741 karma · joined August 9, 2013
It's a really complex system navigating all these different regimes of stability and instability, with variations caused by internal/external forces, yet coming out almost exactly the same every time.
The signal/noise issue is real but as others have noted not the most concerning. The main structural problem IMO is that all the structures that support scientific research (academic evaluation, university admin, granting agencies, publishing industry, etc.) are all re-enforcing untenable illusions. A few benefit immensely from this, but there are many who have the incentive to change, yet they are cornered by their economic circumstances to play along.
It's a thoughtfully executed set of experiments that IMHO should be part of every computational neuroscientist's training.
> boring, dirty and doesn't earn money.
Or even
> If you wake up thinking "Oh fuck, I need to work", yeah then that's a bullshit job for sure.
Instead, it refers to employment that is "so completely pointless, unnecessary, or pernicious that even the employee cannot justify its existence". Prototypical example is the layers upon layers of superfluous middle management.
> The way people in this camp describe themselves is fundamentally pragmatic. They write software because software is useful to other people [...] I think most professional software engineers are in this tribe
Fwiw, I don't think "tribe" is a good word for what the author describes. Maybe "attitude" or "principle" would be more fitting, albeit less catchy and intriguing.
A non-technical detail that the post misses is the human story here. Katalin Kariko, the biochemist that pioneered the idea of delivering vaccines in mRNA form, got nothing but rejections for her grant applications for this very idea and was eventually demoted from tenure track at U Penn.
Whatever you think of this, it's a misleading understatement to describe it as:
> As with other fundamental scientific research we are now reaping the benefits of, the discoverers of this technique had to fight to get their work funded and then accepted.
From the article[0] the post itself cites:
> By 1995, after six years on the faculty at the University of Pennsylvania, Karikó got demoted. She had been on the path to full professorship, but with no money coming in to support her work on mRNA, her bosses saw no point in pressing on.
[0]: https://www.statnews.com/2020/11/10/the-story-of-mrna-how-a-...
Capitalism seems to consistently do the former phenomenally well and make the latter catastrophically worse.
I think that's why we keep having absurd arguments about "the poor having microwaves" while we live in a grotesquely unequal society.
> fundamentally lawless and anti-democratic
If my neighbor, who I believe is a good person, is abusing their spouse, it is absolutely necessary to call the abuse for what it is.
This might be stretching the definition of greed as it's commonly understood.
I've been using Pandoc (and make) daily for over 6 years for all sorts of document writing (letter, report, thesis, design doc, performance review, you name it) and solve the occasional "interesting" format conversion problem. Its robust, reliable, fast, and a pleasure to use (and script).
> the late David Graeber
He died on Sept 2 2020, aged 59, unexpectedly while vacationing. RIP
But does IG allow us to reason about Neural Nets in new ways that could move the needle on open questions about information representation in ANNs or, even better, BNNs?
Given how hard it is to estimate software dev (for oneself let alone for a team), I think the speed with which your team produces quality code is more an emperical observation reflecting your team composition and dev/QA processes rather than a commitment you make. Especially so when the team is new and young.
> or deliver work that ultimately fails the QA process
How many layers of QA/QC do you have? For example, you can think of 4 layers: 0) spec/ticket/story author 1) code author, 2) code reviewer, 3) pre-release manual integration QA (by testers).
These are very different activities (eg reviewer may not actually run code but could be able to tell by looking at code that something's wrong) with increasing cost of fixing issues at every layer. If you leave most of your QA to the last step you end up with a lot of hours spent fixing bugs while you could have avoided most of the same issues (and more) if, say, devs had clear instructions for testing their code and a solid review process.
Granted, it's labor intensive and hard to justify the learning curve just for technical diagrams but I find that once you have built the muscle memory and a good mental map, it's not that slow (and encourages concision and good abstraction to make the diagram simpler) and superior to auto-generated layouts because it gives me the flexibility to make prominent what deserves to be prominent (objects, paths) and complete control over styling. I also hate it when adding a single element might change the auto-optimized layout entirely.
I wonder what folks who use (go-)diagrams at scale think of this. Maybe my solution is untenable for large enough projects.
US foreign policy is not motivated by an empathetic response to address "bombings, massacres, civil wars". That's only relevant in the calculus of managing the public narrative.
Empires don't enter military conflicts to save or better lives. They do it to control resources and counter regional or global imperial competitors.
I suppose you mean the bombed civilians are in Islamic countries rather than the apathy. FWIW, I first thought you meant the latter.
The left sidebar has two killer features for me:
1. Finding canonical translations for technical terms. You need to know what the standard way to translate "hardware acceleration" or "differentiable manifold" is in French? Go to the wikipedia page and hover over the language switcher link in the left sidebar. Done. This has become one of my indispensable tools.
2. Learning about historical events where there are multiple inevitably biased narratives. For example, want to know about the Islamic Golden Age? The English, French, and Farsi versions have significantly different content from different perspectives (I would assume the Spanish and Arabic ones are also equally interesting). I highly recommend this exercise especially for history that one is taught in school, e.g. if you're American and can read Spanish, I bet the Spanish entry on Mexican-American War will teach you a few things you had never heard of.
> We need Howard Zinn now more than ever. Not for the sake of romance or to construct another hero in history. We need his insights, his politics, and his commitment to the struggle for a better world.
Unless your position is that there is such a thing as an unbiased history or that Zinn presented historical falsehoods (not a usual position even among Zinn's critiques), I don't think the valid points you raise on the 1619 project rebut the thesis of the OP.
Doesn't sound like a "deeply pessimistic interpretation of history" to me.
[0]: You Can't be Neutral on a Moving Train, Howard Zinn
From the post:
> With Jenkins, you can try and build a CD platform yourself, but be aware you’ll need a whole team for it, a load of time and maintenance. Microtica enables you to define complex cloud infrastructure through full integration with AWS and Kubernetes. Automate your deployments and deliver on the cloud with confidence.
Not surprising then that it doesn't even mention all the widely used alternatives: CircleCI, GitLab, Travis, etc.
My point was admittedly an aspirational one that requires major changes to the current IP ownership model of (often publicly funded) scholarly work. For example: see how publishers have used the legal system to throw the book at projects like sci-hub or individuals like aaronsw.
I do disagree that the main issue is universities not being keen on having their work circulated openly.
My take is this:
1. Publishers used to have a reasonable value prop, but are now basically robbing the public blind.
2. University admin roles, even at public universities, are more and more occupied by metric-obsessed, run-it-like-a-business types leading to:
3. Academics are stuck on the publish-or-perish hamster wheel that is fueled by a vague amorphous notion of journal prestige, set up and profited by said publishers.
I can't see how we can fundamentally change this situation without either of:
- top-down structural change in academic admin system allowing people to pursue academic careers without h-index and citation obsessions. Major tenure reform?
- bottom-up mobilization of academics or public interest groups, forcing publishers' hands into a more sane (and less profitable) business model. They will fight it tooth and nail of course, hence the need for mass coordinated action (boycotts, public shaming, etc.)
Both these avenues need playing politics on one level or another.
EDIT: formatting and clearer wording.
AFAIK the problem that the OP points out would not have been fixed by Pubmed. It would keep the abstracts, but the full contents would become dead links because the publishers _own_ the full article at the end of the day (which is absurd IMO).
1. cheap it is, considering the size of the institutions that produce and benefit from them.
2. absurdly broken the private publishing industry has become.
1. The importance of a piece of scholarly work need not be immediately apparent to its contemporaries.
2. Being cited is a poor proxy for importance.
3. Work that is cited is rarely paraphrased or duplicated in a meaningful way.
4. Paraphrased citations are a poor proxy for canonical source; papers are often cited incompletely or sometimes outright inaccurately.
5. These are the fruits of people's labour. They spent days and months producing them. To lose them, especially when digital copies are so cheap, is an unnecessary disregard for said labour.
Can't a RF or boosted regression tree theoretically do the same and then some (approximate discontinuous functions)?
But I'm not sure to what extent the language warts are inevitable consequences of what makes it so expressive and powerful. For example: unix pipes are amazing and the bash syntax for them is quite expressive. I can reproduce the same behavior in any of the more "sane" languages but I find that the equivalent code is a less readable procedural soup of string munging, control flow, and error control.