Self-flagellation Friday

Confirmation bias, filter bubbles, etc. mean that it’s easy to convince yourself of falsehoods. I hope that explicitly devoting time and attention to ideas that aren’t intuitively appealing or otherwise aligned with my worldview will help guard against this.

Also, this tag will likely include posts arguing against things that I don’t actually believe, but that I probably appear to believe (e.g. economism).

Assorted links I

1. Dear governments: Want to help the poor and transform your economy? Hold on, recalculating… and follow-up GiveDirectly Three-Year Impacts, Explained

   Reductive summary of claims: Positive impacts of direct cash transfers attenuate considerably over time. Negative spillover effects for non-recipients are substantial.

   Comparing recipients households to non-recipients in distant villages, we find that recipients of cash transfers have 40% more assets than control households three years post transfer. This amount (USD 422 PPP) is equivalent to 60% of the initial transfer (USD 709 PPP). However, we do not find statistically significant across-village treatment effects on other outcomes. This difference could stem … from potential spillover effects at the village level. Indeed, non-recipient households in treatment villages show differences to pure control households on several dimensions. The point estimates suggest spillover households spend USD 30 PPP less than pure control households, or about 16% based on a pure control mean of USD 188 PPP, and score ~0.25 SD less on an index of food security than pure control households. Spillover households also score ˜0.18 SD less on an index of psychological wellbeing than pure control households.

2. Alleviating Global Poverty: Labor Mobility Direct Assistance, and Economic Growth

   The magnitude of the income gains of the “best you can do” via direct interventions to raise the income of the poor in situ is about 40 times smaller than the income gain from allowing people from those same poor countries to work in a high productivity country like the USA. Simply allowing more labor mobility holds vastly more promise for reducing poverty than anything else on the development agenda.

3. Our treatment of animals is stalling human progress

   Overall, I think the essay doesn’t do much to support the titular claim. The closest it gets is a couple paragraphs in the middle stating, “The industry exploits complex, sentient beings as resources, which is a woefully inefficient process.” I’d certainly be interested in hearing a more compelling argument on this topic (i.e. animal welfare arguments that are justified solely in terms of benefits to homo sapiens). Pointers?

4. The optimizer’s curse and On the psychology of prediction

   when comparing actual outcomes to value estimates, we should expect to be disappointed on average, not because of any inherent bias in the estimates themselves, but because of the optimization-based selection process.

   This true story illustrates a saddening aspect of the human condition. We normally reinforce others when their behavior is good and punish them when their behavior is bad. By regression alone, therefore, they are most likely to improve after being punished and most likely to deteriorate after being rewarded. Consequently, we are exposed to a lifetime schedule in which we are most often rewarded for punishing others, and punished for rewarding

5. Adam Rifkin Eyes ‘Peeps’ Classic Candy Treats For Animated Film & TV Franchise

   Now accepting pitches. Mine:

   Peeps resent their tortured existence. But this Easter, they have a plan to end their misery once and for all. They’ll travel back in time to the eve of Jesus’s crucifixion and gear up to stop it. No crucifixion, no resurrection, no Easter, no Peeps. Like 300, but more biblical.

   

A lazy approach to AI safety

According to most ethical theories, the actions of any AI agent we create have moral import. To accommodate this reality, we guide the agent with a utility function and (implicitly or explicitly) an ethical framework.

One of the key problems of AI alignment is that we are uncertain about which ethical theory to encode in the agent because we (philosophers, humans, society) are ourselves unsure of the correct ethical theory. How can we expect our agent to act in accordance with our values when we don’t even know what our values are?

I propose that we wave the white flag of surrender in the battle to find final, certain answers to the hard problems of ethics. Instead, we should reify our uncertainty and our search procedures in agents we build.

The scarcity of cooperatives

Introduction

The predominance of capital-managed firms (CMF) over worker cooperatives (WC) remains an open question in economics. Early explanations relied on a hypothesized comparative inefficiency of cooperatives. Subsequent empirical study has shown that cooperatives are at least as efficient as capital-managed firms (Doucouliagos 1995) (Estrin, Jones, and Svejnar 1987) (Craig et al. 1995) (Levine 1990).

A profusion of hypotheses has since arisen. (Dow and Putterman 1999) offers a good summary (though the term must be used loosely for a 126-page paper). One that I have not seen presented is: capital-managed firms predominate because capitalists have a greater incentive to expand than worker-owners in worker cooperatives. Roughly, for each market segment a capitalist expands into, their income increases by capital’s share of the new segment’s profit. For each market segment a cooperative expands into, the expanders receive no direct remuneration (supposing that the new market segment is also a cooperative). Any new profit goes to worker-owners in the new segment.

Graph of contents

A table of contents can provide a useful overview of a document’s content. However, because of its limited form, some information about the structure of the document must be omitted.

If, instead, we use a graph of contents, we can convey additional information about the relationships between sections. In effect, we combine the table of contents with an argument map.