Archives, page 11

Ranged Computation and Parallelism
Date: 2018-Dec-17 20:48:04 EST

For awhile I've been interested in mathematics on (mostly continuous) ranges, and thinking about building a software package that makes that easy; today in a "tea and chat with the rest of the scientific centres" informal event we have every week, someone was describing simulations of large numbers of molecules in protein folding, and a use for such a library came into mind. The challenge as described was that it's hard to parallelise such structures because each state strongly depends on the prior states. The idea then would be to bucket the possible ranges of values that come out of each step and use your paralellism to calculate out buckets of further steps (using bucketed mathematical operators), and have a "main thread" that pushes forward through those states determining what actually happens, pruning the speculative branches and their children out, and proceeding down the pre-eased pathways. The challenge being making sure that stepping through that pre-mapped branch is more efficient than direct calculation can be, and/or that the range of buckets "within range" of the current verified path has more value than the broader search space (or some simpler way of getting the narrowing).

Fun set of thoughts. No guarantee the basic idea is useful for anything, but I still like chewing on it. Maybe I should actually make that package. Although perhaps it already exists; I doubt I'm the first person to think of this idea.

A life of Purpose
Date: 2018-Dec-23 20:20:20 EST

Now that work for the year is done, and I've used some rarely-spent vacation days to round off a partial week during the new year up into 2 weeks off from work, I have some distance to wind down 2018 and think about career and life and all that. At least, alongside a few remaining work tasks (moving some data around, training a neural network on some GPU nodes that might be very free) and some chores and my own unpredictable health.

I'm glad things have turned up as they have careerwise. I'm still not fast to make ties to people, but it's actually happening at work I think, and more easily than I had in the private sector. There's just a big cultural gap there when I'm out in industry - I'm motivated by different things, I'm not competative, and the main thing that drove me there in practice (the Creed of The Right Thing, and the secondary Creed of Keeping the Ship Afloat) was more a vehicle for exhaustion than anything else. Whereas here, I have a different set of struggles that are survivable (things like staying awake in long lectures), coworkers I can understand, and a place where my actual drives can come out productively. Plus I am exposed to ideas that make me think very hard (a recent internal talk about memory effects in gravitational waves actually made me feel guilty when I realised that while I could kind-of understand what was being discussed, I don't see myself incorporating those ideas into my mental model of the cosmos yet because I have a tough time believing it - I criticise others on this for easier things). I ask questions often in meetings, without shame. And I feel that I'm growing (even though not always at the rate I'd like).

Despite the horrors of watching populists on the right and culture warriors on the left degrading our loyal opposition and the wing I fit best in, this is still a great time for science, and I have pretty great seats and access to the pit for that.

The cats? Still great companions. The apartment? Still acceptable. I sometimes feel the flow of time heavily, but it's no disaster.

I'm sometimes weirded out how so much of historical linguistics is based on reconstruction based on forms; I know it's probably the best we can do, but it feels very speculative. Even history relies heavily on direct artefacts.

Recently been thinking about how often mediocre ideas end up packaged in a way and boldly proclaimed that they get undue influence. Particularly but not only in tech. Makes me worried about packaging over content. I've been pushing against this for a time, in that I've been suggesting that teachers should read source material, think about it, and then much later without rereading the sources, re-teach the material, based on the theory that great ideas will survive restatement, and anything else is just poetry. Recently someone else in the ChaosEng community was weirded out by my not defining things quite the same way as some book, and I offered this as a response. Not sure if he was persuaded, but I think he understood my why and that's good enough.

Yearend Migration
Date: 2018-Dec-31 19:47:02 EST

Took a 2-day trip to Boston by train; earlier than I expected, but those expectation had not yet met the reality of my sister's schedule. Was great to see my little nephew there. He's at an age where his personality is substantial, but still governed by an impulsivity and still mostly on the surface. I think reality is already starting to change that ; the endlessly repeated process of contrast between expectations of an event and how good it actually is inevitably tempers and matures impulsivity. It's a pity that that process is little different from injury, and the deepest thinkers you'll usually meet are adults who have undergone a mix of pain and time to think about it. But still, I don't think these traits can be achieved any other way.

Near the end of the trip I felt a mild migraine coming on; it was present for much of the train ride back and stuck with me through the night even until now, where I've reluctantly left the apartment, preferring relief from the boredom a bit more than the safety of having better coping mechanisms for pain at home (e.g. the boiling shower). It being the last day of the year, it (and the first few days of the new year) is a good time to contemplate). I think I've gotten over the annoyance at attaching significance to markers of things rather than the things themselves, or at least I'm less than abolutely discinclined now. Or perhaps it's just a new way of parsing this that attached to a dreak I woke from this morning (still in mildgraine):

Was a fantasy setting, and I was with one other person (I think her voice and visage were that of Yennifer from the Witcher series, although I don't think I was Geralt nor would it have made sense). She was criticising me because when I was learning magic, I complained that a spell to raise the dead really just summoned some herbs and applied some supernatural shielding and the gods actually did a lot of other things in raising the dead based on the circumstances the spell provided. She felt this was pedantic. I argued back that pulling a rope to summon a servant to prepare dinner didn't actually mean I should claim credit for having made dinner, and she said that the example actually fit her point perfectly, in that if something is fully mechanistic, digging into the details is something only philosophers do and it is a waste of time. And now, when I look back at the tack, I see that the attempt to shift from causality to credit was a sly move that I should have been called on (although I don't think that happened in the dream, although a lot of memory of dream arguments is about summoning a perspective and letting it speak, so there may not be a clean distinction here)

We went to a Lego museum in boston, which was more like a small single-building amusement park than anything else. It was very well done, and it looks like the Lego of these decades understands marketing in ways that the Lego I remembered did not. They've made something called Ninjago which my nephew is really into, and as I already knew from the videogames, they've done crossovers with a lot of other fantasy in making lego-branded versions of the franchises. I amused myself while there in imagining things they did not do - Lego Spaceballs, maybe Lego Occupy. The latter would probably make people the most angry, as it would be the taming and commercialisation of a social movement that tried very hard not to be those things. And even though I was part of that movement, I still think it's important to trample on all sacred ground at least symbolically, so I'd love to see it. Kind of like buying a Che Guevarra T-shirt at the Gap.

Along those lines, I think I've found a way to, if we wanted, take the piss out of the extreme Russian nationalists dominating that nation right now - while Aleksandr Dugin's philosophy is transparently juvenile, I don't think an academic criticism of it would likely be read by many (or if it would be, it would have to be immaculately crafted to generate an alternate path to pride for Russia to be an effective bridge out of regressivism). Instead, taking from satiricists like Kirill Eskov, and from repurposers like Disney, we could alternatively turn classic Russian folktales and literary works on their head, and lift them while purging them of their Russian setting and claiming them as Western. Being a celebrant of cultural appropriation, I'd have no hesitation in the act itself, although I wonder if it's a good use of my time and I haven't angled my life correctly to execute the idea well.

Been thinking more about the content trap in stories and games; there are fantastic films (like The Quarrel) that consist entirely in dialogue between small groups of people, and to be really immersive, those kinds of things would have to be possible in any game, and different as per the person. But that would take an amazing amount of writing even for two people, and things multiply out. So the broader the world, the thinner each part of it. In theory with either enough actors and writers or in an MMO people can make their own content, but that's increasingly unlikely to make an experience for most people. It reminds me though of some people I became close to in a game ages ago - there was a German woman and her husband (initials KM for self-reference) that I got to know in Dofus and had long conversations with. I regret now not trying to build the kind of friendship where we'd go between long successions of games together, the two or three of us.

Hoping my migraine fades soon. Hard to think right now, although I at least have notes of thoughts I wanted to write up to guide me through this right now. I want to preserve/restore/grow, whatever it is, the habit of writing often for more than just my own consumption; my life is too lonely for my existing social ties to exercise my expressive capabilities, and my notes-to-self are too immersed in my own mental mnemonics. There's nothing wrong with that and I accept that probably the vast majority of the things I write will take that shape, works mostly for myself, but I feel it's healthy to at least mildly counterbalance that with something like this. And as G+ is gone, this is the way to do it.

The walking sometimes lessens the pain, so I should probably finish my tea here, possibly do my weekly financial recording into the spreadsheet, and go get groceries before I head back. I dread climbing those stairs with a migraine, but I signed up for this when I left the house today, and before when I signed for an apartment on the 5th floor. So I should soak in my choices.

Procrastination and Fulfillment
Date: 2019-Jan-01 04:29:02 EST

Finally I can be pretty sure this is the last blogentry for the year. The migraine is mostly cleared. Still not feeling great, and in a better world I would have at least went to the Brooklyn Bridge, but at least there's this.

I feel a shared negative experience in having watched the horrors of American misgovernance this year. I keep feeling weird given how unhappy I was with BushJr's presidency, and again being in a state where I'm aghast, but I both misjudged that man and didn't appreciate that we could ever elect a Berlusconi-like figure (and that people I know would support that). Watching things run this way has been exhausting, and I get the feeling that it's been like that for a lot of people. And unlike a number of people I've known, there isn't a hero narrative running against this in that I absolutely loathe a lot of the noiser voices on the left that have continued with their culture wars, largely over trans issues and other excuses to restrict speech. I don't want them to win either, either on the sly (through sneaking codes of conduct and IBT crud into workplaces) or directly (through dominating mainstream politics, which right now I don't think they have the strength to do). I live in a world with no heroes, very few robust liberals who are also deeply committed to free speech, the battlefield of ideas, and mocking everything. That makes this period harder for me, I think. Even though fewer of my values are trampled on.

But I'm still very happy to have coworkers that I largely get along with, people I respect, and people to mentor. And I'm glad to have a mission that I believe in and a loose leash and trust to pursue it. My personal life really sucks, but my work life is fulfilling. I also like having family nearby-enough to visit.

I've been thinking a bit about counterfactural morals - some years back I was kinda-sorta into a TV series called Once Upon a Time (that started good, and a few seasons in kinda got bad). Family is a strong theme in the series, and in the backstory, there were parents that, it having been foretold that their daughter could be a great saviour or a great blight upon the world, found a way to shunt the latter possibility off into another young child (whose mother was already a nominal villain) to guarantee their daughter a great fate. I am intrigued now to step back and reflect how that shunting seems so obviously morally wrong, despite it being based on a number of things that are not real. Should that be possible? There are some areas of counterfactual (or at least have-not-done-it-yet) like time travel that would require a deep rethinking of a lot of morality if the capability became commonplace, but this isn't one of them. Maybe because even though that alter-your-childs-fate-and-doom-another thing is counterfactual, it's similar enough to either existing systems of moral reasoning that it invites easy reasoning-through-analogy. Although I think were I to need to explain this sufficiently clearly to be predictive to a non-human AI, I might have a tough time of it. That's often the bar I use - explain to an imaginary AI or to an entity that has lots of data but little deep comprehension why we believe something. In this case though, the start of the journey comes from understanding one's journey through life as being a tension between self-determination and a societal fabric of mutual responsibility.

A final quick take of the year:

  • When is it OK for archaeologists to dig up the dead? - My answer: Absent any explicit permission, when everyone who directly knew them in life is deceased. This is one of my standard metrics for when a number of concerns become irrelevant - I reject a lot of claims of historical injuries when all the direct victims are no longer alive, even when there are downstream effects. Time washes everything away, from interests to regrets.
And that's all. It's time to get used to writing 2019.

Finding the End of the String
Date: 2019-Jan-12 18:43:18 EST

Today I finally wrapped up some low-grade detective work - some months ago I first heard of a programme called Gangstas to Growers, based out of Atlanta. The idea being that they offer agriculture-centric opportunities to people after they leave prison in an effort to fight recidivism. I believe this is a great thing, but the first few times I went to look for a way to offer them financial support, I wasn't able to find a place to do that or even contact info. So I eventually forgot about it. Late last week, when looking at an upcoming talk where I work, I was reminded of G2G and spent some more time digging, this time focusing on the name of the founder. And with enough poking around I finally found a contact address and dropped them an email. Today I got a response and was able to offer support. Finally. Although this was more difficult than it should've been.

Next week a coworker and I are heading to Janelia in Virginia to learn more about some of the tools we're using. I've tried to make our collaboration non-draining to them by being a good coding citizen ; I've been doing a series of diffs to add type-annotation to some of their python projects. It's a bit challenging in some ways because I don't know Django and they also use some language features I'm not familiar with, but in LINK I'm discovering that maybe I should have been using them all along. Function decorators are pretty cool, and the way it's used in this codebase is to extract boring and repetitive permissions-checking from dispatch functions; the wrap-up is pretty elegant. There is a cost though, in that you no longer can tell at a glance what parameters a function uses. Probably worth it, although it's good to step forward in ways that recognise this pain. Feels like a weaker version of closures. The trip should be interesting; I've heard Janelia described as a sibling institution to where I work, and it'll be good to meet in person with some people we've collaborated with for the time I've been here.

From a number of recent internal talks, I've come to revise my earlier impression of redundancy in codon encoding, in that I've come to appreciate that the chemical-structural encoding of base pairs itself can be significant to gene expression (from "will this ever actually make the protein" to "in what cell lines will this see expression"). That dimension was entirely absent from my earlier understanding, and makes me appreciate that genetic engineering is going to be more challenging than I anticipated. There may be some ways to get around that (in that rather than doing direct low-level encoding, we may treat sequences as a compile target with extensive compiler layers between). Happy to have this chance to revise my mental models.

Idea that Cortez should hold back and learn from seniors in legislature - interesting. Not sure if I agree. On surface it seems good, but when people elect a representative it's not the same thing as joining a business. Yes, there is wisdom to be learned and excess passion to be replaced with prudence, but this is highly informal and unlike employment where these relationships are implicit and intended. When I vote for someone new to office I'm not asking them to be timid.

Still very enthused at the 2018 highlights book I made from my Google Photos; I know plenty of other photo services offer this exact thing, and it's the kind of thing that's a win for everyone. I'm likely to go back through photos from prior years and have highlights books made for each of those too.

Been thinking again about how I'd design a home if I got the chance. In the more distant past I've dreamed of converting an observatory, with a raised middle section and four separate lower sections for different areas of life. Given that I'll likely never have such a home, maybe I'll do a more realistic design this time. Partly because living in NYC means perpetual dissatisfaction with living arrangements.

Ancestral Home of the Future
Date: 2019-Jan-19 15:00:11 EST

For Thu-Fri of this week, a coworker and I visited Janelia research lab in Virginia. Our project is heavily tied to them - we got our dataset from them (from a specimen that in turn was collected by another overseas collaborator), we use a lot of their software, and our PI came from there. It was fantastic to sit down with people we knew as well as people doing similar tasks to ours (on a bigger scale out there). The campus is also amazing, and a bit surreal. In some ways it's the kind of place I've long dreamed of living in - secluded, academic, embedded in nature. Although I realise that I may have too much noise in my soul to ever be content in such a place for more than a few months. Still, it was a great place to visit and I hope we get to do it again at some point. And following my habits from visiting the mothership when I worked at Dropbox, I had a big list of things I wanted to learn from each meeting and made sure I got through most of them, leaving me with a lot of followup for next week. There was one thing - the sharp distinctions from the inside of the building to the outside led me to fantasise about the facility being on Mars instead of in the woods. I think that would work and it'd be pretty amazing.

I had a migraine for much of Friday, sadly. And although it was mostly bearable, it started to get pretty bad near the end of the day, making the transit home nightmarish. Migraines are already pretty nightmarish to begin with, but mixing in travel makes it the worst thing I've experienced; this has occasionally happened before. Glad it's over now, but while that's going on, it's one of thise "I wish I could just die right now" kinds of things. Not sure, were an easy way to actually do that present itself during the travel+migraine experience, that I'd actually resist the temptation.

Deeply fascinated with this bizarre star system. I love how much we're learning about the quirks that nature occasionally produces. I was also thinking on the train ride back, in the tail end after the migraine ended, how I'm actually proud of our star for being massive enough that we think it will fuse helium for awhile. Go Sun.

I've been pretty unhappy about the direction one of my former employers, MongoDB, has taken with their recent license change that no longer makes them simply OpenSource. They're using their license to prevent competition, based on the idea that only they should be able to provide MongoDB as a service. As a result, while Amazon's new compatible database as a service is also not opensource, I welcome it because it still provides some level of alternative to MongoDB and we need that now. More ideally the community will either build more ways to break out, or we'll find a way to kill the company. While I have a lot of fondness of a lot of people I knew while there, I wouldn't mind the latter. I believe it needs either a lot of changes (starting with the CTO and this license stuff) or to end, from the perspective of industry-openness and good governance.

I was weirded out to see the ACLU call for big tech companies not to sell face-recognition to governments, in that it's a pretty clueless thing to ask for on the tech front. The technology is no longer in the "you need a dedicated research and development staff to spend years" phase. It's in a "you need a small number of the right academes and a few engineers" phase - probably 10 or less of the right people to produce something good in a year. And nations can just hire that if they want to. Meaning there's little point in pretending it can be capped. As of right now, it's going to work as well as the surface justification of crypto bans on repressive nations - there's no way in the modern world to keep that code out of a nation (the actual/functional useful thing it does is make it hard to do business with those nations by preventing businesses from using those technologies across those borders).

Looking forward to catching up with someone I used to work for early next week - haven't seen him for around 17 years. I guess I'm old enough now that thse numbers are a thing. Still seems weird to me. Also going to an IQ2US debate (my first) at the end of the month.

I'm wary of Cortez's efforts here to elect more people like her until we see more of how she's going to function in office. Part of it is that she's so new, but part of it is also that I'm hostile to some forms of progressivism and I'm not sure whether she is (or will become, or will remain) in that "we must redo culture and weaken free speech" crowd or not.

I'm disturbed by China's control over Churches and need to control narratives within its nation. It's not that I think Christianity is that great to begin with - it's a largely shallow philosophy tied to a fairy tale - but even highly disturbing things in that shape (like scientology or objectivism) should be evaluated (and ideally rejected) for what they are. Revisionism is terrible enough that even if there are legitimate social harms from a perspective's power, it outweighs them.

Linear Shadows
Date: 2019-Jan-20 05:02:03 EST

Tonight I went to a 59th birthday party for a longtime acquaintence. It was impressive how many people he's met in the NYC area that came out for him, and when he talked about what he did this year, I saw he had a lot to be proud of. It was nice. And it made me a bit jealous. I think were I to try to do the same thing, it'd be a nearly empty room, or alternatively heavily dependent on coworkers.

Had some good conversations with some of the other people there; he's a moderate conservative, and there were a number of others of that sort at the gathering. As I'm the sort of liberal that prefers the company of moderate conservatives over that of radical liberals, this was kinda refreshing - some of them were trying to find ways of buiding consensus on dealing with climate change on their side. They complained that the Kochs dominated the elected pols and believed that many conservative voters share their views. I can just wish them luck ; I believe our nation only functions well with two main parties that see each other as the loyal opposition, and which prefer reaching across the aisle than reaching out to their own extremists. And while perhaps I am an extremist in some sense on economic issues, I'm pretty moderate on social norms, and that's where I think this is most important.