1. This doesn't account for undiscovered reserves.
2. This doesn't account for the ability to substitute one mineral with another in some applications (copper was replaced with zinc in pennies, and aluminum in wires when the price went up).
3. This doesn't account for consumers responding to higher prices as a resource becomes scarce by substituting an entire product or technology for another (as the price of energy increases, more people in dry climates switch from AC to swamp coolers).
There was a 20x performance difference between the 2 machines they tested simplejson on. Having used both, I feel that simplejson has a pretty sizeable reliability advantage (unless cjson has changed a lot in the last year), simplejson is pure python (with optional c speedup module) so it works on appengine, and simplejson is far far more widely used.
15 people for 100k/month would mean you are paying an average of 74k/year. Who would willingly work for a startup for that, without lots and lots of equity (answer: the exact people that you would never actually want to hire)? And you can't give lots of equity to 15 people, and even if you give them 8% each you would be left with nothing yourself.
$12k per month on rent? You would have to be retarded to pay that as a startup. Rent an old house and get lots of card tables.
Servers/bandwidth at 20k/month? What kind of site is this? That's $240k per year of bandwidth. If you are serving up that much content and aren't making a profit already, you are either going to get bought out in a few months or go under in a few months after that.
My hypothetical startup budget:
Salary: Pay myself enough to pay bills and live on, as well as 2 or 3 talented confederates (if I happen to have 2 or 3 talented people, with skills I need, on tap. Otherwise, wait until some are located before hiring anyone. Nobody is getting my equity unless they totally rock.) Estimate: 6k-25k per month.
Rent: $0, I happen to have an office large enough for 4 people to comfortably work in at my home. If I didn't, I would charge a one time expense of $2000 to put some carpet in the garage, install an AC out there, and run some wiring and so on.
Furniture, equipment, etc: $500. Find your nearest very large public university. They probably have a depot that just sells old furniture and equipment. You can drive up with a truck and get enough furniture to outfit a small army for under $500 (one time fee). http://www.purchase.umd.edu/ttrader/items/showroom/index.htm...
Servers/bandwidth: This obviously depends on what site you are making. A Youtube needs way way more bandwidth/servers/rackspace than a yelp for the same size audience. Using EC2 as a guide, a small instance is $70 per month if used constantly. If you used say 100 instances constantly, and transfered 1TB a day, you would be paying around $12000 per month. That is a metric assload of servers and bandwidth, so the $20k number is probably retarded also.
PR/Advertising: I'm not sure. 10k per month buys a fuckload of adwords. At a CPC of .40, you are talking 40k * 2.5 = 100k impressions a month with that advertising budget. In reality you can get clicks for .20, and lately even less, unless you are selling asbestos litigation.
So my point is you can get a hell of a site out there for like $300k per year or way way way less, unless you are an idiot.
Making a really big website does cost money. You can't host yelp on your VPS account. On the other hand, startups that have raised money when they aren't basically failing due to inability to serve up content due to too high of demand are idiots (and the people who hand them money are double idiots).
4% of 0 is 0, so I wouldn't worry about it too much. If you don't have a no-compete, which it sounds like you don't since there isn't any company to have an agreement with at this point, I would offer him a full 40% and walk away.
There is no such thing as 'very knowledgeable in technology and the web'. Either they know how to program or they don't. If they don't, odds are they will be expecting you to put in long long hours while they surf news sites and dick around and pretend to be really busy.
And if they knew all about business they wouldn't have needed to take 50k from an angel, since they could have written a check with the proceeds of all their prior business success. 50k would pay for a single programmer to work for 4 months or so...
Oh, there is an alternative. Agree that when you get a first round of funding or make a profit that you'll be paid for all the hours you work at some reasonable rate (maybe $60 per hour). They should get paid for the hours they work also, but obviously since you are the only one actually doing anything I wouldn't let them bill more than 5 or 10 hours a week.
$80 per month, 10mb/s unmetered, 2048mb RAM, 80GB disk.
Slicehost gives 400GB, 1024mb, 40GB for $70, which isn't too bad but 400GB is a garbage amount of bandwidth, and they charge $0.30/GB for overage.
vpsfarm doesn't seem to have any way to set up a backend network, so you are stuck on the one machine from what I can tell. I haven't asked about it, but for me this is fine anyway.
I've used slicehost, gandi.net, and vpsfarm's xen, and webfaction's shared hosting. I have to say, slicehost is very very nice, but the amount of bandwidth they offer is garbage. gandi.net is ok, but they seem to use raid 6, and my '2 unit' vm had terrible disk IO (otherwise it seemed ok, they charge $16 per unit, you can dynamically change it from 1 to 16, for example to use more resources during peak times. you get 5mb/s unmetered per unit, which is pretty cool, but the servers are in France, so add 100ms of latency to something in the US).
vpsfarm however gives very good cpu and disk performance, unmetered bandwidth, and it is a fraction of the cost of AWS if you want to run a single dedicated box. It is super bare bones, basically nothing is provided but a root login. So far I am really liking it.
The stock market is zero sum? That is hilarious. I guess you haven't heard of dividends?
I totally understand people being wrong about basic facts. But what motivates you to go out of your way to post your ignorant nonsense to the internet?
I agree. Also, you can't patent or trademark just an idea. So just wait out the year or whatever on your NDA and make the company yourself. Unless the people who want you to build the prototype have millions of dollars in funding, they need you much more than you need them. It is nearly impossible for a nontechnical person to manage to get a web application out the door without huge piles of money. Someone with the ability to build the site can do it for whatever time it takes them to do the actual work.
Really, any nontechnical 'partner' had better be bringing in leads, selling ads, or otherwise causing cash to come in, or they are basically dead weight (and they'll have lots of little changes they want made to the site, aka the what color to paint the shed principle). Having spent hours and hours programming, while the 'partner' does basically nothing, thinking long and hard about usability and how things should flow, and then having a 'partner' look at the site for 30 seconds and start listing off changes that need to be made is not tolerable.
The two paths are not equivalent at all. When two people bring you in, they might bring someone else instead. So the outcome should be reduced by how easily you can be swapped with someone else.
If you have one of a kind domain knowledge or ability, then they might have to pay you much more than 1/3 of the company. After all, without you they would have a much reduced chance of success. If you are a cleaning lady, they can just pay you the usual cleaning lady rates, since there are a billion more cleaning ladies ready to take your job with equal ability.
Plus, you further assume that holding ownership of the existing company is worthless?
You can't make a reasonable self study. People will say you can, but they are full of crap. There are maybe a few hundred people in the world who can do something like that, and they are all top caliber geniuses.
In reality, we need the pressure of school to push us into spending the unpleasent effort to get the hard stuff into our heads. Otherwise you skim books and dick around and have nothing to show for it other than an ability flip out buzzwords on internet forums.
After all, why do you think there are colleges that teach this stuff? If you could just pick up books and learn it just as well as you can at college, why do people who are at college have to cram and study so much? After all, they are reading the material just like you, and they have the advantage of TA's, other students, and professors to guide and tutor them along. The fact is that you have basically no chance if you go it alone. Good luck.
I've developed a handful of things using APEX and their webservices API.
I've found APEX is a weird combination of a Java-like language, and you query their database with a goofy, SQL like syntax. But the language isn't Java, and the syntax isn't SQL. This is very odd, because you are locked in to their weird crap, and it is pretty much crap, for a task that could easily be done using regular sql and a regular database. Instead you trade simple and known techniques for something obscure, incompatible and poorly documented, which is basically just different enough from sql and java to be incompatible (you use single quotes instead of double quotes in APEX, to make writing the weird sql like queries easier for someone who doesn't know how to escape a character I guess).
And what do you get for all this tradeoff? Basically simple CRUD operations on simple 'contact' datatypes.
It is a miserable piece of shit, hodgepodged together with seemingly no rhyme or reason, and the 'documentation' is basically a list of examples you can cut and paste and modify without really increasing your understanding.
Visiting their forums, you suddenly realize the meaning of the term 'cargo cult programming'. I have seen many examples of code submitted as 'answers' that even a cursory glance at revealed line upon line of unreachable or otherwise no-effect code, and so on.
Salesforce.com is, in other words, MS Access for the internet age.
The neighborhood I live in is almost exactly 50% Black, 50% White (at least my cul-de-sac is exactly 50%, if you count per-house, probably about that per-person). I have yet to see anybody I would consider scary at all, and that is not because I have a high tolerance or something. So your generalization is complete and utter bullshit right on the face of it.
I think you need to be much more careful in your assertions about race. When people are subsidized into a neighborhood, you are getting people who's life skills aren't up to par with the other members of that neighborhood, race has nothing to do with that whatsoever. If you managed to get some German tribesman in 500 AD and put him up in a well to do Baghdad neighborhood of the time, the hard working Arab neighbors would probably find them vulgur, dirty, and likely to steal when the opportunity allowed for it.
Hell, we Europeans walked on to the stage of history as backward celts getting dumped on by Romans and worshiping trees, or crazed Asian nomadic barbarian hoards smashing the greatest cities and civilization to that point (talk about bad neighbors).
Anyway, I have a title I would like to recommend for your post. "Protocols of the Elders of Africa."
Postgresql can scale far far better than mysql on a single multiprocessor machine. Postgres scales up to about 16 cpu's without much difficulty. If you put heavily used tables on lots of seperate spindles, you can go very far.
Also, postgresql supports partitioning of tables. This lets you effectively 'shard' a single database, without having to change your queries. As long as the planner can determine which partitions the data could possibly come from, it will ignore any other partitions. So if you have your table in 30 partitions, and the planner determines the query could only return rows from 8 of those tables, it doesn't scan/check indexes for the other 22. If you partition the data such that it is effectively 'archive' and 'live' data, in that the 'archive' data partitions are rarely accessed, this is a HUGE win, because the data is indexed per-partition, meaning the relatively small 'live' indexes get kept around in memory, and combining with other indexes for bitmap-indexed joins is therefore that much faster as well.
If you don't need concurrency, use sqlite, if you do, use Postgresql or Oracle.
<conspiracy>Why do you think Sun (a hardware company) bought Mysql? Maybe because you need a server installed for every 100 users on your site? </conspiracy>
1. This doesn't account for undiscovered reserves. 2. This doesn't account for the ability to substitute one mineral with another in some applications (copper was replaced with zinc in pennies, and aluminum in wires when the price went up). 3. This doesn't account for consumers responding to higher prices as a resource becomes scarce by substituting an entire product or technology for another (as the price of energy increases, more people in dry climates switch from AC to swamp coolers).
And on and on and on. 4.