lawtalkinghuman

I have tested Lexis AI once for a legal research point. I wasn't particularly keen on putting the exact details of an actual problem in, but I gave it a summary version.

It didn't feel drastically different from using ChatGPT with the ability to search the web, except it was searching material on Lexis, both statute/case law and commentary. It dug out some commentary that confirmed my prior hunches, but also pointed to some cases that weren't in any way relevant.

Otherwise, all the experimentation I've done is with non-confidential material using public LLMs.

Just because the citation exists, what the LLM says it stands for and what it actually stands for are not the same.

For testing, I've asked (admittedly last-gen) LLMs to generate legal opinions regarding issues in commercial English civil litigation, and I received back cases where the citation is real, but the area of law (family law) is not relevant as family courts apply a very different set of procedural rules.

(If you squint a bit, they sometimes might be relevant... and could be useful for a particularly creative litigator to make a novel argument on behalf of a very risk tolerant client. But you would very much want to go read those cases and think quite hard about them.)

It’s almost as if all the guardrails aren’t real.

The answer is: the market will work it out eventually. Clients will push for more work to be fixed-fee/outcome-based rather than billed hourly. There'll be some small firms who'll successfully grab lots of lower value clients who are willing to use digital tools to handle their work and don't particularly care about having a big fancy office in London or New York if it means lower bills (and they can then basically use the relationship they've had providing the supervised online service to be the first point of call when said client wants something that's less off-the-shelf and needs more work).

Also, an interesting example: in English litigation (where, broadly, loser pays unlike America where each side pays), maximising billable hours is not always a viable strategy for anybody if those costs aren't recoverable on success. Someone involved in large-scale commercial litigation involving disclosure of millions of documents who doesn't use algorithmic document classification (now pretty broadly accepted as normal) potentially runs the risk of a judge determining that the costs of going through all the documents by hand isn't recoverable. Insurers/litigation funders aren't going to want to risk padding the costs so much that the judge prevents them from recovering their stake in the litigation.

Customers using their own LLMs: yep, they might do that. I think the pitch from the legal LLM providers is "we've got legally trained people doing RLHF to make it more accurate" mixed in with "also we've got a partnership with Lexis/Westlaw/etc. so we can do legal research that's better than what's on the open web", with a little bit of "if you get sued for professional negligence, 'I used the legal AI thing that's built into Westlaw' is gonna be more convincing to a judge and jury (and your insurance company) than 'I used ChatGPT, yes, like the app you've got on your phone'...".

That's why the EA guys are off beng quants.

Or in prison for fraud.

Yes, that's the general rule.

You then get lots of interesting exceptions and difficulties in applying that rule: who pays when you have multiple potential defendants (but some of them have died or gone out of business or are in a different jurisdiction), or there are fiddly causation issues, or the defendant is an organisation that's vicariously liable for their employee/contractor/authorised religious leader/any number of other relationships, or there's insurance involved, or there's some third party interest, or the purportedly tortious act was mandated by some law or other obligation, or the government decides to set up some kind of alternative process of compensation which may or may not indemnify the tortfeasor, and so on and so on.

When dealing with the not-so-straightforward cases, appellate courts do look at questions like "is this a fair and equitable distribution of the cost?" and legal scholars compare the pros and cons of tort liability to other mechanisms (compensation by the government, industry self-regulation, no-fault/strict liability, mandatory insurance etc).

If my goal is not seeing AI slop, I don't particularly care whether it is honestly labelled or not.

The difference is OneDrive is moderately useful.

I think we must make it clear that this is not related to AI at all

There are clear AI-specific reasons why it's being crammed down everybody's necks.

Namely: someone in management has bet the entire strategy on it. The strategy is not working and they need to juice the numbers desperately.

Keeping the lights on is fine.

But if they remove a feature I rely on, I can't put it back.

If they add a feature I hate, I can't remove it.

If they jack the price up, I have no real solution to this.

If they move features I rely on from the standard tier to the 5x more expensive pro tier, I have no real solution to this.

Why, yes, this is an echo of the old argument for open source software.

you'll slowly get dependent on things that will eventually break in ways you will have no capacity to fix

If the commercial provider charging you $10 a month breaks it, you also have no capacity to fix it.

Your options are: send them an email, or unsubscribe and use something else.

Same in the UK.

Votes close at 10pm. Might be a few stragglers left in the queue, so call it 10:15pm. (Exit poll results are embargoed until 10pm.)

Ballot boxes are transferred from individual polling station to the location of the count. The postal votes have been pre-checked (but the actual ballot envelope has not been opened or counted) and are there to be counted alongside the ballots from the polling stations.

Then a small army of vote counters go through the ballots and count them and stack together ballots by vote. There are observers - both independent and appointed by the candidates. The returning officer counts the batches up, adjudicates any unclear or challenged ballot, then declares the result.

The early results come out usually about 1 or 2. The bulk of the results come out about 4 or 5. Some constituencies might take a bit longer - it's a lot less effort to get ballot boxes a mile or two down the road in a city centre constituency than getting them from Scottish islands etc. - but it'll be clear who has the majority by 6 or 7 the next day.

I can appreciate that the US is significantly larger than the UK, but pencil-and-paper voting with prompt manual counts is eminently possible.

The actual data is being held by GPs, hospitals, other secondary care providers, and pharmacies. Enough of those providers use systems that all conform to a bunch of common standards and APIs that the NHS app can get that data (and the idea is it puts pressure on the remainder to switch to systems that are accessible).

The capabilities the NHS app offers will depend on what subset of the functionality the GP practice has implemented (on, in reality, the commercial vendor that makes the software they use).

NHS has pretty reasonable developer documentation which explains most of the high level pieces of the system - https://digital.nhs.uk/developer/guides-and-documentation

The metaverse is clearly the future. Zuckerberg said so, after all.

Browsers without metaverse integration will be a non-starter.

The metaverse is here to stay! Blockchain is the future!

Without integrating metaverse and blockchain features into Firefox, Mozilla is at a significant disadvantage compared to other browsers. Don't get left behind!

Mozilla shouldn't need to worry about adoption numbers though.

They could even make the AI features available as extensions, downloadable from addons.mozilla.org

That way, the users who want them can download them, and the users who don't, don't.

the numbers speak for themselves

What numbers? Have Mozilla published any numbers showing their AI experiments have been warmly received by users?

When I'm unconscious in an ambulance, I'm definitely in a position to appreciate all that price transparency the free market has provided, so I can rationally weigh up all my options calmly and objectively while my organs are shutting down.

Yes, but there would have to be an error of law in the High Court judgment.

(And you'd usually appeal to the Court of Appeal first.)

[citation needed] because that’s not how the OSA works.

The reason people (myself included) rather like good Git commit messages is evident when one compares them to the alternative.

You're working in a commercial/closed source environment and want to find out why line 57 in src/blah/db/utils.py does that. Where do you look?

- inline code comments. Usually non-existent. Often out-of-date, sometimes misleading, frequently tells you no more than you can discern from just reading the code itself (especially now type annotations are trendy again). Rarely explains why the code exists. There's a reason people caution against too many comments, and that translates into people probably not putting enough commentsin.

- calling code? Helpful, but thanks to microservices and increased levels of abstraction (APIs, DI frameworks, messaging buses, config parsing) you've got to go check 900 different repos out to work out what is going on.

- email? Give up. You'll find invitations to the company Christmas party and Q2 sales figures but actual tech explanations are in short supply.

- Slack etc - same problems as email, plus developers who hide away all the interesting stuff in private team channels

- Google Docs - you probably don't have access to the relevant doc, and there's no way to know that you don't

- wiki/docs? Half baked, wrong etc. Or it'll be autogenerated JavaDoc type stuff that'll tell you what you already know or can reasonably infer from the code. Also, findability sucks. Or the developers just avoid the whole thing because the software is nasty and corporate and barely usuable.

- bug tracker/ticketing system? You ask around and someone says "oh yeah, Dave made that change two years ago" and then you search for tickets that match related keywords only to find out that those tickets weren't brought over from Trello into JIRA, and now you need to go ask IT to give you access to the legacy Trello board which they don't want to do because then it'll put them over the five users per month limit or whatever.

- Architecture Decision Records / decision logs / whatever you want to to call them - nice if they exist, I guess.

- ask the person who wrote it? This assumes they still work there and can remember. Plus you gotta do the asking around routine which takes days and destroys all hope and joy in the world.

By a process of elimination, commit messages are the closest you're going to get. They're right there - on your computer, neatly integrated into your editor, hopefully. You can search them fast in a terminal window rather than in some slow web-based monstrosity. If you're lucky, they're actually useful. Even if they aren't, they're at least contextually useful in helping you narrow down your search strategy for the inevitable plunge through email/slack/JIRA/Trello/internal wiki etc.

Ideally what should happen is the really useful commit messages get copied into stable technical documentation like decision logs or a properly maintaned wiki. If people did that, great, but it's pretty rare. A culture of sharing weird interesting tech things in a Slack-type system can help because future devs can at least search but you do that at the cost of more interruptions for colleagues now.

The broader issue is of all the bad options you can choose, it often tracks the wrong thing. In something like Trello/JIRA/whatever, if you're looking for the technical reasons, it'll have the business reasons without the technical stuff, or vice versa. You generally want both, and most systems only give you half the story.

In addition to the RSPCA example, banks/financial institutions bring a fair number of complex fraud prosecutions because it is in their financial interest to prosecute people who've defrauded them. But private prosecutions are indeed an anomaly and can lead to some weird outcomes.

A few years ago, someone brought a crowdfunded private prosecution of Boris Johnson for "misconduct in public office". The whole thing was incredibly silly and a massive waste of time. (MIPO isn't there for politicians lying/misleading people, though that does indeed suck.) https://en.wikipedia.org/wiki/Ball_v_Johnson

Also, following the murder of Stephen Lawrence, the CPS decided to not prosecute for lack of evidence. The Lawrence family brought a private prosecution which failed. Jeopardy attached at that point.

Later evidence emerged which made the case a lot stronger, but had the Criminal Justice Act 2003 not been passed allowing limited exceptions to the double jeopardy rule, it would not have been possible to bring much stronger cases against the perpetrators.

See https://en.wikipedia.org/wiki/Murder_of_Stephen_Lawrence

The CPS have the power to take over and stop an ongoing private prosecution, but the ability to bring private prosecutions sits really uncomfortably with the principle of double jeopardy. Like, someone could do a bad job of bringing a prosecution in a case where the CJA2003 double jeopardy exception doesn't apply and then the person can't be retried by the CPS.

pex is a build tool built on the top of the zip trick. A pex file doesn't contain a bundled interpreter, but it zips up the entire virtualenv. Also no Windows support.

https://pex.readthedocs.io/en/latest/

Central Line is due next ~2030. Then Bakerloo: ~2033.

It's kind of reasonable to deprioritise the Bakerloo, even as someone who used to use the Bakerloo very frequently. Bakerloo gets significantly less usage than other lines: Piccadilly has double the number of annual travellers than the Bakerloo, for instance.

(It'll be interesting to see if Piccadilly usage numbers go down now Crossrail has been finished and some portion of Heathrow travellers switch over.)

The Bakerloo itself is also kind of a weird, redundant line. The NW segment doubles up with the Watford DC line, and for a lot of NW destinations like Wembley, the Jubilee and Metropolitan are better choices.

There was also the issue of the proposed extension out to New Cross and/or Lewisham. They probably wanted to put off making a decision on the rolling stock upgrade until after they'd decided whether they were doing the SE extension.

Court dress isn't based on seniority—it is based on which court you're in. If you're working in criminal practice, it's pretty routine. If you're doing civil or commercial work, it's not quite as frequent.

Current court dress guidance from the BSB: https://www.barcouncilethics.co.uk/wp-content/uploads/2017/1...

Yup, I do. It's a great place to talk to myself and nobody else.

A bunch of people I know who play video games are on Telegram. I'm pointing out that I can talk to them without sharing my phone number which I can't do on WhatsApp/Signal.

Not being stalked by weirdos is a higher priority for a lot of people than anything involving spooky Edward Snowden stuff involving the NSA, FSB, GCHQ, Mossad and Five Eyes etc.

Meanwhile WhatsApp and Signal's privacy model ensures everyone you talk to has your phone number which means they have the ability to make your device buzz at any time of the day and night. It also excludes people who do not have a phone number.

On Telegram, I can be in rando messaging groups of people and have a username rather than a hard-to-change, legacy voice traffic routing number that's widely used for tracking.

The only improvement is if they could get rid of the requirement that you use a phone number to sign up for these services, which is a regression from AIM, MSN, ICQ, IRC and Jabber/XMPP.

I mean, contractual terms that are implied by statute exist. In English consumer contract law—which since 2015 has been extended due to the EU Consumer Rights Directive to cover digital content—includes an implied term that the goods are of satisfactory quality, and when it's a continuing service (including something like a digital content service like Netflix or Spotify, or a software product with updates), it doesn't radically depart from what's initially offered.

See https://www.legislation.gov.uk/ukpga/2015/15/part/1/chapter/...

Most jurisdictions have something broadly similar (albeit often not quite up-to-date around software and digital products). Everywhere in the EU will have laws that implement the EU's Consumer Rights Directive.

Which is great and would apply if you'd paid money for it. NightOwl is free (as in beer). The expectations the law sets out regulating the sale of goods and services do not apply when no money has changed hands.

Which I'd argue is pretty much right: while it sucks that companies get taken over and have spyware crap put into products, the idea that, say, a teenager who is hacking around and building stuff to learn how to code, puts up a project they've made as open source or a freeware download, does something silly like the left-pad debacle, then gets sued—potentially by a big corporate behemoth with very deep pockets and very scary lawyers—for a series of acts which involved them writing some software for no money. Regulation of technology should rest far heavier on the shoulders of Google, Microsoft, Apple and so on than it does on a hobbyist or small indie dev creating freebie menubar utilities or Chrome extensions or whatever.

The difficulty of ensuring those little freebie and open source apps don't become a vector for supply chain attacks remains difficult. Much better sandboxing and OS app-level permissioning, good network monitoring and anomaly detection on a per-app level, and building trust into packaging/distribution processes - these are all slow, grinding, incomplete ways to improve this. Lawsuits probably aren't.

It isn't a criminal offence. It's a civil penalty that can be levied if they require you to register and you don't.

The legislative text (it's an amending regulation—think a diff—but the underlying statute hasn't been updated on Legislation.gov.uk yet) is here: https://www.legislation.gov.uk/en/uksi/2013/3198/regulation/...

The Electoral Commission's guidance - https://www.electoralcommission.org.uk/running-electoral-reg...

I can't find any reports of people actually receiving the penalty. It'd certainly be useful if the Electoral Commission collected these statistics, but they don't. It seems fairly rare, and it's likely to be one of those things that's there to encourage people to register to vote rather than something to be actively enforced.

I'd also note that given there is a requirement for an in-person home visit to encourage registration before service a formal notice, it's pretty likely that COVID restrictions and guidance meant quite a lot of the home visits didn't happen over the last few years.