I was recently inspired on this front by an interview[1] with a recommender system engineer that explored the idea of a "good" recommender algorithm:
> It feels like smartphones have saturated the available time. It's like the famous quote from the Netflix CEO: "Our main competitor is sleep." There aren't many more biological hours in the day to capture. At this point, it's mostly a war between recommender systems for your attention, as they've already consumed roughly all the available time.
...
> For me, the big issue with recommender systems isn't that they will destroy our minds, though that is a possible risk. It's the incredible waste of potential. Billions of hours of human time will be allocated today, guided mostly by clickbait incentives. The goal is to entertain people, not in a joyful way, but to help them dissociate.
> You have such an opportunity. There's probably a video on YouTube right now that, if I watched it, would inspire me to call my dad, talk to a stranger, or start a new relationship. Google could probably introduce me to a good friend, a co-founder, or my future life partner.
> They have the data, but they aren't using it that way. Instead, they're optimizing for a few more cents of advertising revenue, which is a colossal, civilizational-level failure.
...
> That is the crux of the incentives problem we've been discussing. One thing that gives me hope is we're no longer in the era of free software. Paradoxically, now that intelligence is cheap enough, people are willing to pay for software. It's more reasonable to charge for a subscription now because you can provide measurable value to someone's life. Paying $10 or $20 a month for a social media service that actually helps you live according to your goals is a much less crazy proposition than it was 10 years ago.
Just wanted to chime in and say thanks for including the full transcript. I strongly prefer reading to listening and so I was able to read this over my morning coffee and send it to friends who prefer podcasts before they started their morning commutes.
And generally, I'm a big fan of the weekly newsletter from successful forecasters. Keep up the great work!
I just got a pair of TCL Rayneo air 2 display glasses since I'm farsighted and my eyes become fatigued after a day of working on a conventional monitor. The increased focal distance seems to help, but the nose piece is weirdly designed and the pressure under the pads becomes a little painful after an hour or two. Also the field of view is too wide and so the edges are blurry (hard to see clock, corner buttons in fullscreen windows, health bar in video games, etc).
Worked great to avoid eye fatigue/posture issues on airplanes though. I'm happy I have them, but in hindsight I'd have gotten a Viture or something with a better nose bridge and a narrower field of view.
Thanks for sharing this. I basically agree, and have a lot to say about the neither-black-nor-white state of academia but have never managed to communicate my thoughts as well as you did here.
I'm mostly out now, but I would love to return to a more accountable academia. Often in these discussions it's hard to say "we need radical changes to publicly funded research and many PIs should be held accountable for dishonest work" without people hearing "I want to get rid of publicly funded research altogether and destroy the careers of a generation of trainees who were in the wrong place at the wrong time".
Even in my immediate circles, I know many industry scientists who do scientific work beyond the level required by their company, fight to publish it in journals, mentor junior colleagues in a very similar manner to a PhD advisor, and would in every way make excellent professors. There would be a stampede if these people were offered a return to a more accountable academia. Even with lower pay, longer hours, and department duties, MORE than enough highly qualified people would rush in.
A hypothetical transition to this world should be tapered. But even at the limit where academia switched overnight, trainees caught in such a transition could be guaranteed their spots in their program, given direct fellowships to make them independent of their advisor's grants, given the option to switch advisor, and have their graduation requirements relaxed if appropriate.
It's easy to hem and haw about the institutional knowledge and ongoing projects that would invariably be lost in such a transition, even if very carefully executed. But we have to consider the ongoing damage being done when, for example, biogen spends thousands of scientist-years and billions of dollars failing to make an alzheimers drug because the work was dishonest to begin with, or when generations of trainees learn that bending the truth is a little more OK each year.
It may have to do with how liability insurance requirements are quite low - generally a few tens of thousands of dollars. So serious crashes (both those involving small and large vehicles) might easily blow past that, causing insurers to charge about the same amount for both.
Thanks for sharing these numbers. I'm actually baffled to see the on-time performance so high and the passenger injury rate so low. I wonder how these numbers would look if they broke it down into greyhound vs. other operations, and whether cancellations are factored into their on-time performance. I've seen verbal aggression on so many of these buses and stations that I can't square it such a low injury rate. Maybe I've just been a major outlier in my travels.
I think you're misunderstanding. GP isn't saying "drawing randomly from all videos of unsafe situations, a large fraction of them occur on long-distance buses". They're saying "drawing randomly from all long-distance bus videos online, a large fraction of them have unsafe situations".
There are a number of travel vloggers who do a trip exactly once and report on what happened. The same vlogger will make videos about taking ships, trains, airplanes, and long- and short-distance buses, in a variety of countries, on a variety of budgets. Among these vloggers, it is generally agreed that long-distance buses in the USA are the worst form of transit in the developed world. Their videos on other forms of travel rarely (if ever) show the kinds of unsafe experiences they have on long-distance buses in the US.
I'm also a numbers person and I've looked around to try and find stats for how dismal the safety and on-time performance of US long distance buses are for you, but none are published. I can just report that the cancellation rate is well over 10%, the on-time performance is maybe around 50%, and personally speaking the experience is frequently unsafe and miserable.
I don't mean to make this personal, but if you're in the US, consider driving to the local greyhound station/pick up and just waiting there for particular bus. It's really one of the worst experiences you can have in a city.
Read my other comment. I have ridden inter-city busses in the US on many occasions and picked up people at the stations. They've generally been on-time. My personal experiences have been pretty alright. A few times the WiFi didn't work but that was about it when it comes to my own negative experiences. The bus stations haven't exactly been in the nicest parts of town, but I've never experienced anything like violence there. Outside of chartered trips I haven't done a multi-day bus trip, most have been straight city pairs. Houston <-> Dallas. Dallas <-> Austin. Austin <-> San Antonio. Etc. But I'm not then saying that's always typical, to actually judge the performance I'd look at the larger statistics.
> the on-time performance is maybe around 50%
I posted Greyhound's on-time statistics which was 90%.
> who do a trip exactly once
What a way to collect statistics. I just stepped outside. It wasn't raining. I guess it'll never rain.
> drawing randomly from all long-distance bus videos online, a large fraction of them have unsafe situations
Well yeah, once again, are there really going to be a lot of popular videos of "I took a five day bus ride, nothing happened, here's an hour-long video of the travel!" And are those really random videos or ones the algorithm have bubbled up to the surface? Think a video with crazy stuff happening would bubble up in the algorithm rather than an hour long video where nothing of note happens? Maybe analyzing statistics by one-off example YouTube videos made for clicks isn't the best way.
reply