Ethical issues in moral and social enhancement

7.7. Why is the God Machine an undesirable way of making a better society?
7.7.1. Introduction 
In the previous part of this chapter I focused on the question of the powers that can 
be justifiably given to the state. Given that in our world things can go wrong and 

---

176 
protections from the arbitrary and unchecked interference may fail, Harris’ critique 
of Savulescu and Persson’s (2012) vision is convincing. The question about the 
exact limits of state power, however, remains open. In this section I will consider 
what I think is a crucial foundation of Savulescu and Persson’s (2012a) argument: 
that the state’s freedom-impairing interference is justified if it prevents harm. 
Savulescu and Persson (2012a) refer to Mill’s harm principle to support their point. 
Firstly, I will review aspects of Harris’ (2014a) rebuttal. Secondly, I will propose 
that Savulescu and Persson (2012a) misconstrue the purpose of the harm principle, 
which weakens their case for the desirability of the God Machine. In the second 
part of this section I will present an approach which extends the more convincing 
aspects of Harris’ (2014a) argument and, in my view, is more successful in 
grounding an in-principle critique. 
7.7.2. The harm principle
Harris (2014a) first quotes Dworkin, who insists on a distinction ‘between the idea 
of liberty as license, that is, the degree to which a person is free from social or legal 
constraint to do what he might wish to do, and liberty as independence, that is, the 
status of a person as independent and equal rather than subservient’ (cf. Harris 
2014a, p. 258). I have previously suggested that the argument from non-domination 
(and the same will apply to ‘subservience’) is not convincing. That leaves us with 
the importance of liberty understood as independence. I agree with Harris that 
Savulescu and Persson’s invocation of Mill’s harm principle seems to overlook the 
importance of the distinction between liberty as licence and liberty as 
independence.
Harris writes:
‘So Mill did not advocate the sort of freedom to do wrong that 
the law controls. But he recognized, as Savulescu and Perrson do 
not, that the law is not infallible, and the room, the 
independence, it leaves citizens to form their own values and 
choose their own way of life is vital for a free society—a society 
in which even basic laws may be changed for compelling 
reasons. The God machine takes away the independence of 
decision making, of thought that can lead to action; this is why it 

---

177 
is incompatible with both independence and autonomy, 
incompatible with both liberty as license and liberty as 
independence.’(2014a, p. 258)
However, the first problem is that this in itself does not take us far in undermining 
the Savulescu and Persson (2012a) argument if we take away the support of the 
other arguments presented by Harris: that we might fare poorly if we believe in any 
instruments’ infallibility and that the ability to change laws is important (I think this 
ability is largely preserved in the God Machine scenario). This is because although 
liberty as ‘independence’ is important, and it too can be restricted – at least insofar 
as independence includes acting on one’s life plans. For example, Mill argues that 
freedom of expression, thought and discussion is a fundamental, and fundamentally 
important, liberty. For Mill, its importance lies in keeping true beliefs from 
becoming dogmatic, which is necessary if we are to fulfil our nature as progressive 
beings (see: 
On Liberty
,
II 20). Yet, even the exercise of basic liberties is limited 
by the harm principle, which justifies restricting liberty 
to prevent harm to others. 
Freedom of expression can be restricted on the basis of harm, as in this well-known 
passage from 
On Liberty
: 
[E]ven opinions lose their immunity when the circumstances in 
which they are expressed are such as to constitute their expression 
a positive instigation to some mischievous act. An opinion that 
corn dealers are starvers of the poor, or that private property is 
robbery, ought to be unmolested when simply circulated through 
the press, but may justifiably incur punishment when delivered 
orally to an excited mob assembled before the house of a corn 
dealer, or when handed about among the same mob in the form of 
a placard. (J.S. Mill, ON, III 1) 
Second, according to Mill, the harm principle is something that we can apply 
prospectively to prevent someone from acting in certain ways and causing harm. 
Although in many cases what we could only reasonably know is that a given 
action 
risks
harm, this seems to be all that Mill requires (ON, IV 10). We have 
previously assumed that the God Machine can assess the danger well enough 

---

178 
(infallibility is not needed). 
Thus, 
Mill’s harm principle can justify interference 
even with the liberties that Mill considers basic. As a result, introducing the 
distinction between liberty as independence and as licence in general cannot carry 
the weight required to defeat 
Savulescu and Persson’s (2012a)
appeal to the harm 
principle. A further argument is needed.
There are two arguments I wish to make here. The first argument concerns the 
nitty-gritty of the application of the harm principle. Mill argued that the harm 
principle outlines the sphere of self-regarding actions that is protected from ’others 
meddling’ and 
not
to argue for a converse claim that all that brings harm to others is 
for that reason only open to interference (i.e. harm is a necessary but 
not
sufficient 
condition for a justified interfering action of the society or the state), a mistaken 
interpretation that seems to be at the core of 
Savulescu and Persson’s (2012a) 
confusion. In fact, Mill clarifies that: 
[I]t must by no means be supposed, because damage, or probability 
of damage, to the interests of others, can alone justify the 
interference of society, that therefore it always does justify such 
interference. (J.S. Mill,
 
ON, V 3)
Thus, Mill's position is that causing harm is always 
prima facie 
and a non-
negligible reason in favour of interference, but that this reason might be outweighed 
by reasons not to interfere. As a result, 
Savulescu and Persson (2012a) 
argument 
based on Mill’s harm principle (at least as it stands) fails.
Secondly, it is highly unclear why in the God Machine scenario the interference is 
to come in the shape of a direct and surreptitious change of intentions, rather than 
simply prediction and a last-minute preventative intervention of a 
SWAT team. 
Surely, if the God Machine can predict people’s actions with sufficient accuracy to 
warrant a change of intentions
41
, why not simply monitor citizens’ thoughts, yet 
41
Harris’ (2014a) point about the difficulty of inferring the actual quality action looking 
from the inside at the intention alone is well taken. For the God Machine’s intervention to 
be sufficiently accurate in changing intentions, it has to have access to the wide knowledge 
base of the individual whose intentions are to be changed to be able to assess not only the 
intention to act but also the agent’s interpretation of this action. This would mean that at 
most the God Machine would know how the actions appear to the agent.

---

179 
leave the freedom to nearly make the tragic mistake and make the intervention 
overt? Allowing agents to take the first steps of the harmful action would make the 
reasons for the intervention clearer, leave the possibility of disputing the 
interference and allow people to learn from their near-mistakes. In that case, 
however, the reason for changing peoples' intentions would not be the serious harm 
of others, but rather the harm to the agent. And this is exactly the kind of 
interference that would 
not
be justified by Mill’s harm principle.
42
Moreover, there are powerful reasons not to allow the interference of the God 
Machine via changing intentions directly, reasons which I will explore in the next 
section.