Hacker News new | past | comments | ask | show | jobs | submit login
Advice I Wish I'd Been Told (1999) (archive.org)
170 points by pmoriarty on Sept 13, 2016 | hide | past | web | favorite | 40 comments



I'm not knocking this article, but when original title is so vague and clickbaity, would you mind filling in some more information in the HN submission?


TIL: A way to measure generality in texts http://research.create.usc.edu/cgi/viewcontent.cgi?article=1...

Bad headlines (cryptic, general, and clickbaity) are common, not only in HN. Is there a good solution? (good = better than moderators)

There have been more than a couple of TLDR applications/apps mentioned here before.


There is a relatively simple solution: tags


Tags are a double edged sword. On one hand they offer quick and easy classification. On the other hand, without a clear classifier system they become erratic and of much less use than originally intended, simply because everyone uses different "internal" classifier system. HN has pretty wide variety of topics making usefulness somewhat doubtful. The bigger problem is not browsing top page, but tag based search.

Actually HN already has most of the functionality required. I always found it strange why I can submit either URL OR text. Sometimes it is actually difficult to express the main point in a title sentence. We could provide short summary for the article in the text section.


Reminded me of a solution by one calculus student to an optimization problem:

"(...constraints...) How should the fence be built to minimize the cost of the fence?"

Student answer: Don't build it. Cost=0.

Sounds almost like a Koan.


Answer: Make Mexico pay for the fence.


Anecdote: In my pinboard.in account, I have 6,123 bookmarks, and 497 tags. With no more analysis than a calculator, that's roughly one tag servicing 12 bookmarks. I don't think that's very good.

(I am, however, still very happy with my pinboard account. I treat my bookmarks as searches that I'm going to want to do in the future, with a high probability of finding something useful.)

How I got here is that I use the pinboard Firefox addon, and when I find something useful, I bookmark it in pinboard. Pinboard has tags, and it will often make suggestions, which are often correct. But of course you can add tags on the fly. As I've used the account over the years, my interests have changed, as well as my awareness of how tags have served me so far, so I would tag some things very differently a few years ago as opposed to now or in the future. That makes older tags potentially less relevant, as they are less likely to come to mind, and therefore less likely to help me find the thing I'm looking for now, with today's focus.

Which is a long way of saying, tags are essentially arbitrary without a well thought out system (mental and/or physical), and sufficient thought and constraints at each tagging event. The last tag in my list is "youtub."

Tags are cool. And tags are hard. The idea is simple, but I wouldn't call them a simple solution.


Tags are a solution, though perhaps not a perfect one. But what solution ever is?

Counter-anecdote for you: I've been using the web ever since it was created, and for approximately the first half of that time I was just using regular bookmarks, with a regular bookmark manager, putting bookmarks in a heirarchichal folder structure that I manually created.

Then browsers developed the ability to tag bookmarks. Since then I've switched over to only tagging bookmarks, and no longer put them in to any folders and don't so don't have to manually create a folder heirarchy for them.

What I do is simply think of every possible way to describe a particular bookmark, with trying to think of every possible keyword/phrase way I might want to use search for them in the future. This has served me exceedingly well, and I've been able to find pretty much every bookmark I was interested this way.

This method is still not perfect. The tag-management features of all the bookmark managers I'm aware of are pretty primitive. It would be nice, for example, for these bookmark managers to automatically show show tag clouds or let you browse tags along with their respective bookmarks. This is technically possible, but afaik hasn't been done yet.

Despite this, tags can work, and work really well. It would really suck to have to search through thousands of bookmarks without tags or any kind of heirarchy. That's effectively how HN is: very poor structure. They could do a lot better, and I haven't heard of any suggestions to improve the situation that are anywhere as good as tags.


My experience parallels yours almost exactly, particularly switching from bookmark folders to one big folder with tags. However, there's one more step in my progression, which is that I all but stopped using my browser's bookmarks, and moved to pinboard (with the painful interim step of delicious, on which pinboard is based).

After my comment, I started thinking about tags again, and how I might clean up my pinboard tags. I decided the easiest way to start would be to delete all my tags. Before I did that, I did some simple tests.

In pinboard, you can:

  - search all your bookmarks.
  - full text search all your bookmarks.
  - click on a tag and get everything it tags.
  - click on multiple tags and get everything they tag.
Here's my unanalyzed results, searching for "python":

  text box: 700
  tag: 556
  
  check "search full text"
  
  text box: 707
  
  After eliminating all tags:
  text box: 700
  
  check "search full text"
  text box: 707
I'm betting that the 700 and 707 are the same before and after I deleted all my tags (yes, I actually did that). Don't know what the intersection is between the 556, 700 and 707.

But I look at that and I wonder, do I really need my tags? What are they doing for me. Similar question to what made me, and possibly you, to eliminate folders in your browser's bookmarks.

And I really am wondering now, do I want to bother at all rebuilding a smaller tag set? Of course that's a question for me, and it's anecdotal, everyone has different needs and solutions that work for them, according to some definition of "work."


For privacy reasons, I don't want to share my bookmarks with the owners of any website or corporation.

I've been meaning to import all of my bookmarks in to an emacs org-mode file, and just use its more sophisticated tag management, and possibly build on that to write other features that I may need.


> For privacy reasons, I don't want to share my bookmarks with the owners of any website or corporation.

Fair enough.

If I had stayed on "bookmarks in my browser in a single folder with tags," I might be asking myself the same questions I've been asking myself about tags on and off all day. It's interesting, sometimes surprising, when I question assumptions that I didn't realize I was making.

A good conversation.


> This method is still not perfect. The tag-management features of all the bookmark managers I'm aware of are pretty primitive. It would be nice, for example, for these bookmark managers to automatically show show tag clouds or let you browse tags along with their respective bookmarks. This is technically possible, but afaik hasn't been done yet.

If I understand you correctly, pinboard does all that.


> There is a relatively simple solution: tags

Tags don't and have never worked.

(I'd argue a hashtag is not a tag)


I want to hear that argument.


It's probably irrelevant to HackerNews, but whatever. Thanks, it's exactly what I needed today. I've been telling myself to write clearly, directly, and in utilitarian language for so long that I almost forgot the love of language.


>Irrelevant

Not quite https://news.ycombinator.com/newsguidelines.html

Thought: recommendation engine based on your upvote profile?


No thanks. I like knowing that everyone gets the same articles. It's like a newspaper where we are the readers, writers and editors. I'm exposed to good ideas because the rest of the community thinks they are worth hearing about. An algorithm might prevent me from seeing those ideas.


Agreed. I run into this on Reddit from time to time - I see lots of the things I'm used to and would have read in the past, but I don't see the unexpected. When I occasionally browse before logging in I discover that there's a lot I'm missing.


Vehemently no. I enjoy the unfiltered, random view and don't want somebody else deciding how to tweak the personalisation algorithm.


Right. The personalisation algorithms are one reason for Facebook and Twitter confirmation bias bubbles.


HN is filtered. The main page is a list of article which have been selected and upvoted by small unrepresentative subset of the population. I agree that it's not personalized though.


>HN is filtered.

Not relative to what everyone else is seeing.


Then click on "new" (it's only one click away), and you'll see everything that's submitted. If you change your profile to have "showdead" = "yes", you'll really see everything - even the stuff that's flagged or downvoted to oblivion.


What if the feature could be turned on and off (like with a radio button), so you could see both "views" (random vs. upvote history based), depending on your preference, or curiosity?


I personally prefer clickable thematic tags, which HN is already somewhat implementing with "new" and "show".


Ditto


Of course! I am so tired of seeing posts filled with microaggression, totally violating my safe space. Just make sure that recommendation engine doesn't brake, or horrible things may happen:

http://www.theonion.com/article/horrible-facebook-algorithm-...


I'm glad this sort of lateral-thinking-esque thing can still make it to the front page. I don't write poetry, but this seems serendipitously related to other things I've been thinking about.

I've been getting into lifting weights a lot over the last year, and something I read recently changed the way I've started performing the exercises. I've started focusing more on the movements I'm doing, resisting the lazy desire to just go through the motions. It seems to me to be a kind of "avoiding the abstraction," instead lasering in on what's actually going on with my body. It has the tendency to bring out an ecstatic state as well, which is a nice bonus. It feels like a much deeper relationship with the thing I'm doing.


That deep attention to what you are doing sounds like "flow" -- Mihaly Csikszentmihalyi's concept of optimal experience. See his talk here: http://www.ted.com/talks/mihaly_csikszentmihalyi_on_flow?lan...


[Writing] advice I wish I'd been told.


They wrote about advice for writers, but with a grain of salt this seems apropos to any creative art, design and construction of digital systems included.


Whilst reading, I wondered how the first guidance, "avoid abstraction", could apply to creating programmes, especially in commercial settings.

"Purple", "poetic" or "tangental" function, variable and class names litter my personal projects. I enjoy coming up with these things immensely, at the time. Often, later, I rename them to more directly communicate their intended use. Which way exhibits "better" poetry?


That also applies to life and whenever you try to create anything.


Is the hyphen meant to be there in "ab-stracted" ?

Or is it some artifact of publishing perhaps?

If it is meant to be there what do people think is intended by it? Just to emphasise it? Something else?

(also, no need to use archive.org; the original is now at http://www.valpo.edu/vpr/mcdonaldessay.html)


In the archive.org version, the word "abstracted" is broken across two lines. That's why there's a hyphen.


Thanks but it's not for me; it's on a single line in both.

So .. some publishing/web artifact as I supposed.


I skipped over the first paragraph as was semi-confused to see "Resist Abstractions" as point #1 in a HN article.


* A poem works best, for me, when the writer doesn't tell, but when he or she invents combinations of specific words to show us old facts in new ways. *

I have a similar saying for philosophy; that it is to put the wrong words in the right order. I guess it applies to poetry too, or only.


Advice about writing is oft written in some of the most purple and interminable prose. This article is no different.


Buy Google stock would have been Good advice then.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: