This is how science actually works. It’s less the parade of decisive blockbuster discoveries that the press often portrays, and more a slow, erratic stumble toward ever less uncertainty. Our understanding oscillates at first, but converges on an answer. 1/ theatlantic.com/health/archive…
There’s a lot of expertise around, but fewer tools than ever to distinguish it from everything else. Pure credentialism doesn’t always work. People have self-published terrible pieces on Medium, but some of the best early explanations for laypeople were from tech guys. @zeynep 2/
The issue is that modern expertise tends to be deep, but narrow. But pandemics demand both depth and breadth of expertise. No one knows it all, and those who claim to should not be trusted. @edyong209 3/
The pandemic's length traps people in a liminal space. To clarify their uprooted life & indefinite future, they try to gather as much information as possible-and can't stop
Pandemics actually unfold in slow motion & there's no event that changes the whole landscape on a dime 4/
“Journalists still think of their job as producing new content, but if your goal is public understanding of COVID-19, one piece of new content after another doesn't get you there. It requires a lot of background knowledge to understand the updates" @jayrosen_nyu 5/
Case counts don't show how many people have been infected on any given day. They reflect insufficient testing, reporting lags, and false negatives. 6/
"I think people underestimate how difficult it is to measure things. For us who work in public health, measuring things is like 80 percent of the problem." @nataliexdean 7/
• • •
Missing some Tweet in this thread? You can try to
force a refresh
One depressing aspect of the pandemic is how countries refuse to learn from other countries. Within a country, states refuse to learn from other states. Many refuse to learn from history. Many believe in exceptionalism, that they won’t face what everyone else has. 1/
I still remember first seeing the images of tent hospitals in Lombardy and realizing that this could happen everywhere. Jeremy & I did a data analysis and wrote at the time 2/
75% of people aged 16+ in UK have both doses of covid vaccine & there are currently 700 covid deaths PER WEEK in UK
Some Aus leaders want to reopen when vaccines for ages 16+ hit 70-80%. If our death rate is proportionate to UK, that would mean 266 Australians dying PER WEEK. 1/
Many in UK have already had covid, so it's likely that the AUS death rate could be higher than that 266 ppl per week
75% of ppl 16+ is only 60% of the whole population. 60% against Delta is not enough. We need to vaccinate children & we need rates ~90%. 2/
Some point out how society accepts deaths from flu. In 2019, there were 486 flu deaths in Australia (averages to 9 per week). 2017 was particularly bad with 1,255 flu deaths (avg 24 per week).
What we are facing with covid is over 10x more. These are not the same. 3/
I’m hearing more people in Australia talk about wanting to "live with Covid", even though only 22% of the population is fully vaccinated. #LivingWithCovid (combined with low vaccination rates) means… 1/
Living with Covid (+ low vaccine rates) is:
- Delaying surgery for cancer, organ transplants, brain tumors
- waiting an hour to get an ambulance after heart attack
- turning medical emergency into a catastrophe, b/c the hospital is maxed out
- millions disabled with LongCovid 2/
Living with covid is not just the death count, it is 10-30% of so-called "mild" cases becoming permanently disabled with LongCovid, which can include debilitating neurological effects and constant pain. 3/
An overall lack of recognition for the invisible, arduous, & taken-for-granted data work in AI leads to poor data practices, resulting in data cascades (negative, downstream events)... “Everyone wants to do the model work, not the data work” 1/
Data quality issues in AI are addressed with the wrong tools created for, and fitted to other tech problems—they are approached as a database problem, legal compliance issue, or licensing deal. 3/
5 Myths of Co-Design for Ethical ML
- ‘Better’ involvement➡️ 'better’ design outcomes
- Co-design increases agency of patients
- Representation reduces risk of harms
- Co-design is an inherently ethical approach
- All problems can be co-design problems @josephdonia@jayshaw29 1/
Novel challenges when co-designing AI:
- participation is often unwitting (eg as training data)
- AI technologies can be repurposed after deployment
- black box nature
- hard to account for how data produced by system will be used in future 2/
"Better" involvement does not imply a stronger focus on the whole system (including consequences related to data commodification & surveillance), much of which is out of view for both users & designers 3/
Jeff even includes Black in AI, a fantastic org co-founded by @timnitGebru, whom he fired and then tried to portray using the angry Black woman trope. 2/
One of the 3 conflicting stories that Google has provided about why Dr. Gebru was fired is that it was for being honest about how working on diversity initiatives at Google made her life HARDER. 3/