Already looking forward to Tday
Last week HN had a post about a blog selling for $36k. Ordinarily I’d rate it mildly interesting - blogger does good SEO, regurgitates a lot of content, sells ads and then sells the website for a little over 2x the yearly income.
What I found interesting were some of the comments. One in particular pointed out that the site is a manual content aggregator that could automate most of the aggregation. The user suggested auto-scraping the top 100 related sources for new content on a regular basis, using the Mechanical Turk to rate and tag each article and then posting a generated excerpt for each article, tweeting about the post and posting each picture to pinterest.
I have a feeling that if a person were to listen to HN and literally do what the HN folks suggest, that they have a pretty good chance at something working out. I plan to test this theory out by taking plausible-sounding, low-hanging fruit suggestions, like this one, and trying them out. What could go wrong?
For this suggestion, I’ll write a quick feed scraper and article rating system, put it in a cron job and start automating it. Using the MVP style approach, I think I’ll just automate the scraping and make a rating webapp and do the rating/posting myself at first. This will let me get a grip on the most time consuming parts of getting content out on the web so I have a better idea of what to automate next. It also avoids paying the Mechanical Turk until I actually know what I want to buy.
What a scraper may look like
The world must know - Hungarian synth music of the 70s isn’t bad. I think this is under appreciated
Beautiful sugar powder dump, first snow to stick down low. R and I celebrated by an inagural x-country ski run from our front door. Having Dredge Lake so close is fantastic
The github Windows app is pretty slick, except that it doesn’t seem to want to let changes to the global config stick. I have the problem where I can’t clone a repository from Windows because I have a self signed ssl cert in the certificate chain between me and github (GitHub.IO.ProcessException:Cloning into ‘jogging-ledger’…fatal: unable to access ‘SSL certificate problem: self signed certificate in certificate chain). I can’t do anything about the self signed cert, but supposedly there is a setting that fixes the problem: “git config —global http.sslVerify false”
Using this setting lets me clone, push and pull from the command line. It even works from the github Windows app until I close and reopen it.
I can see the problem with the ‘git config —global —edit’ command. The [http] entry in the global configuration file is emptied out when the github Windows app restarts. I end up with a useless chain of “[http] [http] [http]” I can’t seem to find a setting in the gui that prevents this emptying so I used “git config —local http.sslVerify false” to set each individual repository. This makes cloning a little harder, since I have to remember to use the global setting before attempting to clone, but gets the job done.
//dang, I’ve got to figure out how to email tumblr so that the linebreaks don’t go all wonky
This is old news. Google has forms that can funnel whatever is entered into a spreadsheet. People use them for signup sheets or phone number collectors, or all sorts of things. I don’t run any mailing lists or things that typically require forms, so I have pretty much gone along ignoring forms, feeling pretty good and not knowing what I’m missing out on. Two things have persistently been bothering me, all the time I was blissfully unaware of the benefit of forms. I grow plants and want to keep track of the progress of certain aspects, like how long cuttings take to root, or how fast something grows or how often I actually water them. I also like to write haiku, and feel like writing them daily is helpful for improving.
I have a hard time actually following through with either task. Number one, I tend to not record the information in the same place or same format, and even if I do, I never get it on a computer, so I can’t analyze or share it easily. Number two, I forget.
I’ll give you the tldr; solution to save you from reading further:
- Characterize a problem that requires regular data input
- Distill the data input down to something you can put in a form
- Hassle someone to do the input with calendar notifications
If you’re still with me, this all started when I was wondering how long pothos cuttings take to root in water, so naturally I turned to Google, where people disagreed with each other and weren’t terrifically helpful. Naturally I thought since I have a bunch of vines sitting in a glass of water, I could resolve the
question once and for all by measuring how long it took. I started to write down observations in my notebook, when I was struck that I could just as easily put them in a spreadsheet. I was further struck that a spreadsheet is unwieldy, but a form is pretty easy to use and puts stuff in a spreadsheet for me. So I made a form that asked what day it was and whether I could see any roots.
This ran straight into my second problem, but since I was already in the headspace of doing the quick and dirty computerized assist, I just made a recurring calendar event with the address of my form and have been happily collecting data ever since.
Immediately after populating my calendar, I started to think of other things I mean to do on a regular basis, but never quite manage, which lead me back to my daily haiku project. I followed the same pattern for haiku and have been churning out mini poems on a schedule ever since. I think both of these approaches have the added bonus that it should prove fairly easy to siphon data out of a google spreadsheet into a more useful form
/Edits to clean up email formatting oddities
I knocked together a quick app to keep track of R and my’s progress on C25K. When I started out, I was using skeleton for the layout, but I got frustrated with its documentation and some behavior with spacing between rows and columns that took (I felt) too much micromanaging.