s fault, but I’ve not seen him complain about that while pushing the programme on Twitter.

Just remember that ‘across the world’ moment when he gets around to agreeing that it’s partly Brexit that’s to blame.

When referring to the post-Nice spike in Nye’s piece:

That’s right. Angry abusive anti-Islamic; all happening in the immediate aftermath of Nice.

Carl is interviewed live on the panel and explains the analysis process, in very simple terms that give no indication of the complexity of the tools used, other than to say they are sophisticated, giving a false impression of their suitability for this task. What’s more  he goes on:

I think it [the results] shows underneath all the amazing brilliant things that we’re so used to seeing on Twitter [LOL] there is actually a problematic and worrying under layer of hate and Islamophobic hate. And this is something that we as a society have to confront, I think.

There you have in such a short space two things. One is the actuality of the dishonesty of Carl’s explanatory post on Demos. He really is conflating hate and criticism. The other is his political agenda – there is no researcher’s unbiased impartiality, since Demos is quite clear in its agenda. Carl has something to say and he’s made his figures say it.

When Victoria Derbyshire brings up the Brexit effect Carl goes on to say:

It began to legitamise the airing of certain views that until that moment some people couldn’t talk about. So we have to realise actually that there are a lot of views we thought to have been confronted and indeed defeatedIslamophobia online is very event specific.

What is it about the criticism of Islam, being anti-Islam, or any system of ideas does Carl think should have been defeated? It’s even more clear here, given that there was some post-Brixit hate spiked against all immigrants, not just Muslims, that there is this clear conflation, linked by the word Islamophobia:

hate -> anti-Muslim hate -> Islamophobia -> criticism of Islam

Carl links them very clearly.

Of course the really worrying and really problematic thing here is, and we can see this in the Twitter response to this very programme [the Nye programme, or the Derbyshire programme?] is that there is no distinction in many peoples’ minds between a legitimate criticism of Islam and Islamophobia. They regard the condemnation of Islamophobia to be the shutting down of legitimate criticism of Islam. And they think that anything they say that is Islamophobic is a legitimate criticism.

And this is a very important point of contention critics of the Demos data have.

  • The use of the term ‘Islamophobia’ does conflate them anyway.
  • Many Muslims offended by criticism of Islam call it Islamophobia – as Ajmal Masroor did [there are a whole bunch of problems with his comments on the Derbyshire programme].
  • Some people have genuine criticisms of Islam AND hate Muslims AND are racist and the Demos study isn’t going to differentiate them.
  • Carl’s contribution to the Derbyshire and Nyre programmes conflates them very clearly, explicitly.

So, YES, condemnation of Carl’s use of ‘Islamophobia’ IS shutting down legitimate criticism of Islam. Because if you already deem legitimate criticism of Islam is Islamophobic, then UNDER YOUR USE OF THE TERM, we will necessarily be being Islamophobic when we criticise Islam legitimately.

Let me spell this out with a metaphorical example:

A: Those rules of Rugby are stupid.

B: Rugbyphobe! Not all rugby players are stupid. Why do you hate all rugby players because a minority are stupid?

A: I said some of the rules are stupid. I didn’t actually say any rugby players are stupid.

B: But rugby players love and follow the sacrosanct rules of rugby, you Rugbyphobe!

A: Yeah, but some of the rules are stupid. Look at this one.

B: That’s Rugby League! That’s not proper Rugby! Why do you presume that all rugby players follow the same rules? They aren’t rugby players anyway. Hey, Carl, what do you think?

Carl: Mmmm. It seems any derogatory and hateful Rugbyphobic thing A says, he thinks is legitimate critism of Rugby. He seems to be conflating hate for rugby players with legitimate criticism of Islam.

… somewhere in rugby heaven …

Bill McLaren: There is no legitimate criticism of Rugby!

Nothing in what Demos has produced so far gives any indication that they can reliably recognise and exclude legitimate criticism of Islam from their results.

Carl’s Nonsense On Dangerous Narratives

There’s more from Carl that amounts to absolute nonsense:

Now this is wrong and the only thing this plays into, this Islamophobia, is the narratives that Islamic State wish British Muslims to hear. None of this makes us any safer. None of this makes terrorist attacks less likely to happen. All it does is play into that narrative.

So, Carl is a terrorist psychologist now? This is utter wishful thinking.

  • Racist hateful remarks, post Brexit, against other EU nationals has not increased terror attacks from Poles.
  • A history of racist hateful remarks against Afro-Caribbeans, Hindus and Sikhs or Jews, has not caused them to increase terror attacks.
  • The one identifiable group that has the notion of Jihad and conflict with un-believers, Christinas and Jews written in its core book, the Quran is Islam. Pointing this out with legitimate cricism of Islam or even actual hate speech towards Muslims isn’t going to make it any worse.
  • Grand appeasement of multi-culturalism and sweet talking and the avoiding of criticism, the language that has “confronted and indeed defeated”, is EXACTLY the condition within which Anjem Choudary has flourished and built his network. And within which Muslim child grooming groups went unchallenged for the political expediency of not being racist BECAUSE people like Carl have conflated criticism of Islam with hate speech and racism.
  • And on top of that it was spelt out very clearly for Carl and many others by Islamic State themselves in issue 15 of Dabiq, their magazine – ‘Why We hate You’. They said clearly that nothing we could do would stop them wanting to kill those that oppose them.

Online Civility and Carl’s Agenda

At 10:45 into the Derbyshire programme Carl gets to sell his agenda. It is entirely about the wider issue of the way people interact online. So much for the impartial researcher that wants to bring us unbiased data.

This particular study by Demos is very specifically biased, to such an extent it ignores the very concerns Carl raises about conflating criticism of Islam with hate speech.

The Demos Data and Twitter

So this brings us to the actual Demos analysis itself.

July 2016 – The most signiEicant increase was in the immediate wake of the terrorist attack in Nice, on July 14th, with another appreciable increase in the rate of anti-islamic  expressions in the aftermath of the killing of Jacques Hamel in Normandy. The Eive most significant spikes are analysed in greater depth, below. This is to try to uncover the triggers, drivers and dynamics of anti-Islamic hatred online.

Conflation right out of the gate. Is the study intended to look at anti-Muslim hatred, or criticism of Islam? Because ‘anti-Islamic hatred’ could be perfectly legitimate hatred of an ideology – just as a hatred of fascism might be.

And here of course we have a problem, because criticising, disliking, and, heaven forbid, hating Islam, as a political-judicial-religious system (as it factually is, even if not lived that way by many Muslims), is guaranteed to wind up most Muslims.

Ahmadi Muslims, the most peaceful and even pacifist of Muslims are often deeply offended when their religion is criticised. This is nothing more than the Offence Defence, and it is used very effectively, and very intentionally, by many Islamic groups and individual Muslims – as Ajmal Masroor does in the Derbyshire programme. The very problem that is presented in Benjamin’s NSS response is there in abundance in the nonsense that Ajmal Masoor comes out with.

We can look at some of the examples the Demos report cites as problematic (already covered by Benjamin Jones):

Morocco deletes a whole section of the Koran from school curriculum as it’s full of jihad incitement and violence. The Religion of peace.

A factual claim about an event, and fair criticism of Islam and the way in which the violent content of the Quran is ignored or hidden.

Sorry to hear about france- These muzzies just dont quit.

This isn’t necessarily a hateful tweet. The derogatory term ‘muzzies’ is used, but the context is important, and we don’t know it. It depends on the person tweeting it, and the content of other tweets around it. It could very specifically be referring to the terrorists, as ‘these’ Muslims.

It’s even possible that such a tweet could have been posted by a Muslim as an indication of what he is expecting from a non-Muslim, as sarcasm or a prediction of what’s about to appear in his twitter stream. This is the nature of Twitter. Does the Demos program account for that? No.

Carl’s Caveats

Back in the Demos report:

The research of large social media datasets is a reasonably new undertaking. It is important to set out a series of caveats related to the research methodology that the results must be understood in the light of:

  • The algorithms used are not perfect: throughout the report, some of the data will be misclassified. The technology used to analyse Tweets is inherently probabilistic, and none of the algorithms trained and used to produce the Eindings for this paper were 100% accurate. The accuracy of all algorithms used in the report are clearly set out in this report.
  • Some data will be missed: Acquiring Tweets on the basis of the keywords that they contain presents two possible problems. First, the initial dataset may contain Tweets that are irrelevant to the thing being studied. Secondly, it may miss Tweets that are relevant to the thing being studied. Researchers worked to construct as comprehensive a list of keywords as possible (these are detailed in the report, below), however it is likely some were missed, and the numbers presented in this report are likely a subset of the total.
  • Twitter is not a representative window into British society: Twitter is not evenly used by all parts of British society. It tends to be used by groups that are younger, more socio-economically privileged and more urban. Additionally, the poorest, most marginalised and most vulnerable groups of society are least represented on Twitter; an issue especially important when studying the prevalence of xenophobia, Islamophobia and the reporting of hate incidents.
  • Overall, this research is intended to be an indicative, Eirst-take of the reaction on Twitter to these important events. It is not presented as either exhaustive or deEinitive; and it is very much hoped that it will stimulate further research on this vital topic in the future.

The algorithms aren’t really the problem, except in as much as they may not be providing the analysis that’s needed in this case.

I’m a software developer and I have some old background in AI research – no, really old. I appreciate many of the issues, but I have no particular experience of the systems used. However, there are some general issues about the applicability of this sort of software.

It might be very good at contributing to marketing analysis on Twitter, establishing whether tweets are positive or negative towards a product when it is mentioned. But this isn’t about a simple product. It’s about a complex social problem where the use of words and context are incredibly inter-related, where a tweet can superficially mean one thing, but very often might mean the exact opposite; where the same tweet by two different people maight mean the opposite because of who is doing the tweeing.

So, a tweet could be meant literally by a racist white supremacist; or might be written sarcastically  by a Muslim responding to another tweet:

A: “All Muslms are potential terrorists”

Muslim: “LOL. Let’s kill all muzzies”

It’s not at all clear that the Demos study captures anything like these differences, or that it could. It performs no analysis of the person tweeting, their tweet history, which gives a clue to their overall stance on Islam and Muslims. It performs no analysis of the conversation in which the tweet is made, or who it is directed at.

There is a failure to distinguish the use of the term ‘Islamophobia’ – Is the conflation fault that of the racists, or a fault of people like Carl? I really don’t as yet have any confidence that this study is at all accurate in distinguishing hate of Muslim from criticism of Islam, because Carl doesn’t appear to understand the nature of the complexities – his understanding seems to be superficial, and wrong.

There are plenty of tweets I’m sure it caught that are clearly racist, “Go home Paki”, but which give no indication of any relationship to religion, and specifically Islam, so are bigoted, racist, but not ‘Islamophobic’, even by Carls use of the term.

Note that if this is the standard then a poorly written tweet by a Muslim who does not have a good grasp of English could be included: “Die White MF hate Muslim hate Islam!” The key words are there, but the intended “hater of Muslims, hater of Islam” is poorly expressed.

This is a minefield of potential problems with the use of the term Islamophobia, that will do all of the following:

  • It will attribute criticism of Islam as Islamophobia
  • It will attribute the term Islamophobia to racist ethnic slurs, towards only some ethnicities
  • It will misunderstand the contextual meaning and count many false positives.


Potential False Positive Tweets

I’d be interested to know if this would be included, had the trainers not seen it. How representative of all tweets analysed are the training tweets? Note that the point here isn’t that the trainers could have used this as a negative example, but that had they not seen it, it could have been analysed by the system as a positive example.

There’s so much variety in the language used, and much of it is critptic, and much by people that don’t have English as a first language. Natrual language processing is difficult enough. Is it capable with dealing with this variety?

Even If the Training and Analysis is Accurate

With regard to the study, imagine that:

  • All the issues with training the system and the analysis were correctly distinguishing racist anti-Muslim bigotry from criticism of Islam.
  • All generalised racist tweets that were not specific about Mislims, or didn’t definitely imply Muslims (Indians often call Pakistanis Paki, and some Pakistanis will use the term themselves) were excluded (they were not).
  • That critics accept that if a tweet is both critical of Islam and contains racist terms, it is valid to include it.
  • And as a result, the report was a fair reflection of anti-Muslim bigotry and excluded all tweets that werre genuinely critical of Islam without implying ALL Muslims (an impossible feat since becasue of the Specificity Problem).

Imagine also these more far fetched conditions, that do not in fact apply to the Demos study (as far as I am aware):

  • The analysis excluded all sarcastic tweets X by Muslims and opponents to the persecution of Muslims, aimed at countering geniune bigoted tweets Y.
  • The analyis excluded all sarcastic tweets X by Muslims and opponents to the persecution of Muslims, aimed at countering geniune criticsm of Islam in tweet Y, because the authors of X conflate such criticism with Islamophobia.

For example, would a tweet containing ‘Porch monkey Muslim’ be treated as Islamophobic? Only this phrase is used by Muslims and non-Muslim apologists for Islam to attack Muslims like Maajid Nawaz and other reformists. Because if such slurs are included, then this counts as a point in the Demos Islamophobia 7,000 a month.

Even if everything else is analysed to the satisfaction of critics of the term ‘Islamophobia’, so that the result does represent the trends in anti-Muslim bigotry, particularly around events like Nice, what are the trends in related data?

If other trending sentiments aren’t analysed then it’s not telling us enough. If there are spikes in anti-Muslim bigotry around the Nice incident, and the same increase in genuine criticism of Islam, because the Nice event is understood to be an Islamic terrorist event*, then this spike is nothing more than a response to Islamic terror and isn’t telling us much about the overall level of hate towards Muslims in the UK, or the English speaking world. The spikes are merely a rise in all things Twitter around an event.

[*There’s some dispute about whether it’s classified as ‘Islamic terrorist incident’ (and not just because of  ) See discussion here.]

What if there’s a spike in support for ISIS too, around these events?

All we’re seeing is a more vocal response to an event. That is NOT an increase in Islamophobia (or anti-Muslim bigotry), but merely a more pertinent vocalisation of it related to a specific event. It does not mean there is a an actual increase in hate for Muslims.

Other Issues Not Mentioned

To what extent does the analysis account for multiple tweets from the same account? It’s quite feasible that the same account could be the source of many tens of tweets a day, especially following an incident like Nice. Are they to be counted as individual tweets? If you think this is insignificant bear in mind the extent to which, between them Carl Miller, Victoria Derbyshire, Catrin Nye, Ruqaiya Haris , have tweeted and retweeted about this, and how many quoted follow ups there have been.

A troll will often post the same tweet content, or minor variteies of it, with different target Twitter accounts in the tweet:

“@x @y I hate all muzzies”

“I hate muzzies @z”

Is this one hate incident or two, in the Demos report?

Does the report distinguish tweets from retweets and quoted tweets?

It seems quite possible that if these issues are accounted for the number of ‘incidents’ could eaily reduce ten fold or more.

Remember, this Demos research and the BBC programmes around it amount to a specific issue that’s trending, being criticised and approved – does this show an increase Islamophobia? Only that is exactly what was claimed about the Nye programme. When social media responds to an event, is that a change in the trend of feeling about the event, or merely a change in the reporting of feeling that’s already there?


The actual data analysis method, Method52, is not being questiond here. But there are genuine issues with the following:

The Demos report does not show that Method52 is the right tool for to deal with the complexities that require accounting for: who wrote the tweet, who it targeted, the context of the conversation, multiple tweets, duplicate tweets with minor variations, or even exact duplicate tweets.

The problems around the meaning of the term ‘Islamophobia’, the conflation of criticism of Islam and hate for Muslims and general racism, is acknowledged, but only as a problem for people tweeting multiple sentiments (fair criticism and hate). It is not acknowledged sufficiently as a problem for the researchers and trainers of the system who, given the few examples provided, are classifying fair criticism as Islamophobia.

The public presentation on the BBC Victoria Derbyshore programme very clear has Carl conflating hate and fair criticism, and explictly saying anti-Islam tweets are derogartory ‘Islamophobia’.

The conflation of criticism of Islam with hate for Muslims is one so firmly held with many Muslim communities that Muslims themsleves construct a ‘monolith of Islam’, so that anyone responding to that monolith is then deemed to be Islamophobic in the conflated sense. The Demos study cannot deal with this.

The whole Demos studdy is at the very least highly suspect; and it’s presentation by Carl Miller, and Catrin Nye, is grossly misleading and is contributing to the divisions in society they claim they want to avoid.

It’s odd that Carl thinks that criticisng Islam robustly is not helping, and yet conflating that criticism with hate speech, as derogarory anti-Islam Islamophobia is?

One thought on “Carl Miller of Demos Still Misfires on ‘Islamophobia’

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s