qfwfq_

Geopandas has had spatial indexing available for quite a long time...

https://geopandas.org/en/stable/docs/reference/sindex.html

I think that the challenge for most is that the PostGIS query planner does the indexing for you in most queries, while a naive all-pairs comparison in geopandas/shapely won't tell you to use the .sindex attribute instead.

But, many R tools are already vectorised, so your shift from lapply() to mclapply() is about as fair a comparison as claiming it's "just" a shift from python's builtin map() to pool.map(). Anybody can play this game, and it's not helpful. I've been using+teaching R now for nearly seven years and the number of times I've used lapply can be counted on one hand.

> just

https://justsimply.dev/

I wonder whether this just creates another "underclass" of scientific laborers... I think we need tool-builders as academics. The only thing that academics recognize intellectually are peers. The rise of RSEs is a kind of intellectual outsourcing... implementation of your science should be part-in-parcel to its creation, not deferred to second-class non-academic "technicians." The friends and former students I've talked to who've entered RSE careers have largely been treated as second-class citizens in a research environment that they are integral to!

I think, like many things, the buck stops with academia itself, its metrics, demands, and incentives. We need more research engineering/science about science academics within compute-intensive science departments themselves. Things like JoSS [1] or Scientific Data [2] are awesome first steps at addressing this.

[1]: https://joss.theoj.org/

[2]: https://www.nature.com/sdata/

This also sounds a bit like Planetside 2 [1], which had a similar structure. A relatively large open world where small "provinces" were contested by factions in FPS King-of-the-Hill combat. This meant that any one province action was a part of a larger "front," across which factions would often mass & press offensives. Capturing the entire map led to some kind of reward, and then a reset iirc.

Nothing like rolling up in an APC with 12 people in voice chat on the tip of the spear, or coordinating an entire battery of MAXs keeping the skies clear. Some of the best gaming in general I've ever experienced. Gradually, though, pay to win mechanics pushed me away, and I've not played since 2014.

[1]: https://en.wikipedia.org/wiki/PlanetSide_2

No, nulls matter a great deal. If you want to test a claim in Null Hypothesis Statistical Testing, the "significance" of the claim is in direct reference to the null. Changing a null will change the significance of the alternative. My favorite statement of this is from Gelman:

> the p-value is a strongly nonlinear transformation of data that is interpretable only under the null hypothesis, yet the usual purpose of the p-value in practice is to reject the null. My criticism here is not merely semantic or a clever tongue-twister or a “howler” (as Deborah Mayo would say); it’s real. In settings where the null hypothesis is not a live option, the p-value does not map to anything relevant.

https://statmodeling.stat.columbia.edu/2017/01/07/we-fiddle-...

Given the explosion in the number of journals and the impossibility of effective peer review, being published in a journal does not mean what it used to. This is part of the material drivers for the replication crisis (journals can no longer effectively gatekeep scientific validity), but it also reflects something real about the practice of science: little social cliques come up with pet theories and, over time, "fight" with these theories on epistemic common ground. The successful ones, we'd like to think, are the ones that last the most rounds in the fight, but that probably only holds in the long run. Contradiction, in itself, is normal (and was before!)

We use jupyterbook (geographicdata.science/book), and it has seriously simplified our workflow. The project is building useful features very quickly, and is very responsive to feedback & requests. Big props to their team.

I disagree that it's a "true horror to use." The set of supported built-in classes grows significantly by the day. It's not as good when used for wholly unstructured streams of data (e.g. tuples of mixed type, dicts with complex objects inside of them), but if you can spend the design time to arange things in a structured manner, it's super easy to use, and can seriously boost performance on simple algos.

I've had a ton of success using it in statistical and computational geometry applications.

> This situation is very annoying, especially for a touch typist as my fingers are always on hjkl and my thumb on the spacebar. This makes my thumb knuckle constantly brush the trackpad and activate it.

I thought standard touch typing position was index fingers on the notches (f & j in US QWERTY). With that & some fairly large hands, the trackpad rests comfortably outside of my palm area? the trackpad seems pretty well-designed to just miss the palm in standard position.

I think this is exactly how R's accreted features to make it do things that it was never designed to do. When the "tool you use" becomes the "kind of developer you are," things get a little restrictive.

Cosmicomics has long been my single favorite collection of written work. Calvino's amazing.

The efficiency gap is just one measure of partisan bias that can be used to assess the utility/efficacy of any redistricting plan. The game the authors propose is (basically) an algorithm to draw good districts between two players. So, there's no real or expected relationship between the two... partisan control (the objective the authors cite) isn't necessarily related to the efficiency of that control measured by the efficiency gap. The problem is that so many different measures of fairness exist, have validity, and can be used in different legal or jurisprudential contexts, we're always talking past each other on the meaning of these measures.

I've also seen this happen, where an article was designed in pre-consultation with a journal editor in my field. We submitted (what I felt was) a dubious paper initially. Reviews were very harsh, but the paper was not rejected. I was amazed we had four months to revise and resub. It was eventually published. I was an early grad student at the time, so I didn't know that this was unusual.

I think that, once the suggestion of replacing single member elections with anything approximating proportional representation is made, the focus on managing the electoral system we DO have is lost.

Tons of measures made by political scientists about gerrymandering implicitly value PR or nearly-PR as the standard of justice and fall into this trap. But I think there's opportunity in the SMD systems for fairness, like the Gelman-King stuff that looks at equalizing the winners bonus provided to parties. That is, taken alone, it doesn't matter if seats are awarded disproportionately to votes won, as long as the system responds in the same way when the tables are turned.

I'm a PhD student studying gerrymandering, and I think it's cool this is getting written about in the NYT. But, there are very good models of gerrymandering built off of well-informed counterfactual analysis of generalized linear modelling that can provide both easy to understand indicators of gerrymandering and bias, as well as provide an indication of the uncertainty in those measures.

One thing that's been getting on my nerves in the road to the big D(issertation) is the complete lack of value some in the domain place on that uncertainty.

Elections in single member district systems, at the geography where gerrymandering study becomes meaningful, has to be modelled as a Stochastic system. These kinds of "accounting methods" can vary wildly in their estimates from election to election, and almost always fail to really express how uncertain the indicator is.

I've seen quite a bit when porting a statistical library that used Python 2's default floor/int division behavior for quite a few operations.

Sometimes, though, the 2 version had bugs because of this behavior. So, the bugs were both subtle and bi-directional. I definitely fault Python 2 for allowing that practice, and am glad we're now using Python 3.