I also dislike the title ("What we learned when a user tried to load a massive GML file in a browser") - it's hard to imagine that someone who had tried to build a viewer for geospatial data would not already know the answer (paraphrasing, "tiling the data will be necessary to achieve acceptable performance, and there are standard methods to do this")
This one's website (and a dead comment replying to you) suggests that processing the PDF in the browser, rather than uploading to a server, is a point of differentiation.
However, there are older tools that do this, such as BentoPDF (which is also open source) [1].
I was about to link to the same "What's on the Menu" site, which I remembered being an impressive site for library digitization project years ago, but it was apparently retired in January 2025.
Now [1] redirects to [2], essentially an About page with links to the data.
In such cases I generally prefer displaying counts of points in defined areas, rather than using clustering (e.g., when zoomed out, show counts per country, and when zoomed in more show counts for states or equivalent sub-national areas).
> The angle of attack, α, is the angle between the kite’s sail and the incoming wind. As α increases from zero, C_L increases approximately linearly until a critical angle (typically 12–18 degrees for flat surfaces), beyond which the airflow separates from the upper surface and the kite stalls (DT Online, 2024).
The supporting reference is [2]; this doesn't refer to a linear releationship or a critical angle, but does say that the angle of attack is typically 20 to 30 degrees (contradicting the claim that a kite would stall if the angle is above 12-18 degrees).
So I agree that this website does not seem trustworthy. Specific claims may or may not be correct, but they're not supported by the presented references.
The inclusion of references without hyperlinks suggests it wasn't thoroughly checked: they were probably put there by Claude, and as they aren't links the author probably hasn't read them (they could possibly have read them in hard-copy at a library, but given the rate at which articles were produced this seems very unlikely).
(One such reference is 17 - "Weijers, M. et al. “Heat-induced denaturation and aggregation of ovalbumin at neutral pH described by irreversible first-order kinetics.” Protein Science 12(12): 2693–2703, 2003")
It looks like "Amplify Education, Inc" and "Desmos Studio" (Public Benefit Corporation) are separate entities.
The desmos website still shows that most of the Desmos "math tools" still exist under the Desmos name (graphing calculator, scientific calculator, four function, matrix, geometry, 3D), but that "Desmos Classroom" specifically has been renamed to "Amplify Classroom" [1].
The amplify usage guidelines [2] say that "Amplify does not own but partners with Desmos Studio, the maker of a suite of free math tools, including a graphing calculator used by over 75 million people around the world. (See desmos.com for more information.)"
Apparently the "API Layer" is "competitive", with TanStack Query and FastAPI as the leading options [1]. These are not at all alternatives to each other.
PMtiles is often used with MVT tiles, but it can encapsulate a variety of tile types: the current spec [1] has defined tile types for MVT, PNG, JPEG, WebP and AVIF (plus "Unknown/Other").
> I think this phrasing alone says a lot more about you than anything you typed.
I'm not sure it says anything about them: "inferior court" is the term of art for any court whose decisions can be appealed to a higher court [1]. It's not a derogatory term; 'inferior' is just the Latin for 'lower'.
> - Provides stable pricing in shorter term while accommodating price changes over longer term
How? If you pre-pay $5, your account is credited by $5, and when you make an API request you get charged at whatever the rate is for the model you called at the time you used it. You aren't buying some virtual currency or locking in a specific price.
> - Excites engagement
More accurately, irritates customers by keeping their money without providing any service in return.
I also dislike the title ("What we learned when a user tried to load a massive GML file in a browser") - it's hard to imagine that someone who had tried to build a viewer for geospatial data would not already know the answer (paraphrasing, "tiling the data will be necessary to achieve acceptable performance, and there are standard methods to do this")