Weeknotes: old things, bandwidth, AI, social networks
On older things:
A great article about what happens to old (or older) coders. Where do they go?
(Thanks to Garry Bodsworth)
On another level, coders focussed on new frameworks often fail to maintain software and accrue technical debt. This is a lovely clear article about how a team at Wikipedia went about managing their technical debt. Compare to this history of COBOL and its continued use and maintenance...
A new initiative to use documentation as part of raising the potential to secure funding for maintenance for open source, working with Open Collective. Jessica Kerr writes about open source and this phrase struck me: Useful software comes out of a healthy sociotechnical system. I also came across a new open source licence compliance handbook which summarises a (wide) range of licences (and reminds me of the aphorism that one should never write an open source licence).
A reminder about bandwidth from my old colleague Quentin Stafford-Fraser, and remembering that sometimes cost, latency and volume of data means less obvious solutions for moving it about are the best. (Q also played a part in advancing the web, by putting coffee monitoring equipment on it)
On AI:
Everyone worries a lot about Google and Facebook. Amazon is interesting too - there's an Amazon-watching newsletter now. Amazon's system of APIs, of interfaces, which help it optimise internal functions and remove bottlenecks, is how it has become so dominant in retail, and in computing. Zack Kanter examines Walmart first, then Amazon, to show how relentless focus can create incredible scale. He then goes on to look at how ads - Amazon's sponsored product listings - distort everything. I especially liked this bit -
On people:
Extinction Rebellion have a strategy. Lots of detail on the shaping of nonviolent action, the call for citizen assemblies, collective training and community outreach.
An incredibly long but also very interesting post on social media and its role in status signalling. I'm still digesting some of these ideas, like social proof of work. (HT Jacob Ohrvik-Scott) Why do new (distributed / privacy-preserving / secure / whatever) clones of social media networks not succeed? Do networks offer utility before social value, or social value before utility? How do these change over time? How much does management of multiple social identities, and changing identities, matter? What does all this mean for technically-framed ideas like network portability?
This LSE article contrasts the advice given by successful academics around how scholars should engage with policy, and the empirical literature, and policy studies. Cross-sector work is hard, and the 'reckons' of a few people may not be useful for the next generation. (HT Emma Coonan)
A great article about what happens to old (or older) coders. Where do they go?
(Thanks to Garry Bodsworth)
I am committed to a lifetime as a programmer, but my career path for the decades to come is not well-marked. I know disturbingly few engineers older than me whose examples I can follow. Where have all the older coders gone, and what are the career prospects for those of us who remain?Having had some little insight into the world of archive libraries a decade ago, I was fascinated by this article about "data hoarders" - the more informal digital archiving community. (HT Peter Bihr)
Many people active in the data hoarding community take pride in tracking down esoteric files of the kind that often quietly disappear from the internet—manuals for older technologies that get taken down when manufacturers redesign their websites, obscure punk show flyers whose only physical copies have long since been pulled from telephone poles and thrown in the trash, or episodes of old TV shows too obscure for streaming services to bid on—and making them available to those who want them.And they form collectives to be more effective:
Itdaniher, an experienced Linux system administrator, also runs software provided by the group Archive Team to help download materials at risk of disappearing from the internet and help them make their way to the nonprofit Internet Archive. Founded by the digital archivist and filmmaker Jason Scott in 2009, Archive Team calls itself “a loose collective of rogue archivists, programmers, writers and loudmouths dedicated to saving our digital heritage.” Members frequently scramble to preserve aspects of internet history before they disappear as sites fade from the web. Through a mix of manual labor and distributed bots, the project has archived large swaths of sites including the classic free web host Geocities, the text-hosting platform Etherpad and the blog platform Xanga.It seems like archiving content should be straightforward, but it's not:
The fact is, though, it is often genuinely difficult for users without a decent amount of technical experience to find the right balance. Many systems don’t make it easy to find, organize and back up valuable files, while shunting more ephemeral data to the digital trash heap. Social networking sites are notoriously difficult to search, let alone download content from. Cloud services shut down or change policies often with little notice, said the Archive Team’s Jason Scott,Especially interesting in the week where we learned that MySpace apparently lost 12 years of music uploads. Ars Technica reports:
....“We have consistently been working since the mid-80s to turn every single aspect of life into a digital file in one way or another,” Scott said. “People are suddenly discovering they don’t own their data, and all your life is data.”
Myspace has apparently lost most or all of the music files uploaded by its users before 2015, and it told users that the data was corrupted beyond repair during a server migration. Myspace apparently admitted the problem to concerned users seven or eight months ago, but so few people noticed that there wasn't any news coverage until the past 24 hours.We have to get better at technical debt in software. Or perhaps we need to evolve how we develop stuff at all. On one level, we have my former colleague Antranig Basman, declaiming that all software is fundamentally unmaintainable because of the whole system of how we programme, down to the structures of the languages we use. We need to have a longer chat about that idea some time :)
On another level, coders focussed on new frameworks often fail to maintain software and accrue technical debt. This is a lovely clear article about how a team at Wikipedia went about managing their technical debt. Compare to this history of COBOL and its continued use and maintenance...
A new initiative to use documentation as part of raising the potential to secure funding for maintenance for open source, working with Open Collective. Jessica Kerr writes about open source and this phrase struck me: Useful software comes out of a healthy sociotechnical system. I also came across a new open source licence compliance handbook which summarises a (wide) range of licences (and reminds me of the aphorism that one should never write an open source licence).
A reminder about bandwidth from my old colleague Quentin Stafford-Fraser, and remembering that sometimes cost, latency and volume of data means less obvious solutions for moving it about are the best. (Q also played a part in advancing the web, by putting coffee monitoring equipment on it)
On the repurposing of technical concepts to describe other things: "I don't have the bandwidth."
(HT Cassie Robinson)
On AI:
OpenAI (which seeks to ensure artificial general intelligence benefits humanity) needs money [FT, paywall]. Several people have drawn their change from being a pure non-profit to my attention, and it's interesting to see how what seems to me a reasonably pragmatic approach draws criticism for being selling out. The Register describes the concept of "capped profit" investment in a bit more detail.
Sean Mcdonald dives into the natures of AI and humanitarianism - because “AI and humanitarianism are both changing at a rate that is hard to track.”
As with artificial intelligence, the term “humanitarianism” was initially used to refer to something specific—in this case, the international institutions and legal frameworks built to protect the values of humanity, independence, impartiality, and neutrality, during wartime. Over the course of the last 150 years, however, the term has grown in use and application to often include a broad group of state, nonprofit, and corporate actors, whose defining characteristic is that they seek to alleviate suffering in line with humanitarian principles. As a result, the field of humanitarianism is unbundling, not only in the groups it refers to, but also in the contexts in which it’s used, the problems it may address, the groups of people it benefits, and the tools it has available to achieve humanitarian goals. The characteristics of each aspect of that unbundling are important—not only for the way we understand what modern humanitarianism means, but for how we might train artificial intelligence to help. The most important practical observations to make, based on the current state of humanitarianism, is that the field is increasingly defined by a set of contextually interpreted principles, and it is decreasingly manageable through single institutions, or even federations of institutions.
It's a terrific piece drawing out the parallels and the risks of combining Ai in humanitarianism without understanding.
The worth of humanitarianism, like that of artificial intelligence, lies in its ability to set boundaries around, and optimize for, the values that we hold most dear.
A scan of machine learning and tensors and tools and devices from Tony Hirst, containing this tangential gem:
Just reflect on that naming for a moment: Recurrent Neural Network Transducer. I normally thing of transducers as physical sensors (eg things that continuously convert sound, or light, or pressure, or temperature to an electrical signal). Here, we have the notion of a software transducer that turns a signal into a set of meaningful symbols in a real-time conversion stream.
"Because Discord is a chat, not a feed, it’s also free of the algorithms that creators so often bemoan." - I think we too often forget how the internet was when you just had a chronological feed of things. (HT Nathan Matias)
From the Future of Transportation newsletter - highlight mine -
Reports show Uber was 'burning' about $20M per month on its AV program in 2016 (link). If a similar analysis was done on Waymo spending in the same time frame I imagine it would be equal or greater -- and would be labeled as 'investing' not 'burning' but that is the reason why brands mean so much in this era. Finally, I think spending $20M per month is woefully underallocated for the task at hand.
 |
| https://twitter.com/WendellWallach/status/1107644057063485442 |
On people:
Extinction Rebellion have a strategy. Lots of detail on the shaping of nonviolent action, the call for citizen assemblies, collective training and community outreach.
An incredibly long but also very interesting post on social media and its role in status signalling. I'm still digesting some of these ideas, like social proof of work. (HT Jacob Ohrvik-Scott) Why do new (distributed / privacy-preserving / secure / whatever) clones of social media networks not succeed? Do networks offer utility before social value, or social value before utility? How do these change over time? How much does management of multiple social identities, and changing identities, matter? What does all this mean for technically-framed ideas like network portability?
Most of these near clones have and will fail. The reason that matching the basic proof of work hurdle of an Status as a Service incumbent fails is that it generally duplicates the status game that already exists. By definition, if the proof of work is the same, you're not really creating a new status ladder game, and so there isn't a real compelling reason to switch when the new network really has no one in it.I was at the launch of Nesta's new report "imagination unleashed - democratising the knowledge economy." The report identifies huge inequalities in the knowledge economy (the event started by mentioning Cambridge, both at the forefront of knowledge, and "the most unequal city in the UK"), and the challenge of modern progressive ideas being either trivial, or suggesting a far-off utopia, with no useful suggestions in between. It calls for more connection between knowledge and production, more cooperation and responsibility, and a reinvention of the market economy. The speakers and panel were pretty good, although as ever let down by taking unfiltered audience questions. The idea of unbundling property rights, so that control and ownership could be allocated differently for different actions, was interesting. Mariana Mazzucato wondered how we can get beyond cute inspiring experiments in social innovation, to stuff that actually makes a difference. What is the public sector for? What do we mean by public purpose? Are consulting services and think tanks the only way we can figure out what we need to do, hollowing out public purpose? It all sounded good at the time, but afterwards it felt like maybe it had less substance. Perhaps I expect too much of events like this.
This LSE article contrasts the advice given by successful academics around how scholars should engage with policy, and the empirical literature, and policy studies. Cross-sector work is hard, and the 'reckons' of a few people may not be useful for the next generation. (HT Emma Coonan)