47 lines
3.4 KiB
Markdown
47 lines
3.4 KiB
Markdown
## GiveWell
|
||
|
||
wget --recursive --no-clobber --html-extension --domains givewell.org --follow-tags=a --reject '_.js,_.css,_.ico,_.txt,_.gif,_.jpg,_.jpeg,_.png,_.mp3,_.mp4,_.pdf,_.tgz,_.flv,_.avi,_.mpeg,_.iso,_.xls,_.xlsx,_.csv,_.doc,_.docx,_.mpa,\*mp4' --ignore-tags=img,link,script --header="Accept: text/html" --no-parent https://www.givewell.org
|
||
|
||
grep -ri "Internal forecast" -E "prediction|Prediction|forecast|Forecast" \* | sed 's/^/https:\/\/www.givewell.org\//' > searchresults.txt
|
||
|
||
grep -ril "Internal forecast" -E "prediction|Prediction|forecast|Forecast" \* > searchresults.txt
|
||
cat searchresults.txt
|
||
cat searchresults.txt | sed 's/^/https:\/\/www.givewell.org\//' > searchresults2.txt
|
||
cat searchresults2.txt
|
||
grep -v "print" searchresults2.txt > searchresults3.txt
|
||
|
||
while read line; do
|
||
firefox --new-tab "$line"
|
||
done < searchresults3.txt
|
||
|
||
We are experimenting with recording explicit numerical forecasts of the probability of events related to our decision-making (especially grant-making). The idea behind this is to pull out the implicit predictions that are playing a role in our decisions, and to make it possible for us to look back on how well-calibrated and accurate those predictions were.
|
||
|
||
For this grant, we are recording the following forecasts
|
||
For this grant, we are recording the following <a href="https://www.givewell.org/research/internal-forecasts">forecasts</a>:
|
||
For this grant, we are recording the following <a href="https://www.givewell.org/research/internal-forecasts">forecast</a>:
|
||
For this grant, we are recording the following forecasts (made during our decision process):
|
||
|
||
We are experimenting with recording explicit numerical forecasts of the probability of events related to our decision-making (especially grant-making). The idea behind this is to pull out the implicit predictions that are playing a role in our decisions, and to make it possible for us to look back on how well-calibrated and accurate those predictions were. For this grant, we are recording the following forecast:
|
||
|
||
We’re experimenting with recording explicit numerical forecasts of events related to our decisionmaking (especially grantmaking). The idea behind this is to pull out the implicit predictions that are playing a role in our decisions, and make it possible for us to look back on how well-calibrated and accurate those are. For this grant, we are recording the following forecasts:
|
||
|
||
We’re experimenting with recording explicit numerical forecasts of events related to our decisionmaking (especially grantmaking). The idea behind this is to pull out the implicit predictions that are playing a role in our decisions, and make it possible for us to look back on how well-calibrated and accurate those are.
|
||
|
||
Divide by h2, then pull the second which has forecasts
|
||
|
||
## OpenPhil
|
||
|
||
wget --recursive --no-clobber --html-extension --domains www.openphilanthropy.org --follow-tags=a --reject '_.js,_.css,_.ico,_.txt,_.gif,_.jpg,_.jpeg,_.png,_.mp3,_.mp4,_.pdf,_.tgz,_.flv,_.avi,_.mpeg,_.iso,_.xls,_.xlsx,_.csv,_.doc,_.docx,_.mpa,\*mp4' --ignore-tags=img,link,script --header="Accept: text/html" --no-parent https://www.openphilanthropy.org
|
||
|
||
Find and delete largest files
|
||
du -a . | sort -n -r | head -n 20
|
||
find . -xdev -type f -size +100M
|
||
|
||
find . -type f -exec du -s {} \; | sort -r -k1,1n | head -n 20
|
||
|
||
grep -ril -E "Internal forecast" \* > searchresults.txt
|
||
|
||
grep -v "print" searchresults.txt > searchresults2.txt
|
||
|
||
Note to self: OpenPhil uses h3 headers instead.
|