This game looks so much better than it did a few years ago! I wish people used open source games like this for AI competitions. So much more open and also what a great community effort.
Woah thanks :)!
We had previous experience with writing a compiler for subset of C (during uni), and I also did some learning from the "Crafting Interpreters" book, but regarding the DSL design, I would say it is actually simpler -> we just went kind of ad-hoc and are expanding syntax as we figure out what we need. I am not sure if that is the best process, I would love to learn about the better one if there is one, but it worked so far. Wasp is still pretty simple, really not much more expressive than JSON at the moment, but we hope to add more stuff soon.
While we are using parser combinators right now to parse the language, which is pretty manual process, once it becomes clearer what the language looks like, we will look into capturing it with formal grammar and then use some of the more advanced tools to parse it.
The book I am going through right now is "Language Implementation Patterns" by creator of Antlr, it has chapters specific for designing DSLs!
we also have a DSL design channel in our Discord - feel free to join and start the discussion, we'd be happy to learn from you and also help if we can! https://discord.com/invite/rzdnErX
If you’re looking for actual real life footage of 1980s Japan, then I’m not sure this is it. I’ve watched a bunch of the movies where they are ripping the footage from.
All good films, though maybe acquired tastes unless you are into Japanese film. The second is a documentary by Wim Wenders exploring the work of Yasujiro Ozu, mixed with Wenders' visual impressions of 80s Japan.
This is the one headline I’m constantly afraid of reading... do any HN biotech experts have any comments or feedback? Is this a serious cause for concern?
Please note, that neutralization by antibodies is only one of the many possible ways how our immune system can take action. Antibody recognition is still possible with the found mutant:
"To determine whether 501Y.V2 is still recognized by non-neutralizing antibodies, the binding of polyclonal sera (from Fig.2a) to a recombinant 501Y.V2 RBD+SBD1 protein and an RBD+SBD1 from the original lineage was assessed by ELISA (Fig.2b). These data revealed that binding of polyclonal plasma to 501Y.V2 RBD+SBD1 was only substantially affected in a minority of cases (14 of 44 with ≥five-fold reduction, 32%). Most of the convalescent plasma/serum suffered less than four-fold reductions in total binding activity (as measured by area under the curve), suggesting a considerable non-neutralizing antibody component are still able to bind the 501Y.V2 spike antigen"
This is correct, but just want to add that many may interpret this as a positive sign, it is not necessarily. In many cases, binding antibodies increase cell infectivity. Also, there is some research that indicates the presence of non-neutralizing antibodies without the corresponding neutralizing antibodies and humoral response may be a factor in the development of antibody dependent enhancement. As far as we know, it is the neutralization potency that strongly correlates with disease severity.
"On 7 January 2021 it was reported that Pfizer researchers had found the Pfizer and BioNTech COVID-19 vaccine in tests involving 20 blood assays to be capable of affording protection against one of the 501.V2 variant mutations (N501Y, shared with variant B1.1.7). Further investigation was to be undertaken to ascertain the level of protection involved."
The next section in the article is what's important here:
E484K mutation
The E484K amino acid change, a receptor-binding-domain (RBD) mutation, was reported to be "associated with escape from neutralising antibodies" which could adversely affect the efficacy of spike protein-dependent COVID vaccines.[50][51] The E484K spike mutation was linked to a case of reinfection with the 501.V2 variant of SARS-CoV-2 in Brazil, believed by researchers to be the first such case of reinfection involving this mutation.[29] The possibility of an alteration in antigenicity was referred to as an "escape mutation" from a monoclonal antibody with the capability of neutralising the spike protein variants of SARS-CoV-2.[52][53]
I read that. They're only speculating that this could impact vaccines that rely on the spike protein. They didn't actually test for that.
The links I posted show that testing with current mRNA vaccines against this exact variant are under way. Pfizer's (very small) initial tests results seem positive.
1) it's in the spike protein. I think it's in the area that typically elicits antigen response. That's super bad news. If you look carefully, it's near the cleft of a representative antibody (fig 1a), and the buried surface area is high across the board on (clinically?) sampled antibodies (fig 1e), suggesting its importance.
2) it's a mutation that changes the charge state of one of the amino acid residues. E->K, changes a negative charge to a positive charge, potentially very bad if typically antibodies have a positive surface charge to interact with it, now they will be repelled. Since individual humans make random antibodies, even if there is no charge interaction, K amino acid is bigger than E, and so there is likely a shape interaction that could get disrupted by the "new bump" in the spike.
Is it a serious cause for concern? Yeah, almost certainly. The current round of vaccinations will probably be rendered inactive.
IMO the fastest way around this is to quickly approve a reformulation of the moderna/pfizer vaccines that encode for the new amino acid, or better yet, a cocktail of both.
For the more traditional protein-based vaccines, it could be more challenging, because the gut feeling is that an amino acid change like this could pose more of a risk to the effectiveness of the formulation for delivery of the vaccine (mrna is more "neutral" information medium that typically exerts less of a chemical change to the delivered product when altered).
In any case this emergent phenomenon will be certainly challenging to the regulatory landscape around vaccination.
But CAN you safely reformulate vaccines to include this new protein and introduce it widely into the population without knowing if this protein's antigen is already present in the body and might cause severe side effects?
I think you can't. It's famously difficult to anticipate off-target effects of any novel antigen target. This can't be tested in-vitro. You have to repeat the full range of clinical trials with each reformulation that targets a new antigen.
> You have to repeat the full range of clinical trials with each reformulation that targets a new antigen.
Are you sure? I don't think we do this with the seasonal flu vaccine in general. It certainly wasn't a consideration in the CDC pandemic flu response fire drill (sequence antigen -> manufactured vaccine in ~2 weeks) that I got a sideline seat on.
I agree -- I had no idea they were building custom code and delivering it like this... Very interesting, though also makes sense given where they're at.
Thanks for clarifying this. I was worried about it too. Have you considered also providing a marketplace for bookkeepers, etc? That’s another pain in the butt with this process.
It is like you're looking at my pitch deck. We have zero interest in providing professional services but know that some of our customers will need CFO like help and so a natural partnership for us is outsourced accounting firms. Later this year we will open up reviews for bookkeepers so that our customers can interact to make those qualified recommendations to each other.
Canadian startup here. Yes; our general liability insurance has cyber provisions in it.
Usually upper management asking is either because a customer needs certain provisions or they want to make sure a certain class of risks are covered. I’d suggest you clarify what they need and then talk to your insurance broker.
Case in point: general liability could cover a breach but might not cover ransomware attacks.
I’ve seen a few threads like this here on HN. I feel like the most important thing is talk to a professional advisor. Take money off the tank but it’s not a “keep it all in the company” vs “sell”. You might want to do a rebalancing and keep 50/50 or something.
How much of your ownership stake would you need to sell to never worry about working again?
I could 15-20% and not worry about working. My intention is to understand the optimal strategy. Keep the winner or sell shares and schedule and invest in ~50-60 new companies with larger tickets. I love my work and want to keep working.
If I could lock up a guaranteed "this will see me out the rest of my life" amount safely by selling 20% of a holding, I would do that then do whatever I wanted/have fun with the rest. If that fun involved angel investing in a handful of other companies, so be it. Take care of yourself, then do what feels good, I say!
Interesting! I wonder if there are unintended consequences here? Since BERT clusters words based on broader text examples/crawls, can this reduce search quality in niche topics or ones that use a lot of jargon?
I try to turn it into a game — pick a few random words, or a random time frame, or other scenario, and come up with ideas or stories around that. For example “ten years out, agriculture, global trade resurgence”. What happens? What do we build? Etc...
