Tentative thoughts on which kinds of speech are harmful

TLDR: as­sum­ing that peo­ple are open to polic­ing harm­ful speech, what kinds of speech should they ac­tu­ally choose to sup­press? It seems to me that com­mon tropes about harm­ful speech are just partly cor­rect. Speech-cops should con­tinue to sup­press prej­u­dice, but they should be more cau­tious about polic­ing some ideas which are only vaguely as­so­ci­ated with prej­u­dice, and they should be more will­ing to po­lice speech that con­tra­dicts some key tenets of the EA com­mu­nity.

Note: I haven’t per­son­ally done any se­ri­ous kind of speech-polic­ing.


There seems to be in­creas­ing in­ter­est and ca­pa­bil­ity in the West for polic­ing speech in an at­tempt to sup­press harm­ful at­ti­tudes in so­ciety. Speech can be po­liced ei­ther through le­gal means (e.g. hate speech laws) or via pri­vate ac­tion (protests, firing, milk­shak­ing, on­line cen­sor­ship, se­lec­tive pub­li­ca­tion of re­search, etc). For this post I will ig­nore the dis­tinc­tion and treat speech polic­ing as one is­sue. Also, I won’t ad­dress whether speech polic­ing is ac­tu­ally good or bad.

In­stead, I will fo­cus on the spe­cific choice of which kinds of speech are most harm­ful, i.e. best to po­lice. Speech polic­ing has been ad­vo­cated and performed against a va­ri­ety of differ­ent ideas, of­ten in very differ­ent par­ti­san di­rec­tions. And there are few good ar­gu­ments about which kinds of speech are more harm­ful or benefi­cial than oth­ers. As­sum­ing that one were to po­lice harm­ful speech, which at­ti­tudes most de­serve to be sup­pressed? Most speech-po­licers don’t seem to put the proper amount of care into con­sid­er­ing this ques­tion.

It should go with­out say­ing that in­for­ma­tion should only be sup­pressed among cer­tain kinds of un­eth­i­cal peo­ple, if we can be so pre­cise. The worry with harm­ful speech is al­most uni­ver­sally that it can in­spire un­eth­i­cal peo­ple to take bad ac­tions. An eth­i­cal per­son, by defi­ni­tion, will only use in­for­ma­tion to do eth­i­cal things. For in­stance, if you tell an Effec­tive Altru­ist that im­mi­grants will lower do­mes­tic wages, she will shrug her shoulders and say that the needs of the many out­weigh the needs of the few, un­less it re­ally makes im­mi­gra­tion a bad idea all-things-con­sid­ered; on the other hand, if you tell a nar­row-minded na­tivist that im­mi­grants will lower do­mes­tic wages, that will mo­ti­vate him to be more strongly xeno­pho­bic. When some­one is un­eth­i­cal, we seek to con­trol them; when some­one is eth­i­cal, we seek to em­power them. In­for­ma­tion shar­ing is em­pow­er­ment; in­for­ma­tion de­nial is con­trol. For this rea­son, we must ab­solutely re­ject speech polic­ing and ly­ing to EA com­mu­nity mem­bers. In­stead, I am just hy­poth­e­siz­ing about speech polic­ing that might be pur­sued on col­lege cam­puses, on gen­eral so­cial me­dia, etc.

Note that speech polic­ing is very con­tro­ver­sial, with some ar­gu­ing for a clas­si­cally liberal policy of re­straint and tol­er­ance. I won’t di­rectly ad­dress this con­tro­versy here. I don’t re­ally have a stance on it, and I’m not ar­gu­ing in fa­vor of speech polic­ing. But no mat­ter what your view is, you must re­al­ize that speech polic­ing in­curs ob­jec­tive costs. It takes time and effort, it cre­ates con­tro­versy and grievance, it causes peo­ple to be fired from pro­duc­tive la­bor, it makes vic­tims feel un­happy and op­pressed, etc. Th­ese down­sides might be out­weighed by the benefits of re­strict­ing speech, but they are still real and can­not be ig­nored. There­fore, merely be­ing cer­tain that a par­tic­u­lar kind of speech is harm­ful is not suffi­cient to jus­tify sup­press­ing it. The speech has to be more harm­ful than a thresh­old in or­der to out­weigh the prac­ti­cal harms im­posed by speech polic­ing. I don’t know what that thresh­old is, but it is still an im­por­tant prin­ci­ple to keep in mind when de­cid­ing which kinds of speech are bet­ter or worse can­di­dates for polic­ing.

Another thing to re­mem­ber is the risk of im­plicit cen­sor­ship by as­so­ci­a­tion. If peo­ple who be­lieve X also dis­pro­por­tionately be­lieve Y, then cen­sor­ing X-be­liev­ers can sup­press Y from the pub­lic dis­course.

Keep­ing these is­sues in mind, here are my pre­limi­nary thoughts on which spe­cific kinds of speech are more or less harm­ful. They are in­formed by the con­clu­sions of the Can­di­date Scor­ing Sys­tem, which not only rates spe­cific can­di­dates but also serves as a gen­eral com­pendium of poli­ti­cal judg­ments and pri­ori­ties. If you dis­agree or are con­fused by poli­ti­cal as­sump­tions made in this post, you may find an­swers in a rele­vant sec­tion of the Can­di­date Scor­ing Sys­tem, speci­fi­cally the “policy po­si­tions” and “weights” for rele­vant is­sues. How­ever, some of my views in this post are based on per­sonal judg­ments that are not writ­ten into CSS.

Harm­ful speech

Calls to violence

This one is pretty straight­for­ward. Agi­tat­ing for kil­lings or other vi­o­lence against minori­ties, poli­ti­cal op­po­nents, cops, land­lords, etc can only do harm. An ex­cep­tion would be calls for prop­erly or­ches­trated state vi­o­lence against threat­en­ing ac­tors, such as rou­tine polic­ing and strikes against ter­ror­ists, which are much more likely to have good con­se­quences.


Speciesism, defined as not car­ing about an­i­mals merely on the ba­sis of their mem­ber­ship in a differ­ent species, is clearly wrong and harm­ful.

In re­al­ity, per­haps the ma­jor­ity of “speciesism” is not like this. In­stead, peo­ple be­lieve that an­i­mals de­serve less moral sta­tus by dint of be­ing un­in­tel­li­gent, or be­ing in­tended to serve hu­man­ity, or some other rea­son. But these at­ti­tudes too are good can­di­dates for speech polic­ing.

Vi­cious racism

Racism is tough to define, but a core type (dis­cussed by philoso­pher Jorge Gar­cia) is spe­cific dis­re­gard for the in­ter­ests of peo­ple of other races. This is straight­for­wardly wrong, and its po­ten­tial harm is demon­strated by a va­ri­ety of regimes and non­state ac­tors, both his­tor­i­cal and con­tem­po­rary, who have pur­sued geno­cide, eth­nic cleans­ing, bi­ased polic­ing, harsh sen­tenc­ing, im­mi­gra­tion re­stric­tions, zon­ing re­stric­tions, and other bad ac­tions. Vi­cious racism is not the sole cause of racially de­struc­tive poli­cies, but it pre­sum­ably plays a ma­jor role.

Vi­cious racists don’t seem to uniquely hold other in­sights or valuable ideas, so the risk of im­plicit cen­sor­ship is min­i­mal.

White nationalism

None of the poli­cies fa­vored by white na­tion­al­ists seem to be benefi­cial, and they pose a risk for com­mit­ting ma­jor harms like mass de­por­ta­tions. In ad­di­tion, they don’t seem to pos­sess any unique body of use­ful ideas that liberal so­ciety ought to take into con­sid­er­a­tion.

Trans-ex­clu­sive speech

Claims that trans­gen­der peo­ple are not their tran­si­tioned gen­der, or that they are all men­tally ill, or re­fusals to use their cho­sen pro­nouns, can cre­ate stigma that re­duces their so­cial ac­cep­tance. It re­duces the prospects for an­tidis­crim­i­na­tion laws. It re­duces the prospects for pri­vate and pub­lic fund­ing of tran­si­tion ther­a­pies. It re­duces the prospects for chil­dren with gen­der dys­pho­ria to get any kind of sup­port. It has fur­ther been alleged to in­crease their risk of be­ing mur­dered, but as far as I can tell, there is for­tu­nately no sig­nifi­cant ev­i­dence to sub­stan­ti­ate the idea that there is an epi­demic of hate-mo­ti­vated trans mur­ders.

The biolog­i­cal or philo­soph­i­cal re­al­ity of gen­der and sex is re­ally not rele­vant here: even if all the trans­gen­der the­ory were wrong, that wouldn’t change the fac­tual re­al­ity that af­firm­ing trans­gen­der sta­tus is so­cially benefi­cial.

The risk of im­plicit cen­sor­ship is mild here. There are some sig­nifi­cant con­cerns about peo­ple (es­pe­cially chil­dren) be­ing en­couraged or pres­sured too much into ir­re­versible or difficult-to-re­verse gen­der tran­si­tions, and the broad sup­pres­sion of gen­der-crit­i­cal peo­ple or TERFs already seems to have in­terfered with de­bates about this is­sue; how­ever health and le­gal pro­fes­sion­als can do a fairly good job of work­ing through this on their own. It’s OK if we are a tad over-en­thu­si­as­tic in our early so­cial at­ti­tudes to­wards gen­der tran­si­tions, we will learn more in short or­der. There are con­cerns about women’s sports, as the in­tro­duc­tion of a small num­ber of markedly biolog­i­cally su­pe­rior perform­ers can re­duce com­pet­i­tive­ness and per­ceived fair­ness for the ma­jor­ity of fe­male ath­letes (po­ten­tially out­weigh­ing the benefits of in­clu­sion for a small num­ber of trans­gen­der ath­letes); how­ever this is sim­ply not some­thing that se­ri­ously af­fects a ma­jor part of the pop­u­la­tion. In any case, these con­cerns are perfectly well voiced by peo­ple who are ba­si­cally fine with trans­gen­der iden­tity and ac­cep­tance. It isn’t use­ful to bring ac­tual trans­pho­bes to add to the cho­rus of what is be­ing voiced more re­spect­fully and elo­quently by main­stream liberal ac­tors (al­though some of these main­stream liberal ac­tors are them­selves re­garded, in­cor­rectly in my opinion, as trans­pho­bic in their own right).

So over­all, it seems pretty straight­for­ward that truly trans-ex­clu­sive or trans­pho­bic speech is harm­ful. On the other hand, just 0.6% of the pop­u­la­tion iden­ti­fies as trans­gen­der and a good deal of ther­apy fund­ing and trans-af­firm­ing com­mu­ni­ties and in­sti­tu­tions already ex­ist, so such speech may not always be harm­ful enough to out­weigh the ob­jec­tive costs of cen­sor­ship. For in­stance, Maya Forstater was fired for stat­ing that men can­not change into women, and her ter­mi­na­tion was up­held by a court; while that speech act seems straight­for­wardly harm­ful, it may be much more harm­ful that her em­ployer – a think tank which cam­paigns against global poverty – weak­ened its mis­sion by firing her. Of course the think tank may have been re­spond­ing ra­tio­nally to pop­u­lar pres­sure to fire her, but then the re­spon­si­bil­ity lies with the out­siders who ap­plied that pres­sure rather than af­firm­ing the im­por­tance of their mis­sion and or­ga­ni­za­tional in­de­pen­dence.

Anti-vaxx campaigns

This should need no elab­o­ra­tion.

Un­scien­tific cli­mate change denial

As­sum­ing that some­one is not pre­sent­ing sci­en­tifi­cally valid ev­i­dence to change the con­ver­sa­tion, it is ob­vi­ously harm­ful for them to gen­er­ally pro­mote de­nial of cli­mate change.

In fact, even if man-made cli­mate change isn’t real (though it re­ally definitely is), it might still be good for peo­ple to be­lieve in it, be­cause the more di­rect cog­ni­tive, health and agri­cul­tural harms of air pol­lu­tion already pose an im­por­tant and ne­glected prob­lem for hu­man­ity.

Th­ese poli­cies are clearly and strongly benefi­cial. They are sup­ported by ro­bust com­bi­na­tions of em­piri­cal ev­i­dence, the­o­ret­i­cal mechanisms, and the most com­mon-sense liberal ideas of so­cial fair­ness and free­dom. There­fore, op­pos­ing them out­right is likely harm­ful, al­though it can be benefi­cial to point out spe­cific down­sides in cer­tain con­texts.

Sup­port­ing Don­ald Trump

Trump is a clearly and par­tic­u­larly poor pres­i­dent, and both his im­peach­ment trial and his elec­tion are forth­com­ing. There­fore pro-Trump sen­ti­ment is gen­er­ally harm­ful.

(How­ever, this doesn’t mean that point­ing out par­tic­u­lar virtues of Trump or re­fut­ing poor ar­gu­ments against him is harm­ful. That can be healthy dis­course.)

Selfish­ness and nihilism

Selfish­ness is a col­lec­tively harm­ful point of view, warp­ing so­cial and poli­ti­cal at­ti­tudes in a va­ri­ety of ways too nu­mer­ous to list. Mean­while, it is hard to see any up­sides to self­ish ide­ol­ogy. It’s true that har­ness­ing self­ish in­cen­tives can have good re­sults, for in­stance in cap­i­tal­ism. How­ever, that doesn’t mean that ex­plic­itly pro­mot­ing that ide­ol­ogy over other at­ti­tudes is a good thing. It seems that self­ish­ness/​ego­ism is one of the bet­ter can­di­dates for ideas to be sup­pressed.

Eth­i­cal nihilism could be con­sid­ered similar, lack­ing in up­side, and prob­a­bly func­tion­ing similarly to self­ish­ness for all prac­ti­cal pur­poses.

Dis­parag­ing EA/​EAs

There are some le­gi­t­i­mate crit­i­cisms of par­tic­u­lar prac­tices and ideas which are com­mon in EA. While these are gen­er­ally voiced ad­e­quately within the EA com­mu­nity by other EAs, it can still be benefi­cial to pro­tect their free­dom of speech in a pub­lic con­text, when they are be­ing re­spect­ful and open to de­bate.

It is a differ­ent story when peo­ple crit­i­cize EAs for rea­sons that are less about sub­stan­tive and fair dis­agree­ment and more about iden­tity and af­fili­a­tion. Com­plaints that EAs have the wrong poli­ti­cal opinions, wrong cul­tural back­ground, wrong race or gen­der, or the wrong mind­set on ra­tio­nal­ity can be very per­ni­cious and un­re­spon­sive to hon­est rea­son and eth­i­cal ap­peals. In­sults, mock­ing, etc is of course very prob­le­matic, and ac­cu­sa­tions of racism/​sex­ism and so on are par­tic­u­larly dam­ag­ing in the cur­rent mil­leu. The harm of this sort of speech is that it can lead peo­ple to de­vote less or none of their time and money to im­por­tant is­sues in­clud­ing global poverty re­lief, an­i­mal welfare, ex­is­ten­tial risk re­duc­tion, in­sti­tu­tional re­form and EA move­ment build­ing. Even if some­one makes it their mis­sion to crit­i­cize just a part of EA, if they pub­li­cly do it in a cyn­i­cal or in­flam­ma­tory man­ner, the re­sult is a gen­eral weak­en­ing of the over­all pres­tige and power of the EA move­ment, to the detri­ment of all EA pro­jects.

Be­cause the EA move­ment is uniquely high-im­pact, it fol­lows that poor crit­i­cism of the EA move­ment is uniquely harm­ful, at least com­pared to crit­i­cisms of other benefi­cial ideas, so­cial move­ments and in­sti­tu­tions.

Honor­able mentions

Ra­cial differences

Ra­cial sci­ence could con­tribute to some of the same harms of racism. It seems par­tic­u­larly likely to pro­mote re­stric­tions against im­mi­gra­tion, and such ar­gu­ments are oc­ca­sion­ally made by figures such as Ja­son Rich­wine. From the per­spec­tive of global welfare, this is not a good pos­si­bil­ity. Coun­tries will pre­dictably err on the side of too much na­tion­al­ism and too lit­tle im­mi­gra­tion, and we should seek to steer them to­wards more cos­mopoli­tan be­hav­ior. How­ever, I think it would be in­cor­rect to say that the main xeno­pho­bic poli­ti­cal trends in Amer­ica, In­dia, Hun­gary, and other coun­tries are sub­stan­tially mo­ti­vated by racial sci­ence. To the ex­tent that laypeo­ple on the right are mo­ti­vated by a sci­en­tific kind of racism, it is more of­ten a mat­ter of ig­no­rant as­sump­tion, not the most re­cent pa­pers and blogs.

Carl (2018) de­scribes down­sides of the taboo on racial sci­ence. One is that it can in­crease peo­ple’s propen­sity to equate moral sta­tus with sci­en­tific at­tributes like in­tel­li­gence (and if some kind of in­con­tro­vert­ible proof of racial differ­ences were found, this would be very dan­ger­ous). Another is that the view of hu­man­ity as a ‘blank slate’ can in­spire to­tal­i­tar­ian efforts to re­make hu­man­ity, or other poli­cies which are more be­nign yet still waste­ful. A third down­side is that the taboo could in­terfere with use­ful med­i­cal treat­ments for peo­ple of differ­ent races.

I don’t see a good over­all ra­tio­nale for cen­sor­ing care­ful stud­ies of things like ‘hu­man bio­di­ver­sity,’ if the re­searchers re­ject xeno­pho­bic and na­tion­al­ist ideas and write things re­spon­si­bly. On the other hand, if some­one more ca­su­ally re­marks that mem­bers of an­other race are in­fe­rior at some­thing, that is harm­ful speech.

Stu­pefy­ing narratives

One of the harms of anti-semitism is that it “makes peo­ple dys­func­tion­ally stupid, un­able to un­der­stand how the world ac­tu­ally works,” since they blame ev­ery­thing on a con­spir­acy. Sim­plis­tic ide­olo­gies which blame ev­ery­thing on un­sub­stan­ti­ated con­spir­a­cies by Jews, the main­stream me­dia, cap­i­tal­ists, Ukrainian par­li­a­men­tar­i­ans, etc could be con­sid­ered stu­pefy­ing nar­ra­tives that just dam­age peo­ple’s abil­ity to make sense of ma­jor is­sues.

There is a high risk of im­plicit cen­sor­ship here. For in­stance, it seems that the West’s very strong at­ti­tude of anti-anti-semitism has buried le­gi­t­i­mate crit­i­cisms of Is­rael and the pro-Is­rael lobby. Gen­er­ally speak­ing, it’s hard to sup­press ir­ra­tional at­tach­ment to stu­pefy­ing nar­ra­tives with­out stamp­ing them out en­tirely. We don’t want to make peo­ple ig­nore the in­fluence of cap­i­tal­ists, the main­stream me­dia, Ukrainian par­li­a­men­tar­i­ans, Ge­orge Soros, etc; we just want them to be less ir­ra­tionally fix­ated on them. It is difficult to see how speech polic­ing could be used effec­tively here, even if we ig­nore the ob­vi­ous prob­lem that cen­sor­ing stu­pefy­ing nar­ra­tives will only en­trench these peo­ple fur­ther in their con­spir­a­to­rial mind­set.

(Of course, vi­cious anti-semitism falls un­der the cat­e­gory of vi­cious racism, and is ac­cord­ingly very harm­ful.)

Not harm­ful speech

Scien­tific dis­re­gard for animals

Some peo­ple ar­gue that an­i­mals have lit­tle or no sen­tient feel­ings of pain. Th­ese views may not be ac­cepted by the ma­jor­ity of neu­ro­scien­tists, but they have enough po­ten­tial merit that it can­not be ap­pro­pri­ate to po­lice them. The ques­tions on the na­ture of con­scious­ness are too murky.

Weight­ing an­i­mals less than hu­mans on the ba­sis of their sen­tience (or in­tel­li­gence as a proxy for sen­tience) is al­most cer­tainly the right at­ti­tude to have, so of course it should not be po­liced.


Some peo­ple have con­demned the idea of tran­sra­cial­ism as offen­sive – for in­stance, see the con­tro­versy sur­round­ing Re­becca Tu­vel’s philos­o­phy pa­per on the sub­ject. How­ever it’s not clear if this idea is ac­tu­ally harm­ful. There isn’t a se­ri­ous prob­lem of “race dys­pho­ria” that needs to be solved via tran­sra­cial prac­tices, but it could be mildly benefi­cial for peo­ple to en­joy the abil­ity to switch races/​cul­tures. It might weaken racial and cul­tural iden­tity, which could be bad by weak­en­ing a source of pride and valuable poli­ti­cal mo­bi­liza­tion, or could be quite good by weak­en­ing a source of sec­tar­ian ten­sion and grievance. Gen­er­ally speak­ing there just isn’t a clear case that tran­sra­cial­ist ideas are harm­ful.

Hu­man enhancement

Forc­ing peo­ple to go through eu­gen­ics like ster­il­iza­tion or abor­tions would be a very bad policy, though it is not as bad as some other kinds of bad poli­cies pro­moted by racism. Fear of eu­gen­ics has caused some peo­ple to po­lice speech which ad­vo­cates it. How­ever, there is lit­tle risk that be­nign mea­sures (like sci­en­tifi­cally in­formed dat­ing ser­vices to pre­vent ge­netic dis­eases, or ge­netic en­g­ineer­ing for par­ents to im­prove the hap­piness or in­tel­li­gence of their chil­dren) would lead to such poli­cies. Western so­ciety has very strong laws and norms against non­con­sen­sual med­i­cal pro­ce­dures (ex­cept­ing cases where peo­ple are men­tally ill), plus a deep an­tipa­thy to­wards any poli­cies that evoke Nazism or sci­en­tific racism, so these new tech­nolo­gies are very un­likely to lead to op­pres­sive poli­cies. Mean­while, there is a sub­stan­tial chance that ge­netic en­g­ineer­ing and other tech­niques could im­prove the health, ethics, hap­piness, pro­duc­tivity or other char­ac­ter­is­tics of new chil­dren. Some peo­ple worry that this could cause com­mu­ni­ties to wither – for in­stance, there will be no deaf com­mu­nity if no­body is deaf any­more. How­ever, such or­ganic shifts in so­cial iden­tities and cul­ture are not harms.

So, while calls for co­er­cive eu­gen­ics are harm­ful, speech which fa­vors ac­tions that are su­perfi­cially similar but ul­ti­mately con­sen­sual and non­vi­o­lent is not. In fact, sweep­ing ide­olog­i­cal at­tacks against hu­man en­hance­ment could be con­sid­ered harm­ful speech in their own right.

Crit­i­ciz­ing the Amer­i­can pro-Is­rael lobby

The pro-Is­rael lobby has a very strong and per­ni­cious in­fluence on U.S. for­eign policy. Some peo­ple fear that con­dem­na­tions of the lobby could lead to an­ti­semitic at­ti­tudes, but wide­spread an­ti­semitism is not a se­vere risk in Amer­i­can liberal democ­racy, and the costs to the Pales­ti­nian peo­ple (and even to the Is­raeli peo­ple, in the long run) from flawed Amer­i­can policy to­wards Is­rael ap­pear greater.

A con­clud­ing point

It ap­pears that com­mon as­sump­tions about what kinds of speech are harm­ful are just partly right. They cor­rectly rec­og­nize many kinds of prej­u­dice against differ­ent types of hu­mans as be­ing harm­ful, but can go too far in us­ing only vague as­so­ci­a­tions with prej­u­dice as a jus­tifi­ca­tion for speech polic­ing. Mean­while, peo­ple must get bet­ter at rec­og­niz­ing the harms of ad­vo­cat­ing self­ish­ness, speciesism, and false opinions on key policy is­sues.