These are ~2 years behind state of the art from the looks of it. Still cool that they're releasing anything that's open for researchers to play with, but it's nothing groundbreaking.
No, it is not as good as Veo, but better than Grok, I would say. Definitely better than what was available 2 years ago. And it is only a 7B research model!
> Are other open weight video models also this small?
Apples models are weights-available not open weights, and yes, WAN 2.1, as well as the 14B models, also has 1.3B models; WAN 2.2, as well as the 14B models, also has a 5B model (the WAN 2.2 VAE used by Starflow-V is specifically the one used with the 5B model.) and because the WAN models are largely actually open weights models (Apache 2.0 licensed) there are lots of downstream open-licensed derivatives.
> Can this run on a single consumer card?
Modern model runtimes like ComfyUI can run models that do not fit in VRAM on a single consumer card by swapping model layers between RAM and VRAM as needed; models bigger than this can run on single consumer cards.
The interesting part is they chose to go with a normalizing flow approach, rather than the industry standard diffusion model approach. Not sure why they chose this direction as I haven’t read the paper yet.
I think you need to go back and rewatch Will Smith eating spaghetti. These examples are far from perfect and probably not the best model right now, but they're far better than you're giving credit for.
As far as I know, this might be the most advanced text-to-video model that has been released? I'm not sure whether the license will qualify as open enough in everyone's eyes, though.
I wanted to write exactly the same thing, this reminded me of the Will Smith noodles. The juice glass keeps filling up after the liquid stopped pouring in.
There seems to be an easy solution: use a password manager and save the answer to the question as an additional password.
(This is actually a FR to any password manager's product team: it's time to treat things like 2FA recovery code and secret question answers as first class citizen in your product).
That's what I do as well, but that defeats the purpose of the secret question being something only I know and will not forget. And that's because I am aware of the flaw of this system; someone naive might actually fill out that question with the honest answer and leave himself wide open to being exploited. Password manager are not a solution, they are a band-aid fix to a problem we should not be having in the first place.
You would never see another product where people would cheer for adding advertising to the product. Like everyone has an equity in this product. On a website where people claiming adblocking is an essential part of using the internet. Oh the irony.
Perhaps, but with more transit options that means fewer people on the road which is good for those who have 2+ children to lug around.
On a side note we should drop the public bit of this because it implies a bus is “publicly funded” but highways aren’t. Both are subsidized by the taxpayer.
Why are you equating busses to roads and not cars? Cars are not subsidized and in fact car-related taxes (vehicle sales, gas tax, yearly registration fees, in some cases tolls) have historically covered the majority of roadway infrastructure costs. Without car related taxes, we would absolutely need to charge bus fees to subsidize roadway costs, and they would probably need to be pretty steep.
Most states fail to collect enough in user fees to fully provide for roadway spending. This necessitates transfers from general funds or other revenue sources that are unrelated to road use to pay for road construction and maintenance.
Only three states—Delaware, Montana, and New Jersey—raise enough revenue to fully cover their highway spending. The remaining 47 states and the District of Columbia must make up the difference with tax revenues from other sources.
The states that raise the lowest proportion of their highway funds from transportation-related sources are Alaska (19.4 percent) and North Dakota (35.1 percent), both states which rely heavily on revenue from severance taxes.
that's about what I expected. And that's not even including sales tax from car purchases, and maintenance related spending. Suffice to say, without cars, a year bus pass would need to run ~ however much the average person spends per year on all car related taxes.
Well it wouldn’t because we wouldn’t have as many people driving cars, so there wouldn’t need to be as many roads so costs would be much lower.
In Ohio we just spend $2bn on about 2 miles of road to effectively temporarily ease congestion. That’s $2bn paid for by taxpayers regardless of how it’s paid, that we didn’t necessarily need to spend.
I’d also like to add, yes that “bus ticket” (I’m no fan of busses for short term travel) might be a little more expensive but consumer costs overall would’ve likely to go down. Why? Well in addition to already paying for highway infrastructure you’re paying $30,000, $50,000,
&c . on a vehicle, plus insurance, gas, repairs, tires, maintenance, interest on loans, &c. So while I think it’s hard to compare apples to apples, I think it’s good to have this information in mind as well when discussing this topic broadly.
Thanks. Yea also not accounting for other social costs - obesity, teen deaths, first responders and police spending time rescuing people who are maimed in car crashes.
There are benefits too and all, just saying we don’t really have a full cost readily available for comparison because it’s hard to measure, never mind the literal dollars and cents that go into funding.
> On a side note we should drop the public bit of this because it implies a bus is “publicly funded” but highways aren’t. Both are subsidized by the taxpayer.
Arguably, neither of them should be. Give poor people money, instead of giving free highway access (and bus transit) to rich and poor alike. Rich people don't need our help, and poor people would rather have the money to spend as they wish instead of other people deciding for them what alms they should consume.
Individual cars have worse externalities than busses, so that means we should tax them more than busses. Though I suspect once drivers of cars and busses are paying non-subsidised prices for road access and fuel, busses will naturally look better in comparison, no extra tax differential needed.
The poor I want to help the most are not mentally able to handle money. I know someone who gave money to 'nigerain prince' scams several times - a nice guy but he has no idea scams exist even after that.
Not exactly because as soon as you realize it doesn't solve the most important problems the a side effect of giving to well-off people as well becomes easier to argue against as well. If we need to identify those who cannot handle money then we may as well use that same effort to identify who doesn't need help in the first place. Not that your point doesn't stand, but it isn't as powerful.
Note that I don't know how to identify people who cannot handle money. I know individuals, but how to you fairly do this in a way that doesn't get abused or abuse someone - both have been major problems with every plan to help the poor in the past.
> Note that I don't know how to identify people who cannot handle money. I know individuals, but how to you fairly do this in a way that doesn't get abused or abuse someone - both have been major problems with every plan to help the poor in the past.
Welfare (nor any other undertaking) doesn't have to be perfect. Good enough is good enough. And you can always optimise over time, as you learn. There's plenty of programmes out in the real world and throughout history to learn from: the good programmes as examples to perhaps imitate, the bad programmes as illustrations of what to avoid.
> If we need to identify those who cannot handle money then we may as well use that same effort to identify who doesn't need help in the first place.
Isn't that perfectly compatible with what I'm suggesting? (Though I'm not quite sure they are actually the same kind of effort. Looking at eg people's taxable income is rather different from judging their mental state.)
I'm making two points:
(A) Don't bother giving welfare to the well-off.
(B) Prefer cash payments over in-kind provision, where possible.
You convinced me to add the 'where possible' clause, because there are some cases were people can't handle money. Though I hold that these spendthrifts should be dealt with as a special case, and not be the default template for how we treat everyone else.
> You convinced me to add the 'where possible' clause, because there are some cases were people can't handle money.
I'm not really in favor of or against UBI, but I think that your assessment here isn't quite right. It's not that there are some cases where people can't handle money, it's actually the norm that people can't handle money, at least in America. And when you expand your scope to include folks with steady jobs and such and "can" handle money, they don't really make enough money such that their better handling of money will make a difference either. UBI kind of rests on this assumption that people will "spend their money wisely" or spend their money efficiently - but people will spend their money in obviously stupid ways and then we'll deal with the consequences anyway because as a society we are not willing to let people die on the street. To be fair, not all who would make use of UBI would do so in a poor fashion, and like you said don't let perfect become the enemy of good.
Most people who are competent and kind enough to look at these kinds of problems and want to provide a great solution for them out of the kindness of their heart, in my estimation, just do not have experience dealing with the target population and understanding the true limitations of just giving a little bit of help.
If you grew up in a roughly middle class environment, went to college afterward, etc. and you didn't know any crackheads in your family, you probably should stay far away from trying to find solutions for these problems because you're just going to wind up frustrated and wrong about worthwhile solutions for the rest of your life which will distract anyone from actually making progress. I'm not suggesting this about you or anything, just speaking broadly. Most of the time people working on these issues are like product managers who just care about what the data says instead of having a really good intuition for the "customer" or the "problem".
Back to UBI itself I actually think it should go to everyone, because at the end of the day those who are high earners are just going to net out paying for it, and it'll be simpler to administer if "everyone gets it" and there's less room to complain about it and less hand-wringing and adjustments for whatever excuse people come up with for who gets some and who gets more. If everyone gets it, there's much less arguing about that stuff.
The best part about a potential UBI implementation is that we can replace all of the government workers administering and declaring who is eligible for benefits whether that's SNAP or Welfare and one dude can be sitting at home and push the big red "send checks" button with an offshore team in India in case that guy is too drunk to do his job.
Maybe just providing free healthcare and dental care, job training and education, and a sack of fresh produce is the way to go too. I'm not sure.
I'm not sure why you bring up UBI? Government welfare for poor people in eg Germany (and many other countries) is mostly handed out as money, but it's not a UBI. Whether a UBI is a good idea is indeed a wholly separate discussion.
> If you grew up in a roughly middle class environment, went to college afterward, etc. and you didn't know any crackheads in your family, you probably should stay far away from trying to find solutions for these problems because you're just going to wind up frustrated and wrong about worthwhile solutions for the rest of your life which will distract anyone from actually making progress.
I grew up on welfare in the place formerly known as East Germany. Bleak times. Not sure that will get me past the gate you are keeping here.
(As late as 2006 a government spokesman warned foreign football fans coming for the world cup of no-go areas that they should avoid, if they want to survive. Compared to the 1990s, the late 2000s were the Good Times.)
> Back to UBI itself I actually think it should go to everyone, because at the end of the day those who are high earners are just going to net out paying for it, and it'll be simpler to administer if "everyone gets it" and there's less room to complain about it and less hand-wringing and adjustments for whatever excuse people come up with for who gets some and who gets more. If everyone gets it, there's much less arguing about that stuff.
Someone has to pay the taxes to finance the redistribution. The average person can't get more out of the system than they pay in (by definition of how averaging works).
I would suggest to pay attention to the net effective marginal tax you charge people across the spectrum of incomes. Basically, for every level of income ask: taking account of welfare phase-outs and marginal income tax, how many cents of net income would a person get for an extra dollar in gross income?
This effective marginal tax rate matters a lot more than how you split the rate between welfare phase-outs and other taxes. Yes, for simplicity you could have no welfare phase-outs (aka UBI) and create the whole shape of the taxation graph simply with progressive income tax rates. Or you could merge both systems, and call your UBI a negative income tax.
In any case, what you want to really avoid are sudden cliffs, where an extra dollar in gross income costs you more than a dollar in net income. And ideally, you also want to avoid unduly high marginal rates (even if they are still below 100%) from phasing out multiple, uncoordinated welfare payments.
> The best part about a potential UBI implementation is that we can replace all of the government workers administering and declaring who is eligible for benefits whether that's SNAP or Welfare and one dude can be sitting at home and push the big red "send checks" button with an offshore team in India in case that guy is too drunk to do his job.
Well, as described above at the moment we have (at least) two arms of the government that assess how much you earn: the tax people and the welfare people. My suggestion would be to streamline that into one organisation. At least, if you want to keep an income tax around at all.
In the American context that's the generally agreed upon "alternative". There's no appetite for, say, all of the existing welfare programs and also handing out cash. I don't think there's a meaningful distinction between "handing out cash" and "UBI", but of course I want to acknowledge that cash paid or services rendered (healthcare, etc.) can be a mix in an overall scheme and that UBI typically refers to a specific program. To me it's all the same thing with just a mix of how you want to do it but it's not fair of me to suggest that you think of it the same way so I apologize for that.
> Not sure that will get me past the gate you are keeping here.
If you haven't personally deal with or lived with crackheads, homeless, hustlers, etc. you have little to no insight into how to effectively work with people who are experiencing those issues or lifestyle challenges and you just make the problem worse in general while also wasting money.
It's not gatekeeping to suggest one must have more experience with something in order to make informed decisions or create helpful and fair programs. But if you want to call that gatekeeping I'm happy to gate-keep.
Last Tuesday I called the local police because a homeless guy was standing on the side of the road with a good chunk (the size of an average hand or so) of his leg missing, out in the rain at night, in 40 degree weather. He's not doing that because he just needs some UBI or healthcare, and there's no clinic or pamphlet that will "fix" that person's problems.
The ivory tower will tell you we just need to get that person help, a safe place to live, etc. but they're wrong. I'm sure you see a lot of that in Germany too? What do you guys usually do when you see a homeless person in such circumstances? Do you give the money and does that fix the issues?
> Someone has to pay the taxes to finance the redistribution. The average person can't get more out of the system than they pay in (by definition of how averaging works).
That doesn't make any sense at all. Plenty of people in the United States at least get more than they pay in, even if it's not exactly a cash reimbursement. Not that I have a problem with that.
> In any case, what you want to really avoid are sudden cliffs, where an extra dollar in gross income costs you more than a dollar in net income. And ideally, you also want to avoid unduly high marginal rates (even if they are still below 100%) from phasing out multiple, uncoordinated welfare payments.
Right and we can just avoid those by not having restrictions on the income. It doesn't matter if Bill Gates gets a $2,000 UBI check, he's going to more than make up for it in overall taxes (ideally, unfortunately our government is all too happy to tax anyone but the ultra wealthy). I think the same is true for many other folks too and don't think it's a worthwhile concern.
but not completely - and this is only even talking about maintenance. The initial investment is absolutely not "paid for", because the economic returns from them are privatized, and the tax collection of those private benefits aren't really up to par imho. If it was a private business who did this road/highway investment, they'd be losing money (due to the cost of capital, and the lack of returns from collected tolls/taxes, not to mention the maintenance outlay that comes as a big lumpsum).
> In economics, a public good (also referred to as a social good or collective good)[1] is a commodity, product or service that is both non-excludable and non-rivalrous [...]
That's because roads are rather excludable (see toll roads), and if you've ever been in a traffic jam, you'll notice that road use is rivalrous.
When I had a solar-charged EV, taking transit to SF only made sense if I was going by myself and didn't need to do any transfers. Any additional people or modes and it was always better to drive.
This is misleading. How far is your drive from home to your destination in SF? I bet the total cost of ownership per kilometer driven far exceeds the BART fare.
Like many households, they probably have a car already for other reasons.
To me this is big reason why transit has to be basically free to attract riders. It has to compete with marginal cost per kilometer of private car use, not total cost.
Marginal cost of driving is much higher than people think, because they don't count depreciation, maintenance, and insurance. There's a reason the IRS reimbursement rate is $0.70/mi.
True, though it depends quite a lot on the car. If you're driving a second-hand civic or prius, the marginal cost per mile will be much lower than a Mini or a truck. I don't know what it's like for EVs.
Where I live and with what I drive, once there is a single passenger, the car breaks even on the milage. But parking can shift things back in favor of public transit. It's close enough that traffic is always the deciding factor for which mode I take.
If transit were free, I would probably just take it most of the time.
This. No matter what is said to drivers in HN (and the "Real World"), they always have a good reason to drive instead of take transit. No worries about them. We, transit riders, can be productive (read, work, sleep, etc.) during our commutes.
Very much so. When I was younger I assumed fares were for the cost of the public transport, but after following some local budgeting discussions I was stunned by how little the fares covered operating costs.
Small amounts of cost sharing are a useful technique for incentivizing people to make wise decisions in general, so there’s some value in having token small fares. It’s the same difference that shows up when you list something for $10 in your local classifieds as opposed to listing it as FREE. Most people who use classifieds learn early on that listing things for free is just asking for people to waste your time, but listing for any price at all seems to make people care a little more and put some thought into their decisions. I’ve often given things away for free after listing them for small amounts in classifieds because it filters for people who are less likely to waste your time.
That ignores the massive amount of parking and highway subsidies that make the car-first model in the US viable at all. The absurd amount of space given away for free (or below value) in the city to support cars is actually insane. Its just not properly account for.
In a sane world you would either not have any public parking spots, or parking spots that cost so much that about 10-20% of them are empty, and you would have a road use tax (like Singapore).
And American transit systems are uniquely bad at fare recovery because they are just uniquely bad at everything.
Fares income isn't insubstantial -- just as an example I'm familiar with, King County Metro (Seattle area) was ~33% funded by fares before Covid (which destroyed both ridership and percent non-stealing riders). It is material; not "token."
More or less the same thing as patio11's "pathological customers" premise[0]. People who are unwilling to pay a few dollars to ride the bus tend to be extraordinarily antisocial in other ways. Public transit is for the masses; it's not a homeless shelter.
"While private passenger vehicles contribute 90% of the mileage in the U.S. transportation sector, their emissions share is only 58%. The remaining emissions come from public transit (27%) and other modes including airplanes (13%)."
Diving in, that research is less against public transport in general, more about how the US is just not very good at it:
Our measure of environmental performance is a transit agency's average carbon dioxide emissions per passenger-mile or vehicle-mile.
During the period of analysis, the sector's carbon dioxide emissions declined by 12.8%, while vehicle-miles travelled increased by 7.1% and passenger-miles increased by 10.5%.
Thus, the emissions intensity of public transit has shrunk since 2002 using both measures.
Yet, compared with public transit emissions in the United Kingdom and Germany, we document that the U.S. bus fleets had the highest carbon emissions per mile and the smallest efficiency progress.
ie. US public transport was inefficient and polluting to begin with, and while it improved somewhat when a prior administration finally applied some funding to the task, US public transport stills woefully lags in comparison about the glone.
strong towns is not honest about it though. Urban areas have been maintaining roads for a long time. They seem to think that if you ammortize a road over 20 year you have to replace it in year 21 but most roads are good for 40+ years
As someone who use an eink Android device frequently, I'd appreciate it if every website that uses dark mode by default to have a least a toggle between light/dark.
And do test that the toggle works properly for syntax highlight code blocks: for some websites, the toggle does not change the color of the code block and I had to resort to Chrome's reader mode.
Ideal for what problem? Certainly not for reducing Google's data collection and improving privacy. It would only work with tons of small payment providers, but then you are back at square 1 that users need to subscribe with tons of services for just pennies.
(Side note this is why jj is awesome. A `jj log` is almost as fast as `ls`).
reply