Some problems are so hard they need to be solved piece by piece | Hacker News Reader

Some problems are so hard they need to be solved piece by piece

cdixon.org

89 pointsswohns31 comments

Interesting. It seems like there's some good discussion over the past few days on hunting ground for good ideas. Everything is rightly disclaimed with a "not guaranteed to produce a good idea."

Moving people from general tools to more specific ones is tricky. Most of the things people use spreadsheets or email for could be done with more specific tools, possibly better. Time tracking, data storage, project management, scheduling. General tools that you are already using have a lower overhead.

I can guess at 2 big gotchas. (1) To move people to narrower tools you need a very big improvement. Slightly better isn't enough. For a lot of things there just isn't that much room for improvement. (2) When you move from general to specific tools the software dictates how the user does stuff more. This leaves room to get it wrong.

Using one tool instead of tons of better tools also has the advantage of just one tool to learn.

Have a look at https://pinad.com.au - it might be the big improvement you speak about. It has structure and specialization for each of 81 categories currently available.

I described a bit more in a reply to the main thread (http://news.ycombinator.com/item?id=4824233)

Disclaimer: I'm the creator of pinad.com.au

A while back I was reading some interviews with older famous programmers, like Dan Bricklin, author of Visicalc. Before the spreadsheet, everything was special-purpose. I think the reasons you give are a big part of why the spreadsheet was so valuable. That's not to say special-purpose business applications don't have a market, because obviously they do, but before building something like that I'd double- and triple-check that it's really better enough than a spreadsheet to be worthwhile.

Has anyone had a good experience hiring via StackOverflow Careers? I've found it to be expensive, and it doesn't necessarily provide more or better candidates than craiglist, Indeed or LinkedIn.

I was always afraid that by putting my profile on StackOverflow Careers, prospective employers would see us as individuals liable to spend company time solving other peoples' problems online and therefore be less likely to hire us.

Good point, I find it kind of unnatural to transform a developer community into a recruiting platform. I think it's just companies searching for a solution to solve the resume problem.

It's also an incentive for developers to earn karma as it may reflects on a career opportunity. Other popular developer websites, such as github, also have a way to differentiate /popular/ or strong developers (number of stars/fork). Personally, I prefer the github as a developer and a manager mostly because I'd feel more comfortable working with an employee that crack lots of code rather than one who spend lots of time answering questions.

as a developer and a manager, i'd love a guy who could productively answer questions. it means that he can both solve problems and help other developers who might be struggling with the same issues.

this is the kind of developer who will make sure his code is commented from the perspective of letting someone else see how it works, and that your internal dev wiki is well gardened. if you ignore the value this sort of person can add, it is entirely your loss.

I think I'm with you on this one. Crappy code can still get a lot of stars/forks and great code can pretty much get ignored if it's not the flavor of the month with the majority bandwagon. I think Github is great if you want to look at someone's code, but being able to successfully answer questions, while more academic than utilitarian, in general identifies those with a deep and thorough understanding of the technology. SO gets gamed though.

Ultimately, I suppose that either are a good bet if you're looking for someone who stands out in a certain way. In both places you should vet the quality of the code or quality of the answers. And if you're the type of employer who uses these as metrics for rating candidates, then I think you should also probably give your employees time to work on their Github projects/SO profiles (I didn't intend the irony there, but it's there all the same).

Such a transformation was made with the designer and developer community Forrst and has been quite successful. I'm sure there are many more niche communities I'm not aware of -- to avoid the tragedy of the commons such quality communities are often not heavily advertised. For instance, Forrst is invite-only.

Very unlikely for one reason: If someone is looking for employees on StackOverflow Careers, they have already decided that they want someone who is solving problems for others. If they were against the idea of StackOverflow Careers, there is no way they would find your profile on there to begin with.

The problem might come if a company you apply to using a different method Googles your name and finds the stackoverflow profile, and then decides not to hire you based off of that. Feels a bit unlikely and I'd doubt any HR would understand the issue enough to do that at this stage. I'm not involved in HR and don't fully understand how their illogic works though - some do make hiring decisions based off of Facebook. laughs You probably don't want to work for a company doing that kind of thing to begin with though.

Honestly, if you're the kind of person who enjoys problem solving so much that you spend your free time solving other peoples' problems-I want to work with you. Everybody procrastinates, and I'd rather hire someone who procrastinates in a productive way.

I have been contacted by a rather large and visible prospective employer two months ago. The head hunter seemed to like the "problem solving" angle, actually, but I guess my spoken English wasn't as "good" as my written English.

Yes actually, I found a good employee using StackOverflow Careers. I may use it again but it is kind of pricey. It does beat going through a recruiter though.

Interesting. As someone eluded to in the post's comments, the "horizontal" can almost be thought of as a platform in some ways, where "vertical" are the individual product(s).

So, if the "platform" does not exist yet, it may need to be built first, but I suspect many of them already exist. It's interesting to think of what horizontal platforms are being developed, or are becoming more popular, and how you could target a vertical area on top of (or within) it.

I've seen a few discussions recently on Google Glass and the work Microsoft is doing augmented reality area as well, and wonder what vertical products could stem from that over the long term.

It's very hard for a startup to create a platform without a main popular product. It may be proven wrong, but in my opinion (and experience), it's better starting vertical and then horizontally expanding. Existing companies with revenues can afford wasting resources for a while and trying to attack each market individually.. but a startup?

Agreed, and my comment was not meant to suggest that startups should build the horizontal platforms, but rather to look at what platforms are developing (i.e. being built by others), and look for new verticals within that. So, I think we are saying the same thing.

Feels like you should take the advice even further for a startup. A startup shouldn't be going vertical or horizontal - a startup should be picking one individual problem to solve and focusing on only that.

Horizontal is close to impossible for a startup. The budgets involved don't usually allow for focusing on more than one product market at a time.

Vertical is also very hard. The startup would need to be in at least two different businesses to go vertical. Lots of IT/web startups do try this by being in both the infrastructure business and the software product business, with half of their startup aimed at creating new methods of managing servers and data, and half on providing some user software on top of that. There are lots of success stories here, but you have to assume there are lots of failures also who went overboard on building up a vertical where they didn't need the whole vertical to solve their actual startup goal.

Basically, I'd say in the same way a mature corporation is one that pays out in dividends any money that they can't get good return on, a mature startup would be one that focuses only on a single core business goal of some kind, and fails quickly if that business goal doesn't work. Vericals/horizontals can come after the startup phase and the business goal is met.

I remember the first time I sat down with my C64. That thing was quite scary. Full of buttons that had weird symbols written on every face of the keys. It had a disk drvie that would make drruuuun prrub drruuun sounds while the little light flashed. Boy, did it looked busy. The power button was nowhere to be found by this 7 year old. The cables? Where do they go? Little by little I kept breaking down each issue I faced. Its funny how things never change. One cable went here, the other there. Turns out someone was smart enough to make the cables fit in one way only. Then I found the power button after hitting every possible key on it. Luckily I had managed to plug it in before doing that. The monitor came alive and suddenly I had a working computer in front of me. Little did my oldest brother know that unplugging it would not keep me from using it.

Then came the issue with not knowing how to use the thing. I did not speak or read english and the manual was written in it. What was PRINT? GOTO? Well, that was for me to find out. Turns out, I still use the same method. Break up everything into pieces and work with each piece at a time. I do miss that C64.

But this multitude of narrow solutions creates a marketing and usability nightmare for users. What's the hip place to rent rooms now? And how do I use it, because it has no UI consistency with services for renting cars and objects.

So even if these services have success, it's hard to out-market a go to portal for trading/renting between individuals. Without a unifying interface they'll merely coexist with Craigslist.

There are pros and cons and also implementation. If the pros outweigh the cons and the implementation is good it'll work.

Personally, I use AirBnB. I've never rented holiday accommodation through a classifieds site.

That's true but the problem might be smaller than you think. I have no stats to show but how many people visit more than three or four sub sections of craigslist.org? I personally only use two.

Besides, what is lost in UX compatibility and uniformity might be gained in important features that greatly improve the service offered. For example, Odesk has immensely more usable features than the computer gigs subsection on craigslist.

Yes but who will find Odesk? How will you tell it apart from a bunch of competitors and ZenDesk?

You're limiting Odesk's fancy functionality to the small subset of experienced computer users. Even if you only use 3 subsections of craig that's 3 O/Zen/StandingDesks you didn't have to Google for, memorize stupid names of, compare to competitors and learn where they put buttons.

That's assuming some marketing or universal interface magic doesn't fix it.

Probably the vertical approach started from a need to have better search and be able to describe a certain type of object/activity. Think about it for a minute: to be able to search you have to be able to describe what you want. To be able to match a description you need to have a structure in place (e.g. the attributes of a car is a structure).

Sites like Craigslist offer a GENERIC way to DESCRIBE MANY objects/activities therefore they offer a GENERIC way to SEARCH for many objects/activities. Sites like AirBnB offer a SPECIALIZED way to DESCRIBE ONE TYPE of object/activity and they have a SPECIALIZED way to SEARCH for that one object/activity.

What if we have a SPECIALIZED way to DESCRIBE MANY activities/objects and a SPECIALIZED way to SEARCH for MANY activities/objects?

That's what I'm trying to do at https://pinad.com.au. Currently I have 81 structured categories and hundreds of attributes and options. All of them allow a specialized description and a specialized search. Would this be a better alternative for a centralized marketplace?

Think we're likely to see a similar pattern with Facebook—seems clear that the instagram acquisition happened because they were scared of someone capturing the photo-sharing "vertical" better than they do.

EVERY problem ever solved was solved piece by piece - steam engine happens when it's steam engine time.

The only exception is the moon-landing, I'm still trying to figure that one out.

Craigslist is the epitome of the "worse is better".

On the other side, picking a category that's too specific ties you down to that niche, and can limit the potential of your startup.

This fits with the observation that the structure of production is constantly deepening and specialising.

No matter how niche your field is today, it will eventually break up into smaller, more specialised fields.

And someone will move into that specialty.

This is a good thing.

Someone was browsing Adult.

Reply on news.ycombinator.com