2,815 karma · joined August 26, 2011
It is really hard to get the kind of exponential efficiency gains in this space that a 'blitzscale'-focused startup needs in order to really succeed.
Trucking is an industry that relies heavily on relationships (trucker <-> broker, broker <-> shipper, broker <-> facility, etc), and you're not going to change that (required for true automation) without already being at an absurdly large scale.
Convoy spent too much effort on chasing things that had the potential to give the company those exponential gains without addressing the high-touch nature of the industry (but were just not a good fit, and ultimately fell flat).
And Convoy—conversely—had very little incentive to spend most of its effort on iterative improvements to the core brokerage (which would have probably led to profitability, but not the kind of returns that VCs ultimately want)
…if the company still exists (see comments above and below)
Unless you're manually storing them as strings... (Not ideal, but most dbs are pretty good at dealing with that too)
Can the average dev keep all of the variable states & side effects of the function in your head as they read through it? Great! Linear may be a good fit.
Or does one need to jump up and down in the function to /really/ understand it? Probably time to consider abstracting it.
Consider how much cheaper things might be across the board with lower fees
It's a lot easier for single binary projects (like games) than multi- multi- target projects (web frontend, backend, etc) - plus debugging tools are less mature in those worlds.
If debuggers mostly Just Worked in that kind of environment, people would use them far more often.
And the culture is to not invest in that config
The barrier to entry to the role just got a lot lower, but the barrier to gaining enough experience to become a specialist just dramatically jumped up
(totally speculative) If Bard were to form queries against knowledge graph, and then summarize the results, it could be very up to date for a good amount of information
It can give competitors the ability to estimate the size/success/etc of various aspects of your business.
This was a major motivation for a certain online retailer to generate non-sequential IDs.
Some other interesting examples in an old HN thread: https://news.ycombinator.com/item?id=7278198
Preventing competitors from estimating the size of your business (or of your customer's businesses, if generating sequential IDs on their behalf) is one big reason for having unguessable public IDs.
It is really interesting to see a recent(ish) trend away from this three tier design and back towards tighter coupling between application layers. Usually due to increased convenience & developer ergonomics.
We've got tools that 'generate' business layers from/for the data layer (Prisma, etc).
We've got tools that push business logic to the client (Meteor, Firebase, etc)
* Iran's "morality police" beat a woman to death for improperly wearing a head scarf.
* This has triggered a major uprising throughout the country - people are sick and tired of being oppressed by the regime there.
* Which has led to widespread protests, with security forces responding with extreme force - mowing down protesters with assault weapons, using ambulances to sneak up on & capture protesters, using young kids as security forces, and a lot more. 100s / probably 1000s dead (hard to confirm). Steadily rising.
* The regime has also cut off most internet access, which has made it extremely challenging for people to get word out
The people of Iran could really use help building awareness of this, and getting some international pressure put on the regime that's in charge.
https://old.reddit.com/r/worldnews/comments/xtwnvj/iran_says...
And for some more detail on what happened - plainclothes security forces (basij) surrounded students/professors at the university, have fired on them (rubber and real bullets?) have many of them trapped.
"Iranian forces shoot at protesting students, lay siege to university" - https://www.reddit.com/r/worldnews/comments/xtyias/iranian_f...
If they consistently take this approach for other cancellations, it could change the the common view from:
"why use this? They're just going to shut it down in a few years anyway"
to:
"oh neat, Google's experimenting with something new. Let me try it out. If it doesn't work out, they'll take care of me."