science

The discussion about sexual assault at conferences has been going on for a few years now. Moral reasoning has been discussed a lot, and I will not repeat that.¹

Here I will give a dispassionate, cold and calculating reason why your community cannot afford to tolerate 1% predators:

If in a community of 50 men and 50 women, one person is a predator who attacks one woman every year and causes her to leave, and every year either a man or a woman joins, the community will be male-only after 100 years.

Even 1% predators is far too much.

Farewell to friends -- and a love.

Download: mp3 audio | webm video; Watch on youtube

New recording 2019 at Intermezzo with Rika Körte and Steven Macdonald as recording engineers: mp3 audio

Goodbye my love, I leave tonight,
I know you’re in new hands,
Though I would rather follow you,
That’s not the way that this is planned,
Our destiny will now be watched
by different eyes than mine,
I wish you just the best,
be sure we’ll meet again in time.

Goal: Improve the decentralized spam filter in Freenet (WoT) to have deterministic network load, bounded to a low, constant number of subscriptions and fetches.

This article provides calculations which show that decentralized spam filtering with privacy through pseudonyms can scale to communication systems that connect all of humanity. It is also applicable to other systems than Freenet, see use in other systems.

Originally written as a comment to bug 3816. The bug report said "someone SHOULD do the math". I then did the math. Here I’m sharing the results.

Useful prior reading is Optimizing a distributed spam filter for Freenet.

This proposal has two parts:

1. Ensuring an upper bound on the network cost, and
2. Limiting the cost due to checking stale IDs.

Scientific institutions¹ currently base a large part of their internal evaluation, their comparison to others, and their hiring decisions on counting publication (with a number of different scorings).

And this is dumb.

On the surface this causes pressure to publish as many papers as possible² which drives down quality of publications to the lowest standard reviewers accept.³ And it strengthens a hierarchy of publishers, where some publications are worth more than others based on the name of the journal. That simplifies funding decisions. But makes them worse. And it creates an incentive to get a maximum of prestige with a minimum of substance.

    by Mike Perry (http://nodicemike.com)

We still have to stop CO₂ emissions and plant trees to prevent even worse catastrophes, but since 2022 the most likely future is that there will be catastrophes even if we stop CO₂ emissions right now. This is what climate scientists in the past 30 years hoped to prevent. We failed. Now we must fight to avoid even worse outcomes. We are making progress at that, but we must speed up.

Update 2022: As by the WMO, we’re now at 50% within the next 5 years: “⚀ or ⚁ or ⚂” (1 or 2 or 3). “The odds of at least one of the next 5 years temporarily reaching the Paris Agreement threshold of 1.5°C have increased to 50:50. In 2015 the chance was zero.”

Update (2021-09): According to IPCC AR6, we’re now at 50%.

Update 2018-09-03: As by Aengenheyster et al. 2018, we’re now at “⚀ or ⚁” (1 or 2): »However, reaching the 1.5 K target appears unlikely as MM would be required to start in 2018 for a probability of 67%.« MM means getting a 2% increase of the share of renewables every year.

I don’t know what we rolled, but I sure hope it’s not a 1.¹

For the robust science behind the green future, see Hansen et al. 2017:

Young people's burden: requirement of negative CO₂ emissions.

James Hansen et al. published a paper about the expected costs due to climate change, aptly named "Young people's burden".

→ Young people's burden: requirement of negative CO2 emissions

(Hansen et al. 2017, License: cc-by)

A comment on Amending Published Articles: Time To Rethink Retractions And Corrections? (doi: 10.1101/118356) which asks for making it easier and less of a matter of guilt to change published articles.

Update: Leonid Schneider from forbetterscience notes that there’s a whole dungeon of misconduct which might be facilitated by “living papers”. We need investigate problems in depth before changing established processes. Scientific communication is a complex process. Publication is an important part of it.

Firstoff: The underlying problem which makes it so hard to differenciate between honest errors and fraud is that publications are kind of a currency in science.

Report: Arctic Is Unraveling, discusses assessment Snow, Water, Ice and Permafrost, notes the article rising tide — sounds more like Hansen was right.

AGU publications published "The world's biggest gamble", a short commentary on how to go on with climate change.

• Earth’s Future: The world's biggest gamble (Open Access: cc by)

I am hard pressed not to become sarcastic. Not because the commentary is wrong. It’s spot on. But because we, as a species, are …

I’ll stop speaking my mind for now. Let’s hope that hope wins against frustration and our children don’t have to pay too dearly for the idiocy of my generation and the generation before.

Oh well, Happy Halloween and enjoy Samhain.

I just spent half an hour on finding the references for this, so I can spend 5 minutes providing it for others on the web.

I repeatedly stumbled over needing bibtex entries for the IPCC reports. So I guess, others might stumble over that, too. Here I share my bibtex entries for some parts of the IPCC reports.¹

IPCC 1990 WG1 (physical science basis)

@BOOK{IPCC1990Science,
  title = {Climate Change 1990 The Science of Climate Change},
  publisher = {The Intergovernmental Panel on Climate Change},
  year = {1996},
  editor = {J.T. Houghton and G.J. Jenkins and J.J. Ephraums},
  author = {IPCC Working Group I}
}

On 2015-08-27, Researchers from the Reproducibility Project: Psychology reported that in 100 reproduction studies, only “47% of original effect sizes were in the 95% confidence interval of the replication effect size” (RPP SCIENCE 2015, an overview of the results is available in Scientific American; in german from DLF Forschung Aktuell).

I take this worrying result as cue to describe current challenges to scientific publishing and measures to address them — including reproduction experiments, and what to do if they contest previously published and referenced work.

For or about scientific work.

I just watched the short film Ambition from ESA, and I still have tears in my eyes.

The film is awesome.

Timezones of most active TCCON stations in UTC+x (without daylight saving time (DST). Because I needed it and could not find a simple list quickly.

anmyondo: +9,
ascension: 0,
bialystok: +1,
bremen: +1,
caltech: -8,
darwin: 9, # Timezones2008 says 9 1/2???

The problem

Nature only gives me ris-formatted citations, but I use bibtex.

Also ris is far from human readable.

The background

ris can be reformatted to bibtext, but doing that manually disturbs my workflow when getting references while taking note about a paper in emacs.

I tend to search online for references, often just using google scholar, so when I find a ris reference, the first data I get for the ris-citation is a link.

The solution

If you want to publish your scientific scripts, as Nick Barnes advises in Nature, you can very easily do so with Mercurial.

All my stuff (not just code), excempting only huge datasets, is in a Mercurial source repository.¹

Whenever I change something and it does anything new, I commit the files with a simple commit (even if it’s only “it compiles!”).

science in a dent:

(1) Form a theory. (2) design an experiment to test the theory. (3) do it. (4) Adjust the theory, if needed → (2)

→ written in GNU social.

Please feel free to use it!

If that’s to brief:

→ the scientific method, explained very basically and simply.

and

*That’s not faith. It’s theory.