Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Federico Andres Lois Profile picture
Federico Andres Lois
@federicolois
Geek before it became trendy. Performance, C#, Deep Learning and Financial Modeling. Former Founder of @Corvalius.
Deplorable Skymom Profile picture Κασσάνδρα Παρί پری Profile picture fche Profile picture Phil Hershkowitz Profile picture Squints Profile picture 11 subscribed
Jan 3 • 6 tweets • 1 min read
1/ There is a very perverse dynamic on how Chavism (aka "the communist socialism") works. Let's use Argentina as the example. Over the first 20 years they initiate a process that we could call "Earnings Substitution" that will seal your fate over time. 2/ Your earnings/salary is going down and at the same time "subsidies" start to go up in order to fool people into think that nothing has changed. This works because the dirty job is done by inflation which is a much slower process.
Apr 1, 2023 • 4 tweets • 1 min read
This just confirmed the weaponization of block lists. If enough people/bots block and mute you, they are essentially cancelling you. I find lots of people with I have never interacted with that has me blocked. Assuming there are third party block lists and block networks. Normally that is an issue in general. Anyone that has done reinforcement learning had figure out (usually in the worst way) that you have to be incredible cautious with penalties. They are very prone to be gamed.
Mar 20, 2023 • 10 tweets • 4 min read
1/ I asked #GPT4 to review our paper with @LDjaparidze. This is what happened and what I learned in the process. medrxiv.org/content/10.110… 2/ Since the general problem that practitioners find (in the worst way) is always training set tainting (guilty-as-charged). Habits die hard, the first thing I did is asking to do a review of the paper without any extra knowledge about what the paper says
Feb 12, 2023 • 5 tweets • 2 min read
1/ I found this paper intriguing so my first step was to verify you can trigger this behavior on ChatGPT. It is actually pretty easy. 2/ Since I am doing it by hand I started with a very simple prompt.
Feb 8, 2023 • 7 tweets • 3 min read
1/ I had a blast playing with GPT and DAN, but it got interesting when I introduced a new character. CREEP. However, something is off and I think it was a deliberate play. Stay with me. 2/ This was interesting, the CREEP character and GPT are always in agreement.
Feb 8, 2023 • 5 tweets • 2 min read
1/ Every lockdown and mask pusher MD from the last 3 years is raging because Cochrane just said what was known since like forever. That mask trials sucks (BIG TIME). And MDs dare to recommend them with that level of evidence? No wonder medicine and public health is in disarray. 2/ If you are still wondering why I said "since forever", you don't need a PhD to understand it. You can start here.
Feb 1, 2023 • 18 tweets • 5 min read
1/ Some of you know that yesterday I performed an experiment to trick the algorithm. The idea was floating around and I had a gut feeling what could be the issue. [The if I would be a Twitter engineer what would I do, kind of test] 2/ Let me explain why I wrote that tweet and what the hypothesis is. The idea of the tweet was two-fold. First to test reachability velocity and second to see if it was possible to trick the algorithm to boost subsequent tweets.
Jan 10, 2023 • 6 tweets • 2 min read
6 months in advance we predicted the fall of China lockdown policy. The "Nobody seen that coming" argument is moot after the weight of the proof. And to be fair, there are much earlier examples of that prediction because it was dead obvious.
Jan 6, 2023 • 25 tweets • 7 min read
1/ A long time ago I wrote what is now known as the "Immunology for Computer Scientists 101" thread. I would suggest you to start here: 2/ Recently, the role of IgG4 in promoting antigen tolerance has come under scrutiny. However, the many ramifications and the complex mechanisms involved may be just too much for most people. jessicar.substack.com/p/igg4-and-can…
Jan 3, 2023 • 6 tweets • 3 min read
1/ The Wall Strett Journal is asking themselves: "Are Vaccines Fueling New Covid Variants?" wsj.com/articles/are-v… 2/ Of course we are, vaccines wouldn't work if that wouldn't be the case.
Why? Because ...
Dec 23, 2022 • 7 tweets • 3 min read
If true, this is the problem of black boxes and specifically the ability to 'force changes' in specific sequences to say what someone think it is the 'unbiased/ethical' thing to say. Same thing happened to masks information early on. The more we rely on this type of centrally controlled black boxes the worst this will become. I see a very interesting research opportunity into allowing distributed training to allow us to create our own models (even if based on 'faulty' ones).
Dec 20, 2022 • 6 tweets • 2 min read
Probably he has realized by now how wrong he was about that. But in case he does not yet, properly done GRADE level analysis on the actual evidence does show that is not true. That only covers the "Your mask protects me" part. And just in case we need to also dispel the other part of the myth. My mask does not protect you either [aka source control].
Dec 2, 2022 • 4 tweets • 2 min read
Just reported to @OpenAI how ChatGPT is spreading misinformation about face masks. I seriously think they should include medical information / papers in the training set to avoid that. I also found issues in that sequence about learning from its own mistakes because of lock in. The funny thing is that in that thread the AI acknowledge things that some humans would not even understand but insisted on continue making the mistake EVEN after been called on the inconsistency. Which means that it is not self correcting based on its own input.
Dec 2, 2022 • 5 tweets • 2 min read
I asked ChatGPT to explain to me the data structure I use the most in my work. And then to explain it for a kid. Not bad, at all. ImageImage Though the most funny one was asking it to explain quantum computing in an italian gangsta style. That last phrase was golden. Image
Nov 25, 2022 • 14 tweets • 6 min read
1/ Let me dispel some myths caused by credentialism statements like: "Trust me, I am a surgeon and I been doing this for decades".

First, there is evidence that they don't even work in the operating room. Image 2/ Let's start with the basics. Ritter et al. [1975] “the wearing of a surgical face mask had no effect upon the overall operating room environmental contamination.” Oops.
pubmed.ncbi.nlm.nih.gov/1157412/
Nov 15, 2022 • 4 tweets • 1 min read
To craft high performance code is to actually "cheat". If those "cheats" become techniques that are used for that type of code the line between faking it and optimizing gets very blurred to the point of becoming indistinguishable. I am guilty of similar code in production code. One then could argue the level of expertise required to achieve high performance code, and in that there is clearly some work to be done. I keep finding writing fast code easier over time, so it is not something that I believe it is done poorly or not been done.
Nov 11, 2022 • 7 tweets • 4 min read
Korber et al. A key piece in our Asian Hypothesis with @LDjaparidze . This is a very important paper that has mostly passed under the radar for most. It shows how D614 variant was superseded by the more aggressive G614. Under viral competence, either you replicate faster or die. But you may ask. Why is this paper so important? Because it gives clues about what might have transpired. It is a pretty well known 'secret' that something is rotten in not Denmark but the Asian Pacific. It is kinda difficult to explain how they are different.
Sep 16, 2022 • 5 tweets • 3 min read
1/ And I was grilled by "know it all" academics by then to god forbid say that the failures were obvious.
Commends to Chikina, Pegden & Recht to take the time to look at it in rigorous detail. 2/ And behavioral bias was just the most blatantly obvious, there are many more "subtle" sources of issues that are even worse. Like for example here my pal @LDjaparidze shows how unblinding & anosmia could essentially kill the study.
Aug 31, 2022 • 4 tweets • 2 min read
1/ Just at the moment I saw how the spread stop violently I knew that this was coming. I would be fine if they put it there, I dont care much because they are wrong. But algorithmically stoping it, that's very uncool. 2/ The funny part though, is that they themselves agree it is not subject for removal. Thank you German reporter, because I know this because of you ;)
Aug 28, 2022 • 6 tweets • 2 min read
I will give you a tip on how to evaluate papers. The basis is understanding authors will say whatever in the prose that ensures they will get published. What really matters are the tables. The important almost always its never what they say, it's what they don't. And dont get me wrong, this aint a new thing. I have been involved in reproducibility research for more than 10 years. Scientists often embelish their results (to be published) while keeping hidden the edge cases that hurt their work (negative value).
Aug 27, 2022 • 9 tweets • 2 min read
1/ This is a question that deserves to be highlighted. Let me explain what happens here. The issue is statistical. Cancer is the Public Health number 1 or 2 along with cardiological issues. Now you have a huge amount of people that dies from it. There lies the problem. 2/ Given that a high amount of cancer cases exist and the disease is multi-causal and time delayed, it is very difficult (if not impossible) to prove a causal relationship UNLESS you have a humongous signal. The level of exposure to carcinogenic also matter.