@JosephDaigle@ICooper If I get Yan LeCun right, he suggest we reason based on abstract world model we build through life. On the other hand LLMs generate based on probability.
Interestingly brain has a small language-generation area that seemingly works like an LLM. We use it to voice our thoughts.
Making business projections is important, but you can do that entirely without traditional estimates. To me, an estimate is an analysis of a specification. You can make predictions from estimates, but the predictions are not reliable. For one thing, every time the spec changes, the estimate changes, and no business wants a moving target. So, what's the alternative? I think of a "projection" as something you do by measuring the work—average stories completed per unit time (throughput) and average idea-to-customer's-hands time (lead time)—and then extrapolating from those measurements. No specification is required. The latter approach is a very Lean way of thinking about things and turns out to be much more reliable than estimation. If this is all new to you, Dan Vacanti's "Actionable Agile Metrics…" is a good introduction to Lean metrics.
@zimmed_io@unclebobmartin @industrieapps @StewieMAnderson from my experience it’s impossible to predict if a project going to need framework decoupling in the future. it always seemed unneeded and it always became needed eventually. just recall hooks introduction, when react introduced totally breaking change to itself
Imagine a prompt formalism that is a combination of BDD (Given/When/Then statements) and DbC (Design by Contract). The former describes behavior. The latter describes constraints. The two should completely overload each other, and the solution space.
The programmer should be able to iteratively add to and refine that formalism while the AI builds and modifies the test suite and the code.
The programmer should be able to review both the test suite and the code as it is being refined. Each change to the prompt should create highlighted changes to the tests and the code. Execution of those tests should be automatic; and failures should be noted, but not fixed, by the AI.
@unclebobmartin @industrieapps @StewieMAnderson remember that “mental model” thing Dan Shapir told you? facebook really impose very specific mental model via react “best” practices to the huge chunk of js/ts devs. probably even majority. very hard to tackle that sometimes almost religious devotion.
A horrific Russian ballistic missile strike on Sumy.Russian missiles hit an ordinary city street, ordinary life – residential buildings, educational institutions, cars on the street… And that’s on the day when people go to church – Palm Sunday, the feast of the Lord's Entry into Jerusalem.
According to preliminary information, dozens of civilians were killed and wounded. Only filthy scum can act like this – taking the lives of ordinary people. My condolences to the families and loved ones. A rescue operation is underway. All necessary services are working at the scene.
The world must respond firmly. The United States, Europe, everyone in the world who wants this war and these killings to end. Russia wants exactly this kind of terror and is dragging this war out. Without pressure on Russia, peace is impossible. Talks have never stopped ballistic missiles and aerial bombs. What’s needed is an attitude toward Russia that a terrorist deserves. I thank everyone who stands with Ukraine and helps us defend life.
Hi @allenholub In one of your vids you talk against estimations in agile. instead you suggest to use projections. can you please explain “how to” more or give a reference to study the approach?
@SpeakServeGrow for ukrainian gen x add to that:
91 - soviet union collapse leading to
90-s - decade deep crisis
04 - revolution
14 - revolution
14 - start of the russian hybrid war
22 and ongoing - russian full scale invasion
25 - ...