A Conversation for Algorithms
- 1
- 2
A few comments
Pyriform Started conversation Dec 16, 2002
Trivial correction first - "algorithum" in fifth paragraph (twice)
More substantive points:
I don't think a "stereo manual" is a good example of an algorithm, simply because they are not designed to be used in this way. They may *contain* step-by-step guides (algorithms) to show you how to do certain things, but most of the content (in my experience) is purely descriptive: "what it does", rather than "how to do it".
Levels of detail in an algorithm: This is fundamental. Each of the steps in your hand washing algorithm assumes that the hardware running the algorithm (a person in this case) knows how to perform that step. But "go to sink" is in itself a complex action: you have to know what a sink is, be able to locate it, and then travel to it (itself a complex action!). "Turn on water" is similarly fraught, and so on. On the other hand, I could propose an alternative algorithm with but a single step: "Wash hands". Indeed, this might be step 1 of a "bake a cake" algorithm, for example. So your algorithm is not wrong, but it needs to be qualified.
"A later, better version of this algorithm may include drying your hands, and going to watch TV afterwards"
Now that *is* wrong. Drying you hands is just about acceptable, but going to watch TV afterwards is most definitely not part of *any* algorithm for washing your hands!
Algorithmic Complexity:
I think any article about algorithms should at least hint at the fact that some problems are much more difficult to solve algorithmically than others (inefficiency / intractability) and that some are not amenable to algorithmic solution at all (noncomputability / undecidability). These are not just arcane matters, but place real restrictions on what we can do with computers, no matter how powerful they become.
A few comments
NAITA (Join ViTAL - A1014625) Posted Dec 16, 2002
Trivial correction first.
Typos, bad links and other small errors should be reported through <./>Feedback-Editorial</.>, I already did this with the two algorithums (which were pointed out repeatedly in peer review, the editors have no pride in their work these days. )
Also, once an entry has reached the front page it's too late for changing contents, except through the cumbersome update process. Comments on the contents are therefore best done as comments and not change requests.
Not that your posting suggests the last directly, it's just the impression I get when reading it.
I see from your conversation log that you haven't engaged in PeerReview yet, I think that is something you'd enjoy.
Have fun!
A few comments
Pyriform Posted Dec 16, 2002
"once an entry has reached the front page it's too late for changing contents, except through the cumbersome update process. Comments on the contents are therefore best done as comments and not change requests"
Ah. That explains why nobody was remotely interested when I pointed out a six orders of magnitude error in the Neutron star entry (A877980) Still at least they changed "loose" to "lose"...
I confess to not having investigated the various procedures and processes which permeate this place, other than to note that they don't seem to be very effective. On the other hand, I know from past experience that getting people (especially the *right* people) to review documents is a thankless task, so I shouldn't really be surprised.
Does anyone bother with the update process, or are incorrect entries effectively cast in stone once they've passed peer review? Oh, and what is a "change request" here?
A few comments
NAITA (Join ViTAL - A1014625) Posted Dec 16, 2002
Editorial Feedback works like a charm. The typos I/you pointed out were fixed two hours after my post there.
Peer Review is fairly good here, but some articles seem to get picked and edited without considering their treatment in PR. In some cases this is just fine, some people are just nit-picks, in other cases entries go through that could have been improved a lot by accepting some of the peer review suggestions.
I think maybe there's too much focus on throughput. 5 entries on the front page every weekday means, on average 5 entries picked every weekday, finished by sub-ed every weekday, etc. And at times a lot of them aren't all they could be.
Keeping the Guide in order is a lot of work, and the BBC only spends so much on it, so a long and winding road for Edited Entries to be updated is a given. After all they have been through at least a week of Peer Review, and have passed a sub-ed and the editors before appearing on the front page, so there shouldn't be all that much wrong with them...
Oh dear, what am I doing on this soap-box?
A few comments
Researcher PSG Posted Dec 16, 2002
Hello
Why is life never easy....
OK, this article is imperfect. This is in part down to a quick and unexpected pick, and a few confusions during the sub ed process that where completely unavoidable. And in part trying to write it so all could understand, as I thought it pretty pointless writing for those who already knew. I may have glossed over some of the more detailed points, but I did so mainly for the ease of comprehension.
However there is one aspect I disagree with : Stereo instructions are definately algorithms. They have been mentioned in several other, more respectable, texts as an example.
However this is a small point to quibble over. As they are mearly a means to bring home to readers not experienced in critical thinking the purpose of Algorithms.
A few comments
Researcher PSG Posted Dec 16, 2002
P.s. But if your willing to write a version to replace it I will be more than happy to point out where the flaws are in chucking too much complex detail at an audience with widely varying levels of understanding [and a taste for the light hearted approach]
A few comments
Hoovooloo Posted Dec 16, 2002
Pyriform:
Subscribe to Peer Review, for as long as you can stomach it. It's the best place to get howling errors corrected - although there's no guarantee that the Editors won't add their own after a perfectly sound entry has been picked. But Peer Review NEEDS people who care about accuracy, and it needs even more desperately people who care about accuracy who aren't as disillusioned with it as I am. Go to it...
H.
A few comments
Pyriform Posted Dec 16, 2002
Hoovooloo:
I'm afraid you are to blame for my joining H2G2 (only a week or so ago). I was moved to do so by a series of posts you made in a creation versus evolution thread. As a veteran of a number of such online debates myself over the years, I thought you did a particularly good job, and you made me laugh. I'd hoped that your sort of contributions would be more typical than they in fact are...
Researcher PSG:
I accept that your audience will have widely varying levels of understanding, and of course you cannot go into great depth in an introductory article. But what I think you *can* do is to hint at why algorithms are so important (a naive reader might wonder what all the fuss is about), and at some of their limitations. You don't need to go into the mathematics of NP-completeness to talk about the travelling salesman problem, for example. It's a very easy to understand, real world problem, and people might be surprised that it's so hard to solve. Surely that is one reason people look up articles; to be surprised and to push their understanding a little. And if you can be light-hearted at the same time, so much the better. I have no problem with that.
I've now ploughed through my collection of manuals for various items of hi-fi and video equipment and I'll concede that many of them are indeed set out in an algorithmic fashion. Oddly enough, they aren't the ones I've needed to read! My AV amplifier manuals, on the other hand, are not set out in this way. So it's still not an example I'd choose, but my counter argument isn't as strong as I'd thought.
A few comments
Hoovooloo Posted Dec 17, 2002
Pyriform:
Gosh.
"I'd hoped that your sort of contributions would be more typical than they in fact are..."
Well, if you like them, you can make them slightly more typical, by posting more of the same!
H.
A few comments
Sam Posted Dec 17, 2002
Hi all,
We always check the threads in entries that appear on the front page 24hours after they first go up. (This is what I'm doing now!) Apologies for the errors slipping through. As NAITA points out, the typos etc were corrected in Editorial Feedback (which, after an entry has gone up, is always the best place to report errors, queries etc). However, if you want me to change anything else then of course, I will. One thing though, because I don't know anything at all about Algorithms, I'll have to trust you on the changes you want!
Sam.
A few comments
Sam Posted Dec 17, 2002
Hello Pyriform,
I'm sorry that I didn't address the more technical aspects of the queries you made on A877980. (Fixing 'loose' and 'lose' was the easy bit). However, I've subscribed to that page now and I'll have a go at looking a the changes. Once again though, I can't tell which of the points made by you and the author is the more correct because I don't know much (for 'much' read 'not a lot') about Neutron Stars. However, I'm sure we can sort this out.
Sam.
A few comments
Pyriform Posted Dec 17, 2002
Well, the only thing I think is actually *wrong* is the idea of "going to watch TV" as part of an improved algorithm for washing hands.
I've hinted at other ways in which I think the article could be improved, but if the author is happy with it, then I guess that's that.
A few comments
Pyriform Posted Dec 17, 2002
Sam:
I don't want to appear to be over-critical of other people's finished articles. I now realise that peer review is the place to make these sort of comments (though obviously I wasn't around when these particular articles were discussed).
With the neutron star entry, I think the point about density ought to be corrected (I can produce references if you need them). The other points were just things I thought were interesting and relevant and could have been included in an expanded article. I guess it's for the original author to decide if he agrees.
A few comments
Sam Posted Dec 17, 2002
Pyriform,
I really don't think you're being over-critical at all! You see, it's impossible for us here to constantly monitor all the entries that go up. We really do try our best to guarantee factual accuracy but one of the features of an h2g2 entry is it's threads. Unlike the printed medium, an entry has these sort of 'live' tendrils hanging of it and the reader, for the full benefit of an entry, is encouraged to read the 'live' threads which had a 3D element to the entry and which often raise very valid points not raised in the entry itself. The points you made, Pyriform, are the very things that give credibility to these 'live' threads.
Right, let's go over to the Neutron Stars entry...
A few comments
Researcher PSG Posted Dec 19, 2002
I've looked it over again and have no major objections to leaving it as is (But I would say that wouldn't I)
Researcher PSG
A few comments
Pyriform Posted Dec 19, 2002
How on Earth do you justify "going to watch TV afterwards" as part of an improved algorithm for washing your hands?
What would be wrong with my proposing an *even better* algorithm in which the freshly-scrubbed washee prepares a light supper, feeds the cat, settles back to watch the TV and then reads a few chapters of an improving novel before retiring to bed?
A few comments
Researcher PSG Posted Dec 19, 2002
A) it was a joke (and I didn't write it)
B) it is mearly saying you could extend the algorithm to include follow on tasks, which you could if you felt the need
c) Of all the things in peer review people picked up on you are the only one to have a major problem with this simple throwaway comment.
Researcher PSG
A few comments
Pyriform Posted Dec 19, 2002
A) It wasn't funny. This is more or less an essential requirement for a joke. Even if GuideML was extended to include humour-markup (I'm envisaging a large animated laughing policeman pointing out the amusing text for the humour-impaired), I still wouldn't get it. As well as not being funny, it is misleading. Well, it misled me, and I'm not easily misled.
B) Oh, so it wasn't a joke after all? Back to the argument then. If you extended the algorithm to include "follow on tasks" (and I am genuinely astonished that anyone would routinely wash their hands before watching TV - some kind of religious devotional thing, perhaps), then it would no longer be an algorithm for "washing your hands". It would be an algorithm for "spending a lazy evening" or "living your life", or something.
C) I can't help it if people missed out on this. I wasn't there. I'd quite like to see the thread so that I can see what they did pick up on. Is that still possible? Oh, and I've made other suggestions which you are clearly not remotely interested in discussing - so I am now reduced to trying to get fixed the one thing which is actually wrong, rather than improving the article in general. Sorry for bothering you.
A few comments
Researcher PSG Posted Dec 20, 2002
Hello Again
Fair point, looked at it again and it probably should be changed. As I said I didn't write it, just incorporated Dancers work.
Alot of this is down to me only having a restricted amount of time on line this week, pressures of the outside world. Anyway I've had time to look at the other points and here are the answers:
>Noncomputability/undecidability
This is an entry on algorithms in the general sense, this seems like a very computer based concept, and more likely to confuse than inform. However if you want to write an entry on computing algorithms fair enough, but this isn't about that.
>Level of Detail
This is in there if you'd bothered to look closely. Or rather a simplified version based on what is needed to understand this, along with asking the writer to define tasks they may feel to be obvious. Like I say, this is about algorithms in the general sense, and if I went about it in the way you suggest it will make it text heavy and probably no clearer to a majority of the audience.
And finally the stereo manual, but I've already dealt with that.
Anything else?
Researcher PSG
Key: Complain about this post
- 1
- 2
A few comments
- 1: Pyriform (Dec 16, 2002)
- 2: NAITA (Join ViTAL - A1014625) (Dec 16, 2002)
- 3: Pyriform (Dec 16, 2002)
- 4: NAITA (Join ViTAL - A1014625) (Dec 16, 2002)
- 5: Researcher PSG (Dec 16, 2002)
- 6: Researcher PSG (Dec 16, 2002)
- 7: Hoovooloo (Dec 16, 2002)
- 8: Pyriform (Dec 16, 2002)
- 9: Hoovooloo (Dec 17, 2002)
- 10: Sam (Dec 17, 2002)
- 11: Sam (Dec 17, 2002)
- 12: Pyriform (Dec 17, 2002)
- 13: Sam (Dec 17, 2002)
- 14: Pyriform (Dec 17, 2002)
- 15: Sam (Dec 17, 2002)
- 16: Researcher PSG (Dec 19, 2002)
- 17: Pyriform (Dec 19, 2002)
- 18: Researcher PSG (Dec 19, 2002)
- 19: Pyriform (Dec 19, 2002)
- 20: Researcher PSG (Dec 20, 2002)
More Conversations for Algorithms
Write an Entry
"The Hitchhiker's Guide to the Galaxy is a wholly remarkable book. It has been compiled and recompiled many times and under many different editorships. It contains contributions from countless numbers of travellers and researchers."