nrnrjrjrj

Open source won those battles but the war doesn't end. The next fight is AI and thanks to a leak we have open source (weights and inference) models now.

Without that leak we would not have the ecosystem evolving around Llama.

Thw problem is big tech can offer free as in beer hosted services.

You can use Google docs for free so it takes some dedication to self host that and pay for the server.

Now if big tech charged for everything things would be more like the old days where you might use small tech, such as a local hosting provider that does open source installs.

Digital ocean is good enough for this. You would be trading the power/$ ratio of Hetzner for probably an easier DX with more regional options.

Those exact words (aka blue green deployment) apply to loadbalancers too and they can do it better. They can even do health checks and slowly ramp traffic to the new server and back off if things go bad for an automated rollback.

Not to mention money losing value, unlimited losses on shorts and the occasional gamestonk style manipulation.

Maybe this is why index funds are good advice for most. As dumb as they are they tend to do the things humans can't stand doing and invest more in winners after the fact.

Do you buy NVIDIA after it has gone up and you missed the boat? Hard to do with an ego, but an index fund will do it with your monthly allocation.

Although index funds rely on other players in the market that actually think. They get the edge but following them a few hours later probably ain't bad.

I have a rule. Any pension-invested money has to stay in index funds.

Maybe I have something like thus but admin tasks are a real challenge. Monday morning my heart wants to code and be in flow... but I know I need to plan future initatives, communicate stuff and worst of all make phone calls for personal stuff.

I think they are right. Advanced ML requires both IQ and career dedication. Doing say the intro YT series building yoyr own LLM and understanding and sucessfully using PyTorch data strucutres in 3 even 4 dimensional (as in an array of array of array of arrays) doing matrix algebra...

I would say this gets possible at 120 and only easy at 140 at a guess.

Edit: i am serious

There is an art to writing tests especially getting absraction levels right. For example do you integration test hitting the password field with 1000 cases or do that as a unit test, and does doing it as a unit test sufficiently cover this.

AI could do all this thinking in the future but not yet I believe!

Let alone the codebase is likely a mess of bad practice already (never seen one that isn't! That is life) so often part of the job is leaving the campground a bit better than how you found it.

LLMs can help now on last mile stuff. Fill in this one test. Generate data for 100 test cases. Etc.

If you are smart and think alot you can do well renting and investing elsewhere.

You can also ask what is life about?

This is hard to do because the conclusion may need to break moulds, leading to family estrangement and losing friends.

I suspect people who end up having a TED talk in them are people who had the ability through courage or their inherited neural makeup to go it alone despite descenting voices. Or they were raised to be encouraged to do so.

I want to block some time to grok the WBR and XMR charts that Cedric is passionate about (for good reason).

I might be wrong but I feel like WBR treats variation (looking at the measure and saying "it has changed") as a trigger point for investigation rather than conclusion.

In that case, lets say you do something silly and measure lines of code committed. Lets also say you told everyone and it will factor into a perforance review and the company is know for stack ranking.

You introduce the LOC measure. All employees watch it like a hawk. While working they add useless blocks of code an so on.

LOC commited goes up and looks significant on XMR.

Option 1: grab champagne, pay exec bonus, congratulate yourself.

Option 2: investigate

Option 2 is better of course. But it is such a mindset shift. Option 2 lets you see if goodhart happened or not. It lets you actually learn.

Key orientation is the new dishwasher loading pattern.

My trick:

Car key is obvious.

Other keys are house front, house side, someone elses house (family).

Get some nail polish and paint the top of the house key, the one thay gets used 90% of the time.

Or you can get the coloured covers from key cutter shops but they do seem to break up eventually.

This is an annoyingly good argument. Yet my gut says a bunch of floating point numbers isn't open source code.

I think it goes to show how hard it is to make analogies between thw two fields.

Maybe it is just not source at all. Open or closed.

It is data. Like a csv of addresses and coordinates that were collated from different sources that say are no longer available.

It is a very philosophical topic. What if machines got faster and you could train Llama in 5 minutes, and an SSD could hold all the training data. Then it would feel like a compiled artifact more than data. Not releasing the training data would then feel like hiding something.