Said Achmiz
Karma: 8,738
NewTop
Thank you for the encouragement, and I’m glad you’ve found value in my commentary.
… it’s also important to track which errors seem like part of a pattern of motivated error, and which seem to be mere mistakes. The former class seems much more dangerous to me, since such errors are correlated.
I agree with this as an object-level policy / approach, but I think not quite for the same reason as yours.
It seems to me that the line between “motivated error” and “mere mistake” is thin, and hard to locate, and possibly not actually existent. We humans are very good at self-deception, after all. Operating on the assumption that something can be identified as clearly being a “mere mistake” (or, conversely, as clearly being a “motivated error”) is dangerous.
That said, I think that there is clearly a spectrum, and I do endorse tracking at least roughly in which region of the spectrum any given case lies, because doing so creates some good incentives (i.e., it avoids disincentivizing post-hoc honesty). On the other hand, it also creates some bad incentives, e.g. the incentive for the sort of self-deception described above. Truthfully, I don’t know what the optimal approach is, here. Constant vigilance against any failures in this whole class is, however, warranted in any case.
For example you might feel anxious about something and be unable to sleep because you keep thinking about it. You could think of this as being taken over by a part, or by yourself in a configuration of anxiety and worry.
Or you could think of this as your emotions arising from internal processes which are not under your conscious control, nor under the conscious (or even “conscious”, in some metaphorical sense) control of any “part” or “configuration” of yourself. This view has the virtue of actually being true.
It seems to me that all this “parts”, “configurations”, “sub-personalities”, and similar stuff stems from either an inability to understand, or an unwillingness to accept, the fact that humans, fundamentally, are not agents (in the sense of all of our actions being caused by volitions in the service of goals). We often act like agents; we can usually be usefully thought of as agents; but if you start with the assumption that we actually are agents, you’ll run into trouble. And so (it seems to me) you end up thinking: “Well, if I were an agent, I would act in way X. But I find myself acting in way Y! How can this be? Ah…! Of course! There must be other agents, inside me!”
But no. You’re just not an agent. That is all.
Edit: Another way to describe this particular bias might be “insistence on applying the intentional stance to yourself, even when it’s not appropriate”.
Insofar as your answer makes predictions about how actual “rationalists” behave, it would seem to be at least partly falsified: empirically, it turns out that many rationalists do not respond well to that particular suggestion (“modify chickens to not feel pain”).
(The important thing to note, about the above link, isn’t so much that there are disagreements with the proposal, but that the reasons given for those disagreements are fairly terrible—they are mostly non sequiturs, with a dash of bad logic thrown in. This would seem to more closely resemble the way you describe a “stereotypical non-rationalist” behaving than a “stereotypical rationalist”.)
Indeed; hence the term “transhumanism” (and, relatedly, “posthumanism”).
Change is terrifying. This is to be expected, not least because most change is bad; in fact, change is inherently bad. Any change must be an improvement, must justify itself, to be worthwhile. And any such justification can only be uncertain. When we look out along a line of successive changes, what we see on the horizon terrifies us—all the more so because we can only see it dimly, as a vague shape, whose outlines are provided more by our imagination than our vision.
But the alternative is an endless continuation of the cycle, in its current form, with no improvement or hope for escape, until the heat death of the universe. That is far more horrifying. If that were all humanity had to look forward to, then we would have no need of Hell.
I think that if a user is thinking about these sorts of things at all, then they can, should, and do have the capacity to make an actual heading element. (And if this is at all difficult or unintuitive in the UI, then that is the flaw that needs to be rectified!)
But that’s not a heading, nor would it be correct to treat it as such! (Note that lower down on the page you linked, the tutorial specifies how to make actual headings!)
I literally googled “Markdown bold” and among the first few results this tutorial uses bolded headers as an example
Huh? I just went to that link; I don’t see where it says to use (or uses) bold-as-header.
Edit:
Just because markdown has a heading syntax, doesn’t mean that everyone follows it, and depending on context you might not want to follow it.
That not everyone follows it is, clearly, technically true (though I don’t at all share your impression of this practice’s prevalence). But I’m curious to know why you would ever not want to follow the heading syntax, if what you want to produce is a heading? (Other than cases of “this particular parser and/or renderer exhibits bizarre behavior, so I unfortunately have to produce non-standard Markdown in order for what I write to look sane when displayed”—but that, obviously, does not apply here, since we’re talking about determining what the parser and renderer do!)
Most of the content on LW is historical content, where editor features like that weren’t available. I want to make sure the ToC works well for that content, but I feel hesitant to edit all the HTML of those posts, in case that does change the semantic meaning of the text in some cases (obviously we introduced some of that confusion already with the ToC, but I think editing the old content feels a bit more violating than that).
Also, I just want to point out that one obvious compromise would be to both treat all-bold elements as headings (for compatibility with as-yet-un-updated old content), and convert all-bold elements in newly-created content (that is written with the Draft.js editor) to proper headings. That way, you would not be creating any new standards-violating HTML, while not being under any pressure to edit old content, and having the ToC work as expected for said old content.
Some aspects of the conversion can, and should, be scripted. Some should be manual.
There is no reason to go through each post. Prioritize them: if someone comments on an old post, clearly it’s active, so edit that. If a post is getting a lot of traffic, edit that. Otherwise, leave it alone until there’s a reason to edit it, or you have time to do so.
(Also: crowdsource this stuff. Let trusted volunteers submit edited HTML, then drop it in as a replacement if it’s good.)
I strongly endorse editing the HTML of old LW posts.
I’ve looked at that HTML, and almost invariably been horrified; it’s often an absolute mess. I don’t think that changing the semantics of the post is a serious danger, for three reasons:
It’s usually quite clear what it should be.
The way it’s displayed currently is already not the same as the way it was displayed back then, and there’s no remedy for that; better to make it right and proper going forward, than to compound old mistakes and the effects of an awkward transition.
You can (and should) archive the un-corrected version, in case there’s a need to revert anything.
I think users that are used to Markdown will often use single bold words as heading
Really? I haven’t seen this… but if it’s true, it shouldn’t be encouraged!
Some content is submitted to us via RSS feeds from people’s blogs, where we obviously have no control over their HTML, and I would also prefer to not modify it.
This is understandable, but in general I think it’s better to do it right in the majority of cases and to either make an exception or to allow imperfect outcomes in a minority of cases, than to do it wrong in the majority of cases for the sake of consistency. (How much of this RSS’d content is there, anyway?)
Not 100% sure which thing you’re referring to – the fact that the header-section (at the very top) re-appears when you scroll up has been there for about a year.
Hmm. I never noticed it before. No, something about the behavior of the header has changed. I don’t use LW.com enough to tell you exactly what, but I’m almost sure it’s something…
In any case, change or no, my comments about the current behavior do stand.
An element that is entirely bold counts as a heading
So, I have a problem with this. My problem is that this is deeply antithetical to best practices of semantic HTML and web standards in general.
Now, my understanding is that the motivation for this behavior is that many people find it easier and more intuitive, in the Less Wrong editor UI, to bold some text than to figure out how to make a proper heading. This is understandable. However, if this is indeed the reason, then let me propose a solution:
When someone makes an all-bold element in the LW editor, convert that element into an actual heading when saving the post or comment. (Use whatever heading level is the lowest available level in the given editing context—post or comment.)
That way, the HTML content can better conform to standards and best practices, and the ToC features on both LW and GW can stick to using only actual HTML headings as sections.
Re: the post page redesign: I rather like the new layout of the title / post-meta section. However, the new sticky heady is not great. Sticky headers in general are really quite annoying, especially on desktops or, worse, on laptops (where vertical space is at a premium).
(There is also what appears to be a layout bug where the appearance of the sticky header shifts the ToC down, even though it really seems like it shouldn’t.)
The Table of Contents feature is very well done—better than GW’s, I think. (I expect we’ll take inspiration from this for future modifications to our version of the ToC feature, in fact.) Kudos, guys!
(A quibble, and quite a minor one: the animated scroll-to-section is rather too slow for my taste. I’d prefer it scroll to the clicked section at least twice as quickly.)
Strongly downvoted for totally unnecessary politicization. Grandparent refers to concepts from religion, philosophy, and roleplaying games. If you perceive some obscure political-faction dog-whistling in perfectly ordinary comments, that is hardly a reason for other commenters to contort themselves to avoid setting off your politics detector.
(I mean, consider generalizing this principle. Is anything that happens to be a shibboleth, no matter how contrived, of some faction, no matter how irrelevant, now off-limits? This is tantamount to surrendering control of what language, and even what concepts, we use, to every Tom, Dick, and Harry whose only accomplishment is posting clickbait political screeds on the Internet.)
Consider: “having food is good. Having more and tastier food is better. This is common sense. Transfoodism is the philosophy that we should take this common sense seriously, and have as much food as possible, as tasty as we can make it, even if doing so involves strange new technology.” But we tried that, and what happened was obesity, addiction, terrible things happening to our gut flora, etc. It is just blatantly false in general that having more of a good thing is better.
Conclusion does not follow from example.
You are making exactly the mistake which I described in detail (and again in the comments to this post). You’re conflating desirability with prudence.
It is desirable to have as much food as possible, as tasty as we can make it. It may, however, not be prudent, because the costs make it a net loss. But if we could solve the problems you list—if we could cure and prevent obesity and addiction, if we could reverse and prevent damage to our gut flora—then of course having lots of tasty food would be great! (Or would it? Would other problems crop up? Perhaps they might! And what we would want to do then, is to solve those problems—because having lots of tasty food is still desirable.)
So, in fact, your example shows nothing like what you say it shows. Your example is precisely a case where more of a good thing is better… though the costs, given current technology and scientific understanding, are too high to make it prudent to have as much of that good thing as we’d like.
As for “common sense”: in many human societies it was “common sense” to own slaves, to beat your children, again etc. Today it’s “common sense” to circumcise male babies, to eat meat, to send people who commit petty crimes to jail, etc., to pick some examples of things that might be considered morally repugnant by future human societies. Common sense is mostly moral fashion, or if you prefer it’s mostly the memes that were most virulent when you were growing up, and it’s clearly unreliable as a guide to moral behavior in general.
Now this does prove too much. Ok, so “common sense” can’t be trusted. Now what? Do we just discard everything it tells us? Reject all our moral intuitions?
Yes, by all means let’s examine our intuitions, let us interrogate the output of our common sense. This is good!
But sometimes, when we examine our intuitions and interrogate our common sense, we come up with the same answer that we got at first. We examine our intuitions, and find that actually, yeah, they’re exactly correct. We interrogate our common sense, and find that it passes muster.
And that’s fine. Answers don’t have to be complex, surprising, or unintuitive. Sometimes, the obvious answer is the right one.
That is Eliezer’s point.
I do not read Eliezer as claiming that there are no tradeoffs. Rather, his aim is to establish the desirability of indefinite life extension in the first place! Once we’re all agreed on that, then we can talk tradeoffs.
And, hey, maybe we look at the tradeoffs and decide that nah, we’re not going to do this. Yet. For now. With sadness and regret, we shelve the idea, being always ready to come back to it, as soon as our technology advances, as soon as we have a surplus of resources, as soon as anything else changes…
Whereas if we just shake our heads and dismissively say “don’t you know that tradeoffs exist”, and end the discussion there, then we’re never going to live for a million years.
But on the other hand, maybe we look at the tradeoffs and decide that, actually, life extension is worth doing, right now! How will we know, unless we actually try and figure it out? And why would we do that, unless we first agree that it’s desirable? That is what Eliezer is trying to convince readers of, in this essay.
For example, you say:
How much environmental damage and sweatshop labor goes into “extreme” medical interventions? How much systemic economic oppression is required to have enough low-paid nurses to work the night shift? How many 6-year-olds could have better nutrition and a significantly better life for the cost of extending a 95-year-old’s life to 98 years?
Well? How many? These are fine questions. What are the answers?
Quoting myself once again:
The view here on Lesswrong, on the other hand, treats Alice’s proposal as an engineering challenge. … Once you properly distinguish the concepts of desirability and prudence, you can treat problems with your proposal as obstacles to overcome, not reasons not to do it.
(One important effect of actually trying to answer specific questions about tradeoffs, like the ones you list, is that once you know exactly what the tradeoffs are, you can also figure out what needs to change in order to shift the tradeoffs in the right direction and by the right amount, to alter the decision. And then you can start doing what needs to be done, to change those things!)
Any moral framework that doesn’t acknowledge tradeoffs is broken. The interesting questions aren’t “should we save this person if it doesn’t cost anything?”—truly, that’s trivial as you say. The interesting ones, which authors and ethicists try to address even if they’re not all that clear are, “who should suffer and by how much in order to let this person live better/longer?“.
The problem comes in when people start inventing imaginary tradeoffs, purely out of a sense that there must be tradeoffs—and, critically, then use the existence of those (alleged) tradeoffs as a reason to simply reject the proposal.
And then you have the flaw in reasoning that I described in this old comment:
I think that this post conflates two issues, and is an example of a flaw of reasoning that goes like this:
Alice: It would be good if we could change [thing X].
Bob: Ah, but if we changed X, then problems A, B, and C would ensue! Therefore, it would not be good if we could change X.Bob is confusing the desirability of the change with the prudence of the change. Alice isn’t necessarily saying that we should make the change she’s proposing. She’s saying it would be good if we could do so. But Bob immediately jumps to examining what problems would ensue if we changed X, decides that changing X would be imprudent, and concludes from this that it would also be undesirable.
[…]
I think that Bob’s mistake is rooted in the fact that he is treating Alice’s proposal as, essentially, a wish made to a genie. “Oh great genie,” says Alice, “please make it so that death is no more!” Bob, horrified, stops Alice before she can finish speaking, and shouts “No! Think of all the ways the words of your wish can be twisted! Think of the unintended consequences! You haven’t considered the implications! No, Alice, you must not make such grand wishes of a genie, for they will inevitably go awry.”
Would you mind making the headings into actual headings (instead of just bold text)? That would make the post easier to skim; also, that way, when the post is viewed on GreaterWrong, there will be an auto-generated table of contents!
I share your aesthetic preference (and I consider such preferences to be no less valid, and no less important, than any “moral” ones). But no one here is advocating anything like that. Certainly Eliezer isn’t, and nor am I.