211–220 of 323 posts
1. https://www.nhs.uk/conditions/
Earlier quoted context omitted.
That's a nice page, and I'll definitely bookmark that one. But I don't see most of the public using something like that over WebMD. WebMD is complete trash (I even block it in my Kagi search preferences), but it succeeds over sites like the NHS one because of blog SEO. WebMD is really a glorified blog with posts about every condition you might get paranoid over. I don't know how precisely they achieve this, but they'…
Google's problem is that it ranks pages, not sites. It will much prefer a crappy site with a crappy page that is highly specific to your search query, instead of showing you a good site related to the idea you are looking for. It puts far too much faith in its language parser , which doesn't comprehend .
Examine.com is the closest analogue to what this site is trying to do, and as far as I know, it works. It takes a while, but if people develop trust in a product or website they value that deeply, and are even willing to throw money at it. I don't even consider trying to google a supplement; I'll just go to examine and look at the body of evidence, recommended dosages, etc. I've paid for Examine and I would pay for t…
Earlier quoted context omitted.
I don't know about the NHS, but we can't trust the information vended by the CDC -- they have proven willfully incompetent over the past two years. I will give just one example: they initially told us masks were ineffective against COVID, which they knew at the time was inaccurate. They said it because they didn't want a run on PPE required by medical workers. Result: there was a shortage anyway and they damaged thei…
If one negative example decreases an entity’s credit, would positive examples increase an entity’s credit? What about the proportion of positive to negative examples? How does that impact an entity’s credit?
Lies though, where we find out they knowingly said something they knew was wrong when they said it, don't balance out. The opposite of lying to someone to manipulate them is respecting them enough to let them make their own choices. They need to acknowledge that and start purely providing facts again - even at the cost of some political narratives, before they'll regain trust.
A better question might be: Why is the private sector responsible for providing accurate health information? As this article shows, the incentives for people running medical websites and the people reading them are not aligned. I'd say the UK NHS website and symptoms/medications pages hit the nail on the head - https://www.nhs.uk/conditions/ . It has no advertising and provides short, easily readable and actionable i…
>>Why is the private sector responsible for providing accurate health information? As this article shows, the incentives for people running medical websites and the people reading them are not aligned.>> I don't know that the cause of this particular misalignment is "private sector". But no doubt the current implementations by the private sector exhibit this misalignment. >>I'd say the UK NHS website and symptoms/med…
Perhaps the more likely to be read as benign version would be this: We all agree site A is bad because it's misaligned (which I think all agree is due to being driven by ads). Then we propose that site B is good - but there's no indication about WHY site B is good. We know why site A is bad, but why is site B good. And more specifically since we're describing the value in terms of alignment, what alignment is there in site B that drives the goodness?
Earlier quoted context omitted.
Punishment/vengeance is a popular idea around here, but you have to also remember that a search engine is supposed to bring you the most relevant results. Filtering out, say, Stack Overflow or Reddit because it has ads doesn't help you when it answers your question and is perhaps the only thing on the internet that truly does. People seem to think there's this ad-less replica of the internet, sitting right behind our…
Maybe instead of heavily punishing websites with ads, a search engine could instead punish heavily ad-driven websites. A lot of the SEO-exploiting blog mills are filled to the brim with ads where the goal is to get you to visit to view as many ads as possible, not provide good content that's funded by ads.
Does this site (in general, not just this page) have more than 5 advertisements per page? Between 2 and 4?
Does this site attempt to load 12 trackers? "Only" 4 trackers? Just 1?
Does an AI text analysis of the first few paragraphs match on this nonsense?:
> Fixing your gadget is important. Many people find that their gadget sometimes breaks. Gadget helps us do action easier, and improves our lives. We all hate it when our Gadget doesn't work the way we expect it to. It can be frustrating. Read below for tips on how to fix your gadget. (Followed by 3 more paragraphs of filler before getting to regurgitated gems like "reboot it".
I'm sure we have the AI tech now to semantically see this bullshit and downrank it. Right? (Ok, maybe I overestimate how easy this would be. Forgive me, I'm just ranting here)
Earlier quoted context omitted.
So there's a private option that's terrible (WebMD) and a private option that's terrific (Merck Manuals). And it's the terrible one that ranks well at Google. Sounds like a Google problem and apparently only a Google problem.
That's where I continue to want to see what results in a search engine would look like that heavily punishes presence of advertisements in a result. All the SEO spam pages are ad-driven, so cutting out anything following that incentive should result in removal of all pages that follow that terrible SEO spam pattern that ruins search results.
Because the vast majority of this research does not change the current "standard of care" for most health issues that anyone is likely to experience.
Trying to be your own doctor by reading websites is stupid. Doctors go to school for 8 years and many do years more eductation and practial training in a specialty. You are not going to outsmart them by spending 10 minutes reading an article on a website.
If you are sick, see a doctor. If you are not sick, get an annual checkup if it makes you worry less. This isn't so complicated.
Earlier quoted context omitted.
If one negative example decreases an entity’s credit, would positive examples increase an entity’s credit? What about the proportion of positive to negative examples? How does that impact an entity’s credit?
Mistakes to correct statements, yes. Those balance out. The opposite of being right is being wrong, the proportion of these two is their score. Lies though, where we find out they knowingly said something they knew was wrong when they said it, don't balance out. The opposite of lying to someone to manipulate them is respecting them enough to let them make their own choices. They need to acknowledge that and start pur…
Corruption is bad, lying with good intentions is context dependent bad, mistakes are OK. But if one is going to throw out the baby with the bath water at every instance of corruption/lying/mistake, you are not going to be left with much other than chaos.
Humans are fallible creatures, we have to work and improve with what we have.
The exceptionalism of the USA is the fact that most of its organization are open to criticism and repair. The FDA/CDC/USDA/DoT/EPA/etc and many other non governmental orgs are far from perfect, but they are pretty awesome compared to the alternatives around the world.
There's a difference between symptoms, diagnosis, treatment options, evidence for those, and the audiences for whom these are written. WebMD and friends target a broader market, scientific studies target...scientists and doctors.
I think the hard parts of building a better WebMD (along the article's lines) are:
- screening articles for relevance. This is more than mere search or finding potentially relevant articles, but also making a decision to include them
- extracting _structured_ information from the articles. Frequently we talk about Populations/Problems, Interventions, Comparators (interventions), and (medical) Outcomes, collectively PICOs. Extraction of each component is easy. Assembling them is surprisingly hard. Finding equivalencies between them within documents is surprisingly hard. Finding facets or different parts of a treatment (how do you handle combined treatments for a single study? how about when some studies use them and some don't?)
- establishing equivalencies for your evidence synthesis between documents is even harder: do you care about dosage? combination with other treatments? what happens when a slightly different formulation of a treatment gets used, or a treatment is administered poorly? I don't know of anyone within scholarly document processing community working on analysis of medical methods (possibly ignorance on my part!).
I think real time updates are the smallest problem among these: solving extraction (including the big structured objects) mostly solves this; the statistics of a meta-analysis are not complicated. To be clear, this is still an issue. I have a figure[2] highlighting the lag.
Presenting the evidence in a digestible, meaningful way, seems like a hard HCI problem to do right, and easy to do poorly. Merely giving PICOs (structured!) and findings is easy and a bad UX for the non-technical, a narrative summary is interesting to provide and hard to do well, and a subgroup analysis can suffer problems from differing group sizes and effectiveness findings (tailoring is tricky).
There are several organizations[3] working on these sorts of problems.
[1] https://www.byronwallace.com/
[2] https://www.semanticscholar.org/paper/MS%CB%862%3A-Multi-Doc...
[3] https://community.cochrane.org/organizational-info/people/st...