"Essentially it seems like he is despairing of fixing what is already broken, and warning DOGE to pick winnable battles with big payoffs? But in the long term there is no alternative to fixing the core issues, short of revolution and starting over."
Yes, and he's right, we should nudge them to "pick winnable battles with big payoffs" but absolutely recognize that "big" is not "big enough." Government efficiency is nice and we should do it, but for the real debt crisis, you have to be willing to touch the Unholy Trinity of actual government spending.
The problem is that the administration they're tethered to has made it a very public point that they have no intention of doing that, and was also the president responsible for the largest single blow to the budget/debt in the history of the country.
So DOGE will either:
1) tinker around the edges and maybe make some good improvements in efficiency, but not actually change the Debt Crisis trajectory more than a few months, but be allowed to continue to exist
2) actually try to cut SS/medicare/military, and then immediately be thrown under ten buses by the administration
I think this is very much a One Does Not Simply situation - the various downsides, including it backfiring directly and the reputational issues, are very real, as are the time and health costs.
As in, I notice I wouldn't have plastic surgery on anything like current margins. But if you offered $2300 for 0.4 standard deviations in looks, I would keep pressing that button A LOT of times. At least 100, if you let me.
Possibly I underestimate American health care costs too...
It also must have diminishing returns. Otherwise I'd be very worried about 40-sigma Zvi walking into the room and instantly converting everyone to his cause.
EA cause: reduce stigma from plastic surgery, thereby making more people beautiful, thereby increasing our quality adjusted lifeyears.
Plastic surgery clearly has both diminishing returns and backfires, as noted by various celebrities who took it too far and look...less than human now.
Another issue is that plastic surgery does not pass on to your kids, so either yourself or your spouse getting (really good) plastic surgery has less of an effect than the same level of beauty more naturally. If you and your spouse look amazing but were both ugly prior to surgery, then your kids are more likely to be ugly. I don't think we'd be in favor of surgery for kids, so they would wait until at least teenage years and likely later looking significantly worse than their parents and feeling layers of pain from that.
" . . . thereby making people more beautiful, thereby increasing our quality adjusted lifeyears."
This wouldn't work because our judgements of what is beautiful are relative to other things we've seen. If you made some people more beautiful, other people would just look uglier by comparison.
<<Phryne was a legend in her own right. She earned so much money from her liaisons that she offered to sponsor the rebuilding of the walls of Thebes, so long as they included the inscription “Destroyed By Alexander The Great, Rebuilt By Phyrne The Courtesan” (Thebes declined). While in Athens, she committed some unspecified offense, and got charged with impiety, a capital crime. Athenaeus writes that when the trial seemed to be going against her, she disrobed in the middle of the courtroom. The judges were so awed by her beauty that they became concerned she might be a god, and that something terrible would happen if they harmed her. They voted to acquit.>>
I'd put the study about judges' bias based on "looking trustworthy" in a different category than other sorts of looks-based bias. Judges probably favor attractive people, but they also look for indications that a defendant will reoffend, and I'd expect a bias towards the person who shows up in a suit and looks respectful and attentive as opposed to the equally-attractive person who comes to court covered with gang tattoos, wearing a wifebeater, slouching and apparently intoxicated.
To clarify, the study if anything shows the opposite of what that poster was claiming: this measures the people who worked longer than five minutes to finish it, even though they knew they weren't going to get paid for it. The author has clarified this point, as well as noting there was no increase in sloppiness. The WEIRD countries' respondents stopped working as soon as they weren't getting paid anymore, this has nothing to do with honesty.
> marginal government jobs don’t represent market demand for labor and are reasonably viewed as unlikely to be net productive, given they are likely to represent additional bureaucracy rather than adding classroom teachers, police on the street, firefighters or trash collectors
1. On what basis do you make this sweeping claim?
2. Have you ever read a news story about hiring more cops? About federal financing for localities[1]?
3. Do areas of in-migration, especially of families, ever found new public schools?
4. Are you familiar with the literature that fire safety regulation has made firefighters underutilized but that the politics of cutting firefighter jobs is impossible?
5. Does Baumol's Cost Disease apply to "classroom teachers, police on the street, firefighters or trash collectors"?
Regarding your approval for the FTC right to repair rule: this strikes me as strongly analogous to non-compete agreements, in that it's a fix for allowing private contracts in restraint of trade, in which one side has the incentive to extract maximal concessions, resulting in a sub-optimal and frankly kind of stupid anticompetitive equilibrium.
Reasons that the other side can't prevent this arising may be due any of being unable to systematically value the loss of optionality, being forced into a market where there aren't non-restrictive alternatives (and thus also no marginal cost to new firms adopting noncompetes since everyone else has them), or just being a college student committed to a career path dominated by like five plausible employers with a large candidate pool who needs a job and thus just has a systematically poor bargaining position.
While there are conceivably instances in which bespoke non-competes make sense, the fact that Jimmy Johns even *attempted* to institute one as a contract of adhesion, combined with California's extremely vibrant and liquid tech labor market and the nuclear option of trade secret litigation to pursue the genuine absurdities all strongly incline me towards thinking that CA has it right and everyone else has it wrong, with the self-interest of a job-hopping executive class in switching jobs mitigating the damage in much the same way that a minimum wage below the market clearing wage does.
Regarding non competes, it seems to me there are some possible compromises which have not really been fully explored, like for example requiring non competes to include salary payments during the non compete period, or making non competes only enforceable if the employee had legal representation. A similar argument could be made for arbitration rules. I’m less familiar with Right to Repair.
Indeed, these people are still around. I think it would make for a good journalism piece to see how they feel about this activism with the benefit of hindsight. I would guess that they are mostly still against pay toilets.
> So this is weird, because every calculator says that she should expect in the range of $34k/month in streaming income - but I do not think she would be lying about this.
Spotify pays ~$0.004/stream but Lily's share of that is a function of whatever deal she has with her record label. She's also splitting that most likely with other songwriters and producers on any given song/album. So her take-home could be a tiny fraction of the overall Spotify payout.
> You are not to treat your own information any differently from the information of others. In scenario type after scenario type, participants not following this rule are sneered at.
I would add that the expected epistemic quality level of information from unknown others can be (generously) described as "noise." If your own information about something is also expected to be noise, it may make sense to average the noise out somewhat, otherwise it'll just make things worse. Unfortunately, even then, averaging only works for quantitative estimates, which are rare outside the lab.
Of course this implies that, on average, incorporating information from others is actually good advice, so I do in fact agree with the injunction for most people most of the time and all people some of the time. But, if you're a few standard deviations above average intelligence, you might actually be more likely to make the opposite mistake and overweight peoples' unreasoned and unreasonable opinions by assuming they have much better reasons for what they're saying/writing than they actually have.
I notice my own thinking is much clearer after reading someone else's take, even if I strongly disagree with everything they said. It helps me focus on what I actually think and gives me leads to follow.
Reading Zvi serves this very purpose - surely he recognizes his value is significantly more than just in however often he is correct, but also includes getting people to think about the topics at all.
I mostly agree, but hearing someone and thinking about someone's arguments is very different from deference to what they're saying (as seems to be implied at least by the quoted passages). See for example "read the science" vs "Trust the Experts".
Also note "unknown others" above. It would be for sure throwing out the baby with the bathwater to always discount what others are saying, especially when they have earned your (epistemic) trust. I am posting this in the comment section of a blog post, after all ;)
The ideal would be to discount what someone says as much as the Bayesian evidence one has of their epistemic reliability entails. I was making a point about the prior in that process, but Zvi has earned a pretty high number of bits of evidence above that in my book.
The Experts have also earned a certain number of bits in fields where evidence is relevant to their research - generally speaking, it makes sense to defer to scientific consensus absent other data, while taking into account known biases and keeping up your guard. But not the average guy on the internet (except when they have relevant expertise, e.g. plumbers talking about pliumbing).
I'm familiar with the Bay Area that gets a lot of Asian immigration. My experience is that everyone else leaves a place like Saratoga that is too populated by high income Asians because the schools get "too good". All the Asian kids are studying 24/7, the grades look amazing, with a lot of pressure from the parents and peers. But the kids are stressed out and report being miserable. Other families move to more "normal" areas where their kids will not feel outcast just because they're not doing piano, violin, and Chinese classes 2 times a day on top of regular school.
I wish I could find the link now, but I just came across info suggesting that Selzer told the newspaper a year ago that she would not be renewing her contract, and that therefore, this retirment, awkward timed though it is, is not in response to the bad polling miss.
That of course could just be a version of the "in order to spend time with ones family" that one often hears, but from the little I have heard about her, I would not expect her to take a polling error badly like that.
I heard a theory right after her poll came out that she was 67 and may be putting out a knowingly bad poll to shift the election towards Harris in some way, knowing she was going to retire either way.
I don't know how we would ever falsify that, but it does track with the information we are getting.
So literally in the article announcing her retirement, lol. I wonder how much of the narrative is people who didn't read past the headline (or people like me who didn't even do that, and got it all second or third hand), vs. people who assume it's just normal dissembling and that just don't believe her.
Another in the list of "obscure research that hits big", when I was completing my undergrad, I knew a PhD candidate who was literally getting more grant money than she knew what to do with because she just happened to be studying Geoduck (the large, very phallic looking clams) biochemistry (in regards to how they dealt with changes in salinity and environmental toxins, I think), when it was discovered that the cell membrane transport protein that they had was nearly identical to the one that was implicated in cancer cells gaining resistance to various cancer drugs. Geoducks immediately became an important (or at least potentially important) model organism. I haven't kept up with her career since then, or whether her work ended up being as important as it maybe could have been, but the point was that she was studying something most people would have said was a total waste, and it ended up having implications for the study of cancer in humans.
"I interpret this result as strong evidence that when people talk about ‘morality’ they mostly mean something quite superficial, the superficial surface appearance of morality. If the mere superficial surface appearance of morality - which is all one can possibly hope to measure in milliseconds - is then described as ‘accurate’ then that is all that later judgments are measuring."
It could also mean that people are actually really good at approximating morality from minimal information. To test this, we wouldn't just prove that sometimes people are wrong about morality when more information is available, but actually look at the rates. Humans needs to determine the morality of thousands of things a day, often in very subtle ways. It makes sense for us to have a mechanism to do that very very quickly as long as it's right often enough to be useful.
Keeping in mind that any system that allowed us to quickly determine the morality of a situation is extremely gameable and people defecting against us can and will use it to their advantage.
You can build an ad-hoc community, but you have to opt-in to being the central node. This is a huge carrying cost and single-point-of-failure vs a real village, but sidesteps the tolerance issue. If the node map is everyone connects to A, and some connections between everyone else (but A coordinates) than A can account for B and C having a politics disconnect and too much personal history between D and F and E likes B, so can only connect them sometimes b/c it'll be awkward of if B turns E down.
Ok, that's not new or interesting. Some specific advice though: Build the ad-hoc network (or move to where it can arise) and then ACTIVELY encourage connections between other people without you. There's a very strong implied social sense that if you introduce two people and they do stuff without you later you're being slighted. You've got to constantly fight against this if you want to make the ad-hoc network somewhat self-sustaining.
The other thing is to apply Social Skills 101 to basic interactions. You can turn your ad-hoc network (maybe from a volunteer group, hobby group or social club) into something more real via encouraging people to build connections not dependent on your presence and routinely applying super basic social skills to interactions. Ask about the shared hobby, the local sports team, the kids. Yeah constant Question and Answer is crap form of conversation, so don't be an awkward sod about it. "How are things going? I know it's been a rough back to school season with colds and grand kids" opens the door without prying.
This is very tedious, but if you want to build a network where you have someone you can call at midnight because you're stuck in a swamp and need help, it's how you do it.
"Essentially it seems like he is despairing of fixing what is already broken, and warning DOGE to pick winnable battles with big payoffs? But in the long term there is no alternative to fixing the core issues, short of revolution and starting over."
Yes, and he's right, we should nudge them to "pick winnable battles with big payoffs" but absolutely recognize that "big" is not "big enough." Government efficiency is nice and we should do it, but for the real debt crisis, you have to be willing to touch the Unholy Trinity of actual government spending.
The problem is that the administration they're tethered to has made it a very public point that they have no intention of doing that, and was also the president responsible for the largest single blow to the budget/debt in the history of the country.
So DOGE will either:
1) tinker around the edges and maybe make some good improvements in efficiency, but not actually change the Debt Crisis trajectory more than a few months, but be allowed to continue to exist
2) actually try to cut SS/medicare/military, and then immediately be thrown under ten buses by the administration
"Most people would, I presume, happily pay $2300 for 0.4 standard deviations of improved looks even if it did not impact their earnings directly"
I don't think this is true, because most people haven't had plastic surgery. (Transaction costs? Lack of knowledge? Social stigma?)
I think this is very much a One Does Not Simply situation - the various downsides, including it backfiring directly and the reputational issues, are very real, as are the time and health costs.
As in, I notice I wouldn't have plastic surgery on anything like current margins. But if you offered $2300 for 0.4 standard deviations in looks, I would keep pressing that button A LOT of times. At least 100, if you let me.
I am reading *exactly* the right Ted Chiang short story for this content ("Liking What You See: A Documentary")
Possibly I underestimate American health care costs too...
It also must have diminishing returns. Otherwise I'd be very worried about 40-sigma Zvi walking into the room and instantly converting everyone to his cause.
EA cause: reduce stigma from plastic surgery, thereby making more people beautiful, thereby increasing our quality adjusted lifeyears.
Plastic surgery clearly has both diminishing returns and backfires, as noted by various celebrities who took it too far and look...less than human now.
Another issue is that plastic surgery does not pass on to your kids, so either yourself or your spouse getting (really good) plastic surgery has less of an effect than the same level of beauty more naturally. If you and your spouse look amazing but were both ugly prior to surgery, then your kids are more likely to be ugly. I don't think we'd be in favor of surgery for kids, so they would wait until at least teenage years and likely later looking significantly worse than their parents and feeling layers of pain from that.
An exploration of this theme is the Uglies series by Scott Westerfeld.
" . . . thereby making people more beautiful, thereby increasing our quality adjusted lifeyears."
This wouldn't work because our judgements of what is beautiful are relative to other things we've seen. If you made some people more beautiful, other people would just look uglier by comparison.
What would happen to +40-sigma Zvi:
<<Phryne was a legend in her own right. She earned so much money from her liaisons that she offered to sponsor the rebuilding of the walls of Thebes, so long as they included the inscription “Destroyed By Alexander The Great, Rebuilt By Phyrne The Courtesan” (Thebes declined). While in Athens, she committed some unspecified offense, and got charged with impiety, a capital crime. Athenaeus writes that when the trial seemed to be going against her, she disrobed in the middle of the courtroom. The judges were so awed by her beauty that they became concerned she might be a god, and that something terrible would happen if they harmed her. They voted to acquit.>>
Taken from https://slatestarscratchpad.tumblr.com/post/617802684161982464/the-ancient-greek-city-state-of-knidos-was-famous
I doubt I could have a worthwhile level of plastic surgery for $2300
Podcast episode for this post:
https://open.substack.com/pub/dwatvpodcast/p/monthly-roundup-24-november-2024
I'd put the study about judges' bias based on "looking trustworthy" in a different category than other sorts of looks-based bias. Judges probably favor attractive people, but they also look for indications that a defendant will reoffend, and I'd expect a bias towards the person who shows up in a suit and looks respectful and attentive as opposed to the equally-attractive person who comes to court covered with gang tattoos, wearing a wifebeater, slouching and apparently intoxicated.
The "books are bad for you" bit was done by Plato in _Phaedrus_, if I recall correctly.
You've probably seen this but the item about the 5 minute task is a misunderstanding of the study.
To clarify, the study if anything shows the opposite of what that poster was claiming: this measures the people who worked longer than five minutes to finish it, even though they knew they weren't going to get paid for it. The author has clarified this point, as well as noting there was no increase in sloppiness. The WEIRD countries' respondents stopped working as soon as they weren't getting paid anymore, this has nothing to do with honesty.
yep, this deserves a correction
I've modified the section for this copy of the post (but am not feeling it enough at this point to bother with the other copies)
> marginal government jobs don’t represent market demand for labor and are reasonably viewed as unlikely to be net productive, given they are likely to represent additional bureaucracy rather than adding classroom teachers, police on the street, firefighters or trash collectors
1. On what basis do you make this sweeping claim?
2. Have you ever read a news story about hiring more cops? About federal financing for localities[1]?
3. Do areas of in-migration, especially of families, ever found new public schools?
4. Are you familiar with the literature that fire safety regulation has made firefighters underutilized but that the politics of cutting firefighter jobs is impossible?
5. Does Baumol's Cost Disease apply to "classroom teachers, police on the street, firefighters or trash collectors"?
[1] https://www.justice.gov/opa/pr/justice-department-awards-over-600m-hire-law-enforcement-officers-keep-schools-safe-and
Regarding your approval for the FTC right to repair rule: this strikes me as strongly analogous to non-compete agreements, in that it's a fix for allowing private contracts in restraint of trade, in which one side has the incentive to extract maximal concessions, resulting in a sub-optimal and frankly kind of stupid anticompetitive equilibrium.
Reasons that the other side can't prevent this arising may be due any of being unable to systematically value the loss of optionality, being forced into a market where there aren't non-restrictive alternatives (and thus also no marginal cost to new firms adopting noncompetes since everyone else has them), or just being a college student committed to a career path dominated by like five plausible employers with a large candidate pool who needs a job and thus just has a systematically poor bargaining position.
While there are conceivably instances in which bespoke non-competes make sense, the fact that Jimmy Johns even *attempted* to institute one as a contract of adhesion, combined with California's extremely vibrant and liquid tech labor market and the nuclear option of trade secret litigation to pursue the genuine absurdities all strongly incline me towards thinking that CA has it right and everyone else has it wrong, with the self-interest of a job-hopping executive class in switching jobs mitigating the damage in much the same way that a minimum wage below the market clearing wage does.
Regarding non competes, it seems to me there are some possible compromises which have not really been fully explored, like for example requiring non competes to include salary payments during the non compete period, or making non competes only enforceable if the employee had legal representation. A similar argument could be made for arbitration rules. I’m less familiar with Right to Repair.
The kid leading the anti-pay toilet charge is (and surely was then) a mathematical genius: https://en.wikipedia.org/wiki/Ira_Gessel.
So it's not only the young who are stupid. Also geniuses acting outside their domains, culture- and time-bound to boot.
Indeed, these people are still around. I think it would make for a good journalism piece to see how they feel about this activism with the benefit of hindsight. I would guess that they are mostly still against pay toilets.
> So this is weird, because every calculator says that she should expect in the range of $34k/month in streaming income - but I do not think she would be lying about this.
Spotify pays ~$0.004/stream but Lily's share of that is a function of whatever deal she has with her record label. She's also splitting that most likely with other songwriters and producers on any given song/album. So her take-home could be a tiny fraction of the overall Spotify payout.
> You are not to treat your own information any differently from the information of others. In scenario type after scenario type, participants not following this rule are sneered at.
I would add that the expected epistemic quality level of information from unknown others can be (generously) described as "noise." If your own information about something is also expected to be noise, it may make sense to average the noise out somewhat, otherwise it'll just make things worse. Unfortunately, even then, averaging only works for quantitative estimates, which are rare outside the lab.
Of course this implies that, on average, incorporating information from others is actually good advice, so I do in fact agree with the injunction for most people most of the time and all people some of the time. But, if you're a few standard deviations above average intelligence, you might actually be more likely to make the opposite mistake and overweight peoples' unreasoned and unreasonable opinions by assuming they have much better reasons for what they're saying/writing than they actually have.
More evidence for reversing any advice you hear.
I notice my own thinking is much clearer after reading someone else's take, even if I strongly disagree with everything they said. It helps me focus on what I actually think and gives me leads to follow.
Reading Zvi serves this very purpose - surely he recognizes his value is significantly more than just in however often he is correct, but also includes getting people to think about the topics at all.
I mostly agree, but hearing someone and thinking about someone's arguments is very different from deference to what they're saying (as seems to be implied at least by the quoted passages). See for example "read the science" vs "Trust the Experts".
Also note "unknown others" above. It would be for sure throwing out the baby with the bathwater to always discount what others are saying, especially when they have earned your (epistemic) trust. I am posting this in the comment section of a blog post, after all ;)
The ideal would be to discount what someone says as much as the Bayesian evidence one has of their epistemic reliability entails. I was making a point about the prior in that process, but Zvi has earned a pretty high number of bits of evidence above that in my book.
The Experts have also earned a certain number of bits in fields where evidence is relevant to their research - generally speaking, it makes sense to defer to scientific consensus absent other data, while taking into account known biases and keeping up your guard. But not the average guy on the internet (except when they have relevant expertise, e.g. plumbers talking about pliumbing).
And yes, that obviously includes me.
I'm familiar with the Bay Area that gets a lot of Asian immigration. My experience is that everyone else leaves a place like Saratoga that is too populated by high income Asians because the schools get "too good". All the Asian kids are studying 24/7, the grades look amazing, with a lot of pressure from the parents and peers. But the kids are stressed out and report being miserable. Other families move to more "normal" areas where their kids will not feel outcast just because they're not doing piano, violin, and Chinese classes 2 times a day on top of regular school.
That sounds absolutely miserable. I wonder if that culture contributes to the fertility crisis?
I wish I could find the link now, but I just came across info suggesting that Selzer told the newspaper a year ago that she would not be renewing her contract, and that therefore, this retirment, awkward timed though it is, is not in response to the bad polling miss.
That of course could just be a version of the "in order to spend time with ones family" that one often hears, but from the little I have heard about her, I would not expect her to take a polling error badly like that.
I heard a theory right after her poll came out that she was 67 and may be putting out a knowingly bad poll to shift the election towards Harris in some way, knowing she was going to retire either way.
I don't know how we would ever falsify that, but it does track with the information we are getting.
https://www.desmoinesregister.com/story/opinion/columnists/2024/11/17/ann-selzer-conducts-iowa-poll-ending-election-polling-moving-to-other-opportunities/76334909007/
So literally in the article announcing her retirement, lol. I wonder how much of the narrative is people who didn't read past the headline (or people like me who didn't even do that, and got it all second or third hand), vs. people who assume it's just normal dissembling and that just don't believe her.
Another in the list of "obscure research that hits big", when I was completing my undergrad, I knew a PhD candidate who was literally getting more grant money than she knew what to do with because she just happened to be studying Geoduck (the large, very phallic looking clams) biochemistry (in regards to how they dealt with changes in salinity and environmental toxins, I think), when it was discovered that the cell membrane transport protein that they had was nearly identical to the one that was implicated in cancer cells gaining resistance to various cancer drugs. Geoducks immediately became an important (or at least potentially important) model organism. I haven't kept up with her career since then, or whether her work ended up being as important as it maybe could have been, but the point was that she was studying something most people would have said was a total waste, and it ended up having implications for the study of cancer in humans.
"I interpret this result as strong evidence that when people talk about ‘morality’ they mostly mean something quite superficial, the superficial surface appearance of morality. If the mere superficial surface appearance of morality - which is all one can possibly hope to measure in milliseconds - is then described as ‘accurate’ then that is all that later judgments are measuring."
It could also mean that people are actually really good at approximating morality from minimal information. To test this, we wouldn't just prove that sometimes people are wrong about morality when more information is available, but actually look at the rates. Humans needs to determine the morality of thousands of things a day, often in very subtle ways. It makes sense for us to have a mechanism to do that very very quickly as long as it's right often enough to be useful.
Keeping in mind that any system that allowed us to quickly determine the morality of a situation is extremely gameable and people defecting against us can and will use it to their advantage.
"It Takes a Village You Don’t Have"
You can build an ad-hoc community, but you have to opt-in to being the central node. This is a huge carrying cost and single-point-of-failure vs a real village, but sidesteps the tolerance issue. If the node map is everyone connects to A, and some connections between everyone else (but A coordinates) than A can account for B and C having a politics disconnect and too much personal history between D and F and E likes B, so can only connect them sometimes b/c it'll be awkward of if B turns E down.
Ok, that's not new or interesting. Some specific advice though: Build the ad-hoc network (or move to where it can arise) and then ACTIVELY encourage connections between other people without you. There's a very strong implied social sense that if you introduce two people and they do stuff without you later you're being slighted. You've got to constantly fight against this if you want to make the ad-hoc network somewhat self-sustaining.
The other thing is to apply Social Skills 101 to basic interactions. You can turn your ad-hoc network (maybe from a volunteer group, hobby group or social club) into something more real via encouraging people to build connections not dependent on your presence and routinely applying super basic social skills to interactions. Ask about the shared hobby, the local sports team, the kids. Yeah constant Question and Answer is crap form of conversation, so don't be an awkward sod about it. "How are things going? I know it's been a rough back to school season with colds and grand kids" opens the door without prying.
This is very tedious, but if you want to build a network where you have someone you can call at midnight because you're stuck in a swamp and need help, it's how you do it.
Excellent comment, Thank you