He acknowledges the risks and discusses in some detail towards the end of the video around the 29:00 mark. He just wanted to share a tool he found interesting that he was tinkering on.
Was the whole point of this site to make the cars more visible on the map? B/c they're super hard to see! Would've been so much easier to just use a regular Leaflet slippy map or something and just drop an icon for each car... Could've done that in a day. Sigh.
I was looking at the sourcemap and it appears that they wrote their own tiling library (see Tiles.ts and related files under js/). Is this right? Why not just Leaflet since they're rendering PNG tiles? Or Mapbox?
The whole thing feels a bit over-engineered to me at first glance.
I was happy to see the Crisp chat tool in the bottom right hand corner - the folks who make that do a great job. I used it "early on" in the product's life, back in 2016, and really appreciated how responsive they were, etc.
Interesting - would be interested in a comparison between GraphBLAS (which I had not heard of until just now) and, for example, graph-tool's (https://graph-tool.skewed.de/) underlying algorithms (Boost Graph Library).
It seems that they serve very different purposes. Graphblas is mostly focused on the processing of real-valued functions defined on the vertices and edges of large-scale sparse graphs, and is optimized for this use case (think graphs with thousands of millions of vertices). The boost graph library is not tailored to numeric functions and it will probably be less efficient for this use case.
Can we all just acknowledge that the "AI" aspect of this is gimmicky and - ignoring that part - Scribblenauts did this way, way better back in like '09?
Acknowledged. I would say even Scribblenauts might've been actually a better game without this gimmick. It wouldn't have attracted the same amount of attention though.
Quickly scanning comments I do not think anyone else has brought up: administrative bloat.
More money is being spent on science, but is more money actually making it through the administrative bloat encumbering most institutions to the actual performance of research?
Anecdotally, I have a colleague who has received funding from the NSF and the amount of regulations and paperwork and various travel and meeting-related obligations related to the funding soak up so much of the actual dollar amount supplied. (You have to use your funding dollars to satisfy the various required meetings, travel, and paperwork-filling.) The constraints are so ridiculous that satisfying them consumes nearly all the resources the NSF provided, and the little that remains is actually not sufficient to perform the research with. Worse, he has now wasted months of his time satisfying various oversight requirements administrated by both the NSF and the research institution he works in, leaving him an unreasonably small amount of time to actually achieve any significant progress on his work. Once this round of funding dries up, he will be left with no choice but to repeat the process in order to secure some more funding to continue to barely make progress on his stated research goal.
If I had to make up a number to describe the dollar efficiency of research funding, in some cases I might assert it is negative: Not only is it just being soaked up by self-serving, efficiency-draining administrative requirements, it literally destroys the most valuable resource (time!), leaving the researcher with none to actually engage in their subject matter of expertise.
it's huge. Similar in an european ERC-funded lab: the PI is constantly traveling, there is little oversight of the work let alone actual scientific output. It feels like a large portion of the funding is designed to keep a lot of people people busy doing nothing.
Not to mention the fact that the academic institution usually takes a hefty "tax" on any grants awarded. Where I work, they take around 30-40% I believe.
I think overhead rates at most big state schools are around 50%, and I believe they are higher at a lot of elite private universities. This is the link I got when I googled "Yale overhead rate":
If I am reading it right, it says that on-campus research has an overhead rate of 67.50 or 69.0 percent depending on the funding source. Ouch! I imagine that the federal government knows that it is subsidizing universities as a whole through this system and intends to do so.
My sense is that in a given research project, the proportion of people who do actual technical work vs total people getting paid directly or indirectly has significantly reduced. Even worse, the folks who are not actually on the grounds doing the work gets paid lion share. I'd read about studies on how vast majority of university funds are used on non-faculty staff and tuition increases where directly proportional to increase in highly paid non-faculty.
